## What is customer churn?

Churn refers to the point at which a customer decides to stop using a product or service your company offers. It can happen for a variety of reasons — poor experience, irrelevant outreach, better offers elsewhere — but the bottom line is always the same: a churned customer brings no revenue to your business. No brand can eliminate churn entirely, but the right strategy can reduce it meaningfully and protect the customers worth keeping.

## Why churn prevention deserves your attention

The math on retention is hard to argue with. It costs [5–6x more](https://www.americanexpress.com/en-us/business/trends-and-insights/articles/retaining-customers-vs-acquiring-customers/) to acquire a new customer than to reactivate an existing one. You have a [60–70% chance](https://blog.hubspot.com/service/statistics-on-customer-retention) of selling to a current customer versus a 5–20% chance with a new one. And returning customers spend [67% more](https://www.business.com/articles/returning-customers-spend-67-more-than-new-customers-keep-your-customers-coming-back-with-a-recurring-revenue-sales-model/) on average than first-timers.

Despite this, [44% of businesses](https://www.semrush.com/blog/customer-retention-stats/) prioritize acquisition, while only 18% focus on retention. That's not a strategy problem — it's a data problem. Most brands don't have the context they need to know which customers are about to leave, which ones are worth fighting for, or what it would take to keep them.

That's exactly what machine learning — and the right consumer data — solves.

## What causes customer churn?

Understanding why customers churn is the first step toward preventing it. Common culprits include poor customer service, generic outreach that misses the mark, offers that don't match the customer's actual profile, or simply acquiring the wrong customers in the first place — discount shoppers who were never going to stick around past the promo window.

The challenge is that first-party data alone rarely tells the full story. You know how customers interact with your brand, but you don't know who they are — their financial situation, life stage, household makeup, or likelihood to respond to a given type of offer. That missing context is what makes churn hard to predict and even harder to prevent.

## Churn prevention strategies

Effective churn prevention starts with data and compounds from there.

### Mitigate churn higher in the funnel

Not all churn happens at the end of the journey — much of it is baked in at acquisition. Discount shoppers, casual one-time buyers, and customers acquired through generic targeting are far more likely to leave. Knowing what your best customers look like lets you adjust acquisition spend toward people who are more likely to stick. And if demographic analysis reveals that a specific customer profile churns consistently, you have two options: reposition your brand to serve them better, or stop targeting them altogether. Either way, the insight compounds — better acquisition targeting means lower churn before retention efforts even enter the picture.

### Use data to identify causes of churn

Segmenting customers into meaningful subgroups reveals patterns: what do churners have in common? What distinguishes your best customers from the rest? Machine learning accelerates this analysis dramatically — instead of manually combing through data, you get a predictive model that surfaces those patterns automatically. For a deeper look at how propensity modeling works in this context, see [our guide to propensity models for consumer brands](/blog/propensity-to-buy-machine-learning).

### Create personal and frictionless experiences

One bad experience — an irrelevant email (or too many emails), a missed touchpoint, an offer that doesn't fit — can be enough to tip a loyal customer toward leaving. In high-consideration categories like home services, operational friction is a major culprit: needing to set an appointment, going through a credit check, or simply not getting a call answered can be the moment a customer decides to look elsewhere.

## How machine learning enhances churn prevention

All of the strategies above depend on knowing your customers well enough to act intelligently. That's where most teams hit a wall: the data they have on their own customers isn't enough to drive real personalization, and manual analysis can't keep pace with the volume of decisions a modern retention program requires. Machine learning, paired with rich consumer data, changes the equation on all three fronts.

### Richer segmentation with the Faraday Identity Graph

Traditional segmentation is limited by what you know from your own data. Machine learning expands that picture by unifying your first-party data with the [Faraday Identity Graph](/features/faraday-identity-graph) (FIG) — over 1,500 consumer data points across 240M U.S. adults and their households, sourced from the best data vendors available. FIG data spans demographics, psychographics, property, financials, life events, and more.

The result: your customer Jane Doe stops being just "purchased a dog toy last month" and becomes a fully drawn profile — dog owner, parent, household income over $200k, prefers in-store shopping. That's the difference between generic outreach and an email that genuinely lands.

### Churn prediction at the individual level

What if you could know a customer was about to leave before they even knew they were considering it? Machine learning churn scores do exactly that — predicting, on an individual basis, how likely each customer is to churn. That means you can proactively reach out to at-risk customers with personalized retention offers, rather than reacting after they've already gone quiet.

### Churn modeling in reverse: predicting loyalty and return

Most churn models are built to surface risk — who's about to leave. But the same modeling framework can be flipped to answer a different question: who is likely to *stay*, or who is likely to *come back*?

This inverse approach is especially valuable for brands with long sales cycles or inconsistent buying patterns. A tire retailer, for example, can't expect monthly purchases — but they can predict which customers are likely to return when they're ready to buy again. An eyewear brand can identify who's a loyal repeat buyer versus a one-time purchaser, even if those visits are years apart.

For subscription businesses, likelihood-to-remain scores add a layer of nuance that pure churn risk can't capture: instead of only flagging who might cancel, you can identify who's deeply engaged — and let that shape how you invest in them. A high likelihood-to-remain score is a signal to reward loyalty, not just a reason to deprioritize outreach.

The mental model shift matters here. A high score is a good thing — it tells your marketing and sales teams exactly who to lean into.

### LTV prediction: knowing who's worth fighting for

Not every at-risk customer deserves the same retention investment. LTV prediction adds a second dimension to churn scores — so you can define cutoffs for how deep a discount or how much outreach effort is actually worth it. A customer with a high churn score and high predicted LTV is worth a personalized win-back campaign. One with a high churn score and low LTV might be fine to let go.

## How Faraday predicts and prevents churn

Here's how the pieces come together in practice.

### Connect your data and define your customer groups

Start by connecting your customer data to Faraday and defining two groups: your current active customers and your churned customers. These become the foundation for everything that follows.

### Build personas to understand who your customers really are

With FIG data layered in, Faraday generates personas — distinct customer segments defined not just by purchase behavior but by the full consumer profile behind each person. Age, lifestyle, shopping habits, financial indicators — but also how long someone has been in their home, what sports they follow, what causes they donate to, and hundreds of other signals that marketing teams don't typically think to include in a churn model. That's exactly what makes them powerful: the nuances that make personalization feel personal.

### Generate churn predictions

Define your predictive outcome: who is likely to move from your active customers group into your churned customers group? Faraday builds a custom model on your data, scores every customer, and surfaces the results alongside their persona. The score tells you who is likely to churn — the persona tells you how to reach them. Together, they turn a list of at-risk customers into a targeted, personalized retention campaign.

### Deploy predictions to your stack

Once your churn scores are live, push them wherever you need them — your CRM, email platform, data warehouse, or any downstream tool. Your retention campaigns now have the intelligence to target the right people with the right message at the right time. For example, one fitness brand plugged Faraday scores directly into [Klaviyo](https://www.klaviyo.com/) and bucketed their members into three tiers: green (very likely to remain), yellow (on the fence — intervene now), and red (very unlikely to remain). That simple structure gave their sales and marketing teams a clear, actionable playbook — discounts, messaging, and campaigns calibrated to exactly where each member stood.

### Scale prevention upstream: find better customers to begin with

The most durable churn strategy is acquiring customers who were never going to churn in the first place. Use the same modeling framework — but flip the attainment cohort to your best customers instead of your churned ones — and you get a lookalike acquisition model that finds more people like the ones who stick around. And the cycle compounds: the more you learn about who churns and who stays, the sharper your models get — which means better retention, better acquisition targeting, and a clearer picture of your best customers over time. Churn modeling isn't a one-time fix; it's a continuously improving asset.

## Real results: how a subscription box brand cut churn and boosted revenue

A leading subscription box brand came to Faraday with a familiar problem. They were manually assigning monthly shipments using hand-written rules — logic informed by customer preferences, but still rigid, prone to human bias, and impossible to scale across a growing catalog.

They replaced those manual rules with Faraday's AI-powered recommendation engine, which uses a two-sided matching approach: scoring both customer affinity (how likely a specific person is to engage with a product) and product performance (how well that product has performed with similar audiences). The result is smarter pairings optimized for both personalization and retention — at scale.

The results from a rigorous A/B test:

- **\~5% lift** in revenue per customer  
- **\~3% reduction** in churn rate  
- **5x+ monthly ROI** generated with Faraday  
- **Four consecutive months** of algorithmic recommendations outperforming their manual rules

And because the models improve as more data flows through them, these aren't ceiling numbers — they're a starting point. [Read the full case study.](/blog/monthly-subscription-box-uses-two-sided-recommendations)

## Frequently asked questions

**How much first-party data do I need to build a reliable churn model?**

It varies by industry and subscription type, but in most cases a few months of customer and churn event history is enough to get started. FIG data fills in the gaps on the consumer side — so even if your first-party data is thin, the model has strong signals to work with. Faraday will tell you upfront if your data isn't sufficient.

**What's the difference between churn prediction and LTV prediction — do I need both?**

They answer different questions. Churn prediction tells you who's likely to leave. LTV prediction tells you who's worth keeping. Used together, they let you prioritize retention spend where it actually pays off — high churn risk, high LTV — rather than treating every at-risk customer the same way.

**Can churn prediction work for non-subscription businesses?**

Yes. The subscription context is common, but the same approach applies anywhere customers can stop buying — retail, home services, financial services, SaaS. If you have repeat purchase behavior and a meaningful gap between your best and worst customers, churn prediction adds value.

**How does this connect to acquisition?**

Directly. The same models that identify your highest-LTV, lowest-churn customers can be flipped into lookalike acquisition models — helping you find more people in the market who match that profile before they ever become a customer. Retention and acquisition intelligence, same platform.

**Can Faraday predict who's likely to stay, not just who's likely to leave?**

Yes — and for many businesses, this framing is more actionable. Likelihood-to-remain scores flip the model: instead of surfacing risk, they identify your most loyal customers so you can reward and invest in them accordingly. For businesses with long sales cycles or inconsistent buying patterns — think automotive, eyewear, or home services — likelihood-to-return scores serve a similar purpose, flagging who's likely to come back even if purchases are infrequent. Both approaches run on the same modeling framework as churn prediction.

## Ready to reduce churn?

If you want to talk through how predictive churn modeling could work for your business, [talk to a Context Consultant](https://faraday.ai/contact). Or if you'd rather get started on your own, [try it on buy.faraday.ai](https://buy.faraday.ai).

Learn how consumer brands use machine learning and the Faraday Identity Graph to predict churn, identify loyal customers, and build more profitable retention and acquisition strategies.

Customer churn prediction: a complete guide for consumer brands

Predictive lead scoring uses machine learning to rank your leads by their likelihood to convert — so your team knows who to call, when to reach out, and how to engage before a single conversation starts. It's the single highest-leverage place most B2C teams can add customer context to their sales motion. This guide covers how it works, how the best teams use it, and what it takes to build a system that moves the needle.

## What is lead scoring?

Lead scoring ranks leads by their likelihood to convert so sales and marketing teams can focus on the right ones. But not all lead scoring is created equal — there are two fundamentally different ways to do it, and the choice shapes everything downstream.

**Traditional lead scoring** models often rely on a point-based system where humans determine the lead scoring criteria—typically a combination of basic demographic information and online behaviors such as pageviews, email engagement, or form submissions—and assign points to leads when they meet it.

While these point-based lead scoring models can help marketing and sales teams prioritize their efforts, they come with significant drawbacks. Lead scoring criteria are often arbitrarily decided, resulting in lead scores that may not accurately reflect a lead's true potential. Additionally, because they rely heavily on first-party engagement data, lead scores often end up being a lagging indicator of a lead's conversion likelihood.

**Predictive lead scoring** uses machine learning trained on your actual conversion outcomes. The model learns which combinations of attributes — first-party behavior plus third-party consumer context — predict that a lead will convert. With the right dataset, it can score a lead's likelihood to buy before they ever visit your site or open an email.

![Image of Faraday model report features table](/images/blog/product/feature-importance.png)

Note: the image above illustrates important features surfaced in [Faraday model reports](/blog/faraday-model-reports).

| Feature | Traditional lead scoring | Predictive lead scoring |
| --- | --- | --- |
| **Method** | Rule-based (manual point systems) | Machine learning trained on past outcomes |
| **Data used** | Basic demographics, behaviors, form fills | Hundreds of behavioral, demographic, and contextual signals |
| **Scalability** | Manual tuning required per campaign | Fully automated, adaptable, and always learning |
| **Accuracy** | Assumptions-based, subjective | Outcome-driven, empirically validated |
| **Maintenance** | Static, often neglected | Continuously improved via retraining |
| **Transparency** | Clear but simplistic | Advanced but explainable (e.g., model reports) |

Predictive lead scoring is one application of a broader practice called [propensity modeling](/features/customer-behavior-predictions) — predicting any future customer behavior, not just conversion. The same techniques power churn prediction, LTV modeling, and next-best-offer recommendations.

## What you need for great B2C predictive lead scoring

The best predictive lead scoring doesn't rely on a single source of data. It combines two things: the real-world context of who your leads actually are — their financial situation, lifestyle, and household — and the actual engagements they've had with your brand. You bring one half. Faraday brings the other.

### What you bring

- **Your lead records** (identity data) — names, addresses, phone numbers, and emails — the starting point for connecting your data to real-world consumer profiles.
- **Your conversion history** (transactional data) — who became a customer, when they converted, and what they paid — defines the cohorts the model learns from: customers, leads, stale leads. Without clearly labeled outcomes, the model has nothing to learn from.
- **Your engagement history** (engagement data) — pageviews, email subscriptions, free trial signups — captures how your leads have interacted with your brand and how those behaviors relate to conversion.

### What Faraday brings

Before any modeling happens, Faraday completes your identity data — filling in the gaps in your CRM records with missing addresses, phone numbers, and emails. This alone maximizes your match rates and expands your addressable audience.

But that's only where it starts. Once your records are complete, we combine your first-party data with our third-party consumer records — 1,500+ attributes across demographic, financial, property, and lifestyle dimensions, drawn from the **[Faraday Identity Graph (FIG)](/features/faraday-identity-graph)** covering 240M U.S. adults — to build models trained on your real customers and outcomes.

The result is [custom predictive data](/blog/three-types-of-data-that-power-smarter-decisions): bespoke data points generated specifically for your business. Lead scores are one example. Others include next-best-offer predictions, churn likelihood, and lifetime value estimates.

## How B2C teams use predictive lead scoring

Predictive lead scoring is an optimizer for your sales motion at every stage — giving reps the context to know how, when, and where to engage every lead. Scores deploy in real time into your CRM, ad platforms, and routing tools, so the intelligence shows up where your team already works. Suppression and rejection are part of the toolkit, but they're the floor, not the ceiling.

Here's where the highest-leverage use cases sit, in order:

- **Lead prioritization:** This is the use case that produces the biggest gains. Predictive scores let you tier your leads and route the strongest ones to your best reps, with the most aggressive follow-up cadence. Mid-tier leads move into the right workflow. Lower-fit records shift into nurture instead of taking up agent time. American Standard worked with Faraday and outboundIQ to do exactly this — moving lead scoring from a top-of-funnel blocker to a full-funnel routing optimizer. Their [contact rate jumped from 5.56% to 20%, and high-score leads converted at 3x the rate of others](/blog/how-we-changed-american-standards-mind). [Momentum Solar took the same approach](/stories/momentum-solar) and improved call center conversions by 33%. These patterns play out across consumer verticals, but they're especially powerful in [home services and other industries](/solutions/lead-management) where contact center economics make every dial count.

- **Product recommendations:** If your predictive system supports product-specific modeling, [you can predict which products or services a lead is most likely to purchase](https://faraday.ai/templates/next-best-offer). Reps and marketing systems use those scores to make informed recommendations across every touchpoint.

- **Custom incentives and discounts:** Instead of blanket promotions, predict the likelihood of leads converting to high-LTV customers and [offer targeted incentives only where they'll move the needle](https://faraday.ai/templates/adaptive-discounting). This protects margin on customers who would have bought anyway and concentrates spend where it matters.

- **Lead suppression:** [Rank your leads, set a suppression threshold, and stop spending on the bottom tier](https://faraday.ai/templates/lead-suppression) — whether that's direct mail, paid follow-up, or sales effort. Useful, but a much smaller lever than prioritization.

- **Lead rejection:** If you buy leads from generators or aggregators, you can use predictive scoring to reject leads below a minimum conversion probability before they ever hit your CRM. Saves money, but it's the narrowest application of the technology.

Reject and suppress when it makes sense. Prioritize everywhere else.

## Why Faraday

Most data vendors give you a snapshot of who someone appears to be today. FIG includes longitudinal data spanning years of each individual's history — something no other source in the market provides. A lead who looks identical in a snapshot today may have a very different trajectory, and trajectory is what actually predicts conversion. That's what sets Faraday apart.

It's also production-ready out of the box. The data is normalized to meet modern data science requirements, so there are no expensive licensing negotiations and no raw data to clean and restructure before you can use it.

The data is also responsibly sourced — no third-party cookies, no social scraping.

## Faraday vs. CRM-native scoring

Your CRM only knows what's in your CRM. Klaviyo, HubSpot, and other platforms with built-in predictive scoring are limited to your first-party data — how leads have behaved in your system — because that's all they can see.

Faraday trains on the combination of your first-party data and 1,500+ FIG consumer attributes, identifying patterns about who your leads actually are in the real world: their financial situation, lifestyle, household, and trajectory over time. That's the difference between scoring based on what someone did on your website and scoring based on who they actually are.

## FAQ

#### Isn't predictive AI just a black box?

Not at Faraday. Every model includes explainability breakouts that document the data used, what the model was trained to predict, and how well it performed. Feature importance scores show which attributes are driving conversion likelihood, and segment insights break down what's working in human-readable terms. For a deeper technical dive, see [Seamus Abshere's post on why first-party data and RFM isn't enough](/blog/why-first-party-data-and-rfm-is-not-enough).

#### What if I don't have enough first-party data?

Most brands have more than they think. And when first-party data is genuinely sparse, FIG fills in the gaps with [verified third-party context](/blog/identity-matching-makes-or-breaks-predictions). Models can learn from behavioral and contextual signals well beyond what's in your CRM. Predictive lead scoring doesn't require perfection — it needs patterns.

#### What are the biggest challenges of running predictive lead scoring well?

Two things. First, data quality: the accuracy of your predictions depends on the completeness and consistency of the data you feed the model — particularly identity data, which determines how well your records match against third-party consumer profiles. Second, retraining: customer behavior shifts over time, and models need to be refreshed regularly to stay accurate. A good platform handles both of these for you.

#### How long does it take to get predictive lead scoring up and running?

For most B2C teams using Faraday, a working predictive lead scoring model can be deployed within days, not months. The longer pole is usually data preparation on the client side — getting clean, well-labeled historical lead and customer data into a usable format. Once that's in place, model building, validation, and deployment happen quickly.

## Ready to put context behind every lead?

If you're a B2C team that wants to see predictive lead scoring in your own data, there are two ways to start:

- **Self-serve:** [Get started on buy.faraday.ai](https://buy.faraday.ai) — upload your records and get scored leads flowing into your stack in minutes.
- **Enterprise:** [Talk to a Context Consultant](/contact) — we'll walk through your funnel, your data, and what's possible.

A practical guide to how predictive lead scoring works, what it takes to do it well in B2C, and how the best teams use it to optimize their entire sales motion.

How predictive lead scoring works — and why it outperforms traditional methods

American Standard's contact center was making 20,000 dials a day and reaching almost no one — a contact rate of just 5.56%. After partnering with Faraday and outboundIQ, that number climbed to 20%, and high-score leads converted at 3x the rate of others. Here's how it happened.

When [Eric Kozak](https://www.linkedin.com/in/eric-kozak-11b32550/) returned to [American Standard](https://www.linkedin.com/company/lixil-global/), I heard through the grapevine about his skepticism for lead scoring. And as the "lead scoring" vendor for American Standard, that wasn't exactly the signal I was hoping to receive.

But when we finally sat down to meet, I could tell immediately that I was talking to a guy who knew his stuff. He wasn't skeptical because he thought lead scoring "didn't work". He was skeptical because he thought it was just a simple filter, a "lead blocker", that couldn't do anything else.

And right out of the gate he made this position clear, saying something like: "I get that you can reject leads, but simply rejecting leads isn't really that valuable to us."

But I didn't see this exchange as a problem. In fact it presented a big opportunity. **Eric was a leader with a vision for a totally overhauled customer journey at American Standard, and Faraday was the tool he needed to achieve that goal.**

Fast forward to Q4 2025… Eric and I sat down again. This time with a different story to tell. Eric joined me, our CEO, [Andy Rossmeissl](https://www.linkedin.com/in/rossmeissl/) and [Jason Shatzkamer](https://www.linkedin.com/in/jshatzkamer/) from [outboundIQ](https://www.outboundiq.com/) to walk through exactly what American Standard (powered by Faraday and outboundIQ) did to transform American Standard's outbound process and improve their lead to sale conversion rates by 3x.

But before we get into the results, we need to understand where things started. To see why a 3x lift mattered, you first have to see just how broken the contact center was when Eric returned.

## Inside the silent contact center

_"When I walked through the contact center, it was astounding, the silence that I was hearing. We were making twenty thousand dials a day, but we weren't making contact with anyone."_

On the panel, Eric described returning to American Standard and being pleased to see that the company had grown. But he also noted one part of the operation that had not kept pace: the contact center needed immediate attention.

It did not take long for him to rule out the usual suspects. This was not a staffing problem, and it was not a demand problem. It was a contact problem. A severe one.

**At the time, their contact rate sat at just 5.56%.**

Eric needed answers, so one of his first calls was to [Jason](https://www.linkedin.com/in/jshatzkamer), outboundIQ's CEO. Jason quickly pinpointed the problem: American Standard was optimizing for calls, not conversations.

On paper the team looked busy, making thousands of dials a day, but way too few of those dials turned into connections. The silence Eric heard on the floor wasn't about effort or volume; it was about contact. The tires were spinning, sure, but the car wasn't moving.

Jason had a solution — and once the system was tuned to actually reach customers, the difference on the floor was immediate.

## Then the problem shifted

The dialing fix worked, but it uncovered the next issue. Conversations were finally happening, but conversion rates were not rising with them. The routing system was now healthy, but the logic behind it was still running on basic filtration that relied entirely on limited first party data.

As Eric dug into the numbers, something clicked. The team finally had the throughput they needed, but they did not have the intelligence to match it. **The system could move leads around, but it could not tell anyone who was actually likely to buy.** The more he optimized the routing, the more obvious it became that the data at the center of it all needed to get smarter.

## Lead scoring isn't just a filter

This was the problem Eric was grappling with when we met. He did not need a tool that rejected leads at the top of the funnel. He needed to know how to route them intelligently at every stage of the process. And from his perspective, lead scoring did not solve that problem.

But I knew this was exactly where we thrive, so I wanted to understand where the disconnect was happening. After a quick look at the setup, the answer became clear.

Sure, the Faraday integration existed in their tech stack, but none of the deeper consumer context was turned on. No behavioral signals. No financial indicators. No household insights. And because of how the integration had been structured, Faraday was indeed only acting as a filter at the very top of the funnel. In practice we were functioning exactly how he suspected. We were a blocker and nothing else.

What Eric hadn't seen yet was the depth of context sitting behind that integration. Faraday's lead scores are built on the [Faraday Identity Graph](https://faraday.ai/features/faraday-identity-graph) (FIG) — over 1,500 consumer data points spanning 240M U.S. adults and their households. That's the difference between knowing a lead's ZIP code and knowing their likely purchasing power, life stage, and propensity to convert. The data to power intelligent routing was already there; it just wasn't switched on.

<Pullquote>But Faraday is not a lead blocker. We're a lead optimizer.</Pullquote>

I talked Eric through this paradigm shift and explained that in this use case, the real value Faraday provides is using consumer context to power the routing system outboundIQ had created. High fit leads should rise to the top. Mid tier prospects should move into the right workflows. Lower fit records should shift into slower nurture paths instead of taking up agent time. Smart routing only works when it is powered by smart data. And yes, we can reject the worst leads before purchase, but that is just the tip of the iceberg.

Once we walked through how the lead score could prioritize instead of reject, Eric's skepticism had dried up and was replaced with excitement, he was ready to test it out\!

## Real results

When we circled up with Eric and Jason to hear how the test had gone, the numbers spoke for themselves:

- **Contact rate lift:** Contact rate increased from 5.56% to 20%
- **Conversion rate:** High Faraday score records converted at 3x the rate of others

Nothing had changed about their leads, the only thing that had changed was getting more context to power smarter dialing and routing. Transformational results.

Beyond the top-line numbers, Eric described the shift on the floor as well. When the contact rate increased, the energy changed immediately. As he described it:

_"The energy just goes through the roof… the people are encouraging each other. That unquantifiable morale boost is critical for a contact center… Turnover is down. The momentum is back. People have wins again, and that changes everything."_

This is what it looks like when your stack finally works the way it should. It might not appear on a dashboard, but operators feel it instantly. Real success is not just about the numbers. It is about creating a productive, motivated, and winning environment. You know how to hire the right people, but you need to give them the right tools to succeed too.

## Strategic recommendations for making your contact center actually work for you

If you want to bring the same lift American Standard saw into your own organization, start with the foundation: if your contact rates are struggling before you ever get to lead scoring, that's a dialing and routing problem first — and partners like outboundIQ specialize in exactly that. Get that right, then layer in intelligence.

### 1. Layer consumer context into 1 (or 2) points of your customer journey

Rule-based filters and crude sorting cannot keep up with modern compliance pressure or consumer behavior. Use a custom-built lead score at two points in your customer journey:

- If you are struggling with poor lead quality: add the lead score into ActiveProspect (read our article about how to do it [here](/blog/dtc-homeservices-uses-credit-score-proxy))
- If you are struggling with low conversion rates: add a lead score or credit score proxy to help reimagine how you prioritize leads in your contact center.

A lead score is simply a fancy data point that gives your dialer and CRM the intelligence they were designed to use. It's powered by the same consumer context in the Faraday Identity Graph — the property, demographic, financial, and life-event signals that tell you who's actually likely to buy. Obviously I'm partial to Faraday, but there are others out there. Find someone who can build a model quickly and iterate. If you can't refine the model quickly, you are going to think this process isn't working. It's the wrong conclusion, good modeling takes time and iteration.

If you run a home services contact center specifically, we've written a deeper playbook on where to layer predictive data into your funnel — from rejecting junk leads before purchase to qualifying appointments with financial signals. [Read how home services call centers stop burning money without lead scoring.](https://faraday.ai/blog/homeservices-call-centers-need-lead-scoring)

### 2. Measure everything

Your business is too critical for suffering through the gauntlet of making crappy tools work. If you simplify your tech stack, you can actually measure what matters: the current conversion rate at every stage of the funnel. When high-score leads rise to the top, reps focus their time where it matters, credit declines shrink, and your entire stack becomes more effective without adding new complexity.

## Bringing it all together

American Standard did not solve their problem by dialing harder. They solved it by putting their customers at the center and building a tech stack around figuring out how to reach the right people at the right time. Understanding which records were worth calling and focusing on those ones first enabled them to transform their journey.

[You can hear Eric, Jason, and Andy walk through this transformation in their own words.](https://faraday.ai/blog/smarter-data-dialing-overhauled-american-standards-contact-strategy)

## Frequently asked questions

**If the leads didn't change, why did conversion improve so much?**

That's the key insight from American Standard's story. The leads were the same — what changed was the context layered on top of them. By turning on Faraday Identity Graph signals (purchasing power, household attributes, propensity to convert), the routing system could finally tell which records were worth prioritizing. High-fit leads rose to the top of the queue instead of getting lost in the volume, so reps spent their time where it actually paid off.

**What's the difference between lead blocking and lead optimization?**

Lead blocking is a top-of-funnel filter: reject the worst leads before you pay for them. That's useful, but it's only the tip of the iceberg. Lead optimization uses the same predictive score across the *entire* funnel — prioritizing high-fit leads, routing mid-tier prospects into the right workflows, and shifting lower-fit records into nurture paths instead of burning agent time. Same data, far more value.

**How long does it take to see results?**

It varies by data volume and how many points in the funnel you're scoring, but American Standard saw a measurable lift once the deeper FIG context was switched on and routing was tuned around it. The bigger factor is iteration — good modeling takes a few cycles to refine. If you expect a perfect model on day one and bail when you don't get it, you'll draw the wrong conclusion.

**Does this only work for home services or contact centers?**

No. The contact-center framing is where American Standard saw it, but the underlying principle — using consumer context to prioritize instead of just reject — applies anywhere you're ranking leads or customers. That said, if you do run a home services funnel, we've mapped out the specific plays in our [home services lead scoring guide](https://faraday.ai/blog/homeservices-call-centers-need-lead-scoring).

## Ready to make your contact center work harder?

If you want to chat through how this could transform your organization, [talk to a Context Consultant](/contact). And if you'd rather dive in and start scoring leads yourself, you can [get started on buy.faraday.ai](https://buy.faraday.ai).

Faraday and outboundIQ helped American Standard shift lead scoring from a simple top-of-funnel blocker to a full-funnel routing optimizer, boosting contact rates from 5.56% to 20% and driving 3x higher conversion for high-score leads.

Don't trust lead scoring? Here's how we changed American Standard's mind, and how you can get the same results

A holdout evaluation is the industry-standard method data scientists use to validate predictive models: train on historical data, set aside a portion the model has never seen, then check whether the highest-scoring leads actually converted at higher rates. If they did, the model works.

If you're evaluating predictive data vendors, this is the test worth asking for. With tight budgets and increasing pressure to show ROI, you can't afford to bet on unproven tools or spend months figuring out if a vendor will live up to the promise. You need a way to separate legitimate predictive tools from the noise.

That's what our holdout evaluation does. It's not a methodology we invented — it's an industry standard used to validate [custom predictive data](https://faraday.ai/blog/predictive-lead-scoring-b2c), including [propensity-to-buy scores](https://faraday.ai/blog/propensity-to-buy-machine-learning) used across lead buying, direct mail, and personalization.

Below we've outlined how it works and what to look for when evaluating whether it's delivering real value.

## What is a holdout evaluation?

A holdout evaluation is a method for validating predictive models by comparing their predictions to real-world outcomes. We **set aside a portion of recent conversion data**—data the model has never seen before—and use it as an independent benchmark, against which predictions can be compared.

This "holdout set" acts as a real-world proving ground. It is untouched until after the predictive model has been trained.

Meanwhile, the rest of your historical data is combined with rich consumer data from the [Faraday Identity Graph](https://faraday.ai/features/faraday-identity-graph) ([which includes over 1,500 data points on 240M+ U.S. adults and their households](https://faraday.ai/features/faraday-identity-graph)). We use this combined data to train the new predictive model.

Training a predictive model simply means finding patterns in historical conversions. In other words, what can we learn from your past customers to predict future outcomes? Once trained, the model generates predictions—scoring leads based on their likelihood to convert and only after these predictions are made do we compare them to the **real conversion rates** in the holdout set.

If the highest-scoring leads actually convert at higher rates than lower-scoring ones, the model is validated as accurate and reliable.

Holdout evaluations ensure predictive models aren't just theoretical—they work in the real world, using real data, to deliver real results.

## How holdout validation works at Faraday

We've streamlined this validation process to be transparent and straightforward:

1. ![customer conversion data](/images/blog/holdout_testing_diag/holdout_testing_diag-01.png)

   #### Step 1: Set aside a holdout group

   We start with your recent [lead conversion data](https://faraday.ai/blog/first-party-vs-third-party-data-marketing). Before any processing occurs, we reserve a portion of your most recent leads as a "holdout set"—data that the model will never see during training. This ensures we have an independent benchmark for evaluating predictions.

2. ![merged with FIG](/images/blog/holdout_testing_diag/holdout_testing_diag-02.png)

   #### Step 2: Enrich with consumer data

   With the holdout set reserved, we then combine remaining data with Faraday's [built-in consumer data](https://faraday.ai/features/faraday-identity-graph) sourced from the Faraday Identity Graph (FIG). This expanded dataset provides a deeper view into key predictive signals, like property and demographic information, life events, buying habits, and more.

3. ![model is trained and predictions begin](/images/blog/holdout_testing_diag/holdout_testing_diag-03.png)

   #### Step 3: Train the model and score leads

   Using this enriched dataset (excluding the holdout set), we train a predictive model to identify patterns in conversion behavior—learning from consumer signals, demographic data, and behavioral patterns.

   Once the model is trained, it scores each lead based on its likelihood to convert, generating a ranked list of opportunities—from high-value leads to those less likely to convert.

4. ![model is validated](/images/blog/holdout_testing_diag/holdout_testing_diag-04.png)

   #### Step 4: Compare predictions to real outcomes

   Finally, we put the model to the test: we compare its predictions against the real outcomes in that untouched holdout set. If the highest-scoring leads don't convert at significantly higher rates, we go back to the drawing board.

## How to know if a holdout validation was successful

A holdout validation succeeds when predicted scores align with actual conversion rates across segments. Look for these three signs:

1. **Clear correlation between predicted probability and actual conversion rates**  
   The model's predicted likelihood to convert should directly align with the actual conversion rates. If the predictions are accurate, high-probability leads should convert at a noticeably higher rate than lower-probability ones.

2. **High-scoring leads convert at significantly higher rates**  
   The top-scoring leads should consistently perform better than those ranked lower. This demonstrates that the model is identifying truly valuable leads.

3. **Consistent performance across different customer segments**  
   The model should show reliable results across various segments—whether you're targeting different demographics, regions, or other customer profiles. If the model works well across a broad set of data, it proves its reliability and scalability.

When a model meets these criteria, it's ready to generate custom predictive data points—whether that means optimizing [lead buying](https://faraday.ai/docs/how-to/use-cases/leads), refining your [direct mail strategy](https://faraday.ai/templates/lead-suppression), or enhancing your email [personalization](https://faraday.ai/templates/personalization) approach.

## Questions you should ask us

We want you to push us on what matters most for your business. Here are key questions to evaluate if we're the right fit:

1. **How does this model's performance compare to the lift you are seeing with similar businesses in our industry?**

   While we can't share client specifics, we can provide insight into broader trends and expected performance benchmarks.

   → **Why this matters:** Understanding industry benchmarks helps you gauge whether custom predictive data is truly a competitive advantage for your business.

2. **How much historical data do we need for reliable predictions?**

   The answer varies by business model and conversion volume. We'll help determine whether you have enough data—or how to supplement it if needed.

   → **Why this matters:** Predictive models work best when trained on strong data. Knowing the threshold for reliability ensures you're making data-driven decisions with confidence.

3. **What happens if the model doesn't perform as expected?**

   We have a structured iteration process to refine models when necessary, ensuring they reach meaningful predictive power.

   → **Why this matters:** No model is perfect on the first run. Understanding the process for continuous improvement ensures long-term success.

4. **How will this integrate with our existing marketing stack?**

   We'll provide technical specs, discuss implementation timelines, and outline the resources needed to get up and running.

   → **Why this matters:** Seamless integration means faster time-to-value and less friction in your workflow.

## Why holdout evaluations work

Holdout evaluations work because they test a model against data it has never seen — making it impossible to game the results. Any predictive data vendor can show you strong in-sample performance. Holdout testing is the check that confirms those patterns hold in the real world.

At the end of the day, three things matter to us:

1. **Confidence before commitment:**  
   You see proof before you commit

2. **Higher marketing efficiency:**
   Your marketing dollars go to the leads most likely to convert

3. **Stronger sales performance:**
   Your sales team focuses on opportunities worth their time

If the validation doesn't show meaningful lift, we'll tell you. If it does, we'll quantify exactly what that means for your business.

Interested in generating value you can trust? [Talk to a Context Consultant](/contact).

## Frequently asked questions

**What's the difference between a holdout evaluation and an A/B test?**

A holdout evaluation validates that a predictive model works—it confirms that high-scoring leads actually convert at higher rates before you commit to anything. An A/B test measures the impact of a decision you've already made (like sending a campaign to scored vs. unscored leads). Faraday runs the holdout evaluation first so you have confidence in the model itself, not just the downstream campaign.

**Do I need a large dataset for holdout testing to be meaningful?**

Not necessarily. The threshold varies by industry and conversion volume. In many cases, a few months of lead and conversion history is enough to build a reliable model. We'll tell you upfront if your data isn't sufficient—and if it's borderline, we can discuss ways to supplement it with data from the [Faraday Identity Graph](https://faraday.ai/features/faraday-identity-graph).

**What happens to my data during the validation process?**

Your first-party data is used solely to train and validate the predictive model—it is never shared with other clients or used to enrich Faraday's broader data assets. The holdout set is kept completely separate from training and is only used as an independent benchmark after predictions have been made.

Faraday uses holdout testing to validate custom predictive data before you commit, ensuring it delivers real, measurable results.

Predictive data generates tangible value, and we can prove it

"If one record is wrong, how can I trust the rest?" 

We sometimes hear this from clients, and we get why. You open a freshly enriched customer file, spot something you know is wrong — your own name at an address you left five years ago, a coworker tagged with an income that's clearly outdated — and suddenly the whole file feels suspect.

But that instinct — to judge the whole file by one bad record — is exactly the **wrong way to evaluate consumer data**. And it leads to worse decisions, not better ones.

## Spot-checking is the wrong test for data quality

First things first: spot checking for 100% accuracy is not a valid way to review data quality in a customer file.

Whether you have spotted an incorrect record in a 100k person audience for direct mail or you've identified an incorrect attribute from your customer file, the key is to go back to first principles: your ultimate goal is likely to drive an increase in conversion rates or reach more people. One bad row isn't going to spoil the bunch.No dataset at scale is 100% accurate, however, our mission is to have the best data set on the market. We rely on many upstream vendors and getting feedback on their data quality only helps us get better at our mission. 

## What to do when you spot a bad record

When you find one bad record, here’s what to do: 

- First, consider your use for the data. If the file is 95% accurate instead of 100% accurate, can you still achieve your goal?  
- Second, report it to us via [faraday.ai/data-corrections](http://faraday.ai/data-corrections). When a client flags something, the issue almost always originates with one of our upstream vendors — but because it ended up in your file, it's our job to act on it. Our team reviews every submission, escalates with the upstream vendor when needed, and applies corrections back into the pipeline. It's not a quality guarantee — it's a correction loop, and it's how the dataset stays trustworthy over time.

Our number one goal is to provide data that drives meaningful lift on your campaigns, and ultimately delivers significant ROI for your company. (Side note: we strive to be assessed by the ROI we deliver, check out some of our [case studies](/stories) to see just how much lift we generate). 

The reality is that chasing 100% accuracy at this scale would mean slower files, longer QA cycles, higher prices, and less freshness — all in service of a number that doesn't change the campaign outcome. So the right question isn't "is every record correct?" — it's "does this file help me reach more of my best customers than a random list would?"

## How to actually evaluate consumer data

If you want to know whether your file is doing its job:

* **Run a holdout test.** Pick a random subset of your enriched audience and a comparable subset without enrichment. Run the same campaign to both. The lift between them is your real answer.  
* **Look at downstream metrics.** Response rate, conversion rate, CPA, ROAS. These are the only metrics that reflect whether the file is working.  
* **Evaluate match rates in context.** Match rates depend on what identifiers you bring to the table (check out our [thoughts on this](https://faraday.ai/blog/wheres-my-99-match-rate)). Compare against realistic benchmarks, not vanity numbers.

None of these require inspecting individual rows. All of them tell you something a spot-check cannot: whether the file is actually delivering value.

## The bottom line

It’s just like life, don’t let one data point throw you off your game\! Stay focused on measuring what matters: response rate, CPA, lift versus a random list. If the numbers clear your benchmarks, the file did its job. If they don't, that's the signal worth investigating — not the one weird row that caught your eye.

When you do find a record that's off, [flag it](/data-corrections). We want to know. And if your campaign results aren't where they should be, [talk to us](/talk-to-sales). That's the conversation we want to be having.

When you spot a bad record in a 150,000-person file, your instinct is to question the whole dataset. The math says otherwise. Here's how to actually evaluate data quality.

Found a wrong record in your data? Here's what it actually means

Updated: May 2026

In [our last post](https://faraday.ai/blog/b2c-customer-data-stop-guessing-start-executing), we made the case that B2C brands are leaving value on the table because the data they have about their customers isn't complete enough to act on. Brands know what customers bought. They have a name, probably an email address and physical address. What they’re missing is the broader context — who the customers actually are, what their financial situation looks like, where they are in life — and that's exactly what's missing when campaigns underperform and outreach falls flat.

The first step to closing that gap is data enrichment: appending verified, high-quality external datapoints to your existing customer records so you're working with a complete picture instead of a partial one. That's what Faraday can do for your brand, and this post covers what data is available, how the enrichment process works, and how to get started.

## What data is available for enrichment through Faraday

Faraday's data enrichment is powered by the [Faraday Identity Graph (FIG)](http://faraday.ai/catalog) — a comprehensive, continuously maintained dataset covering **240M US consumers and their household, with over 1,500 verified, pre-cleaned, and feature-engineered datapoints.** Every record is tied to a verified consumer identity, which means the data you enrich actually belongs to the person you're trying to reach.

FIG’s datapoints cover a range of categories including:

- **Demographics** — age, gender, household income, education level, marital status  
- **Property and housing** — homeownership status, home value, roof type, garage spaces, length of residence  
- **Financial signals** — Credit Score Proxy, spending percentiles, investment behavior, debt indicators  
- **Life events** — recent move, new homeowner, new parent, marriage, retirement  
- **Lifestyle and interests** — hobbies, brand affinities, vehicle ownership, charitable giving patterns  
- **Contact and identity** — verified name, address, email, phone, household composition

We recently completed a major refresh of the entire FIG catalog as part of FIG v2, updating datapoints across every category. You can browse everything that's available at [faraday.ai/catalog](https://faraday.ai/catalog) or purchase data directly at [buy.faraday.ai](https://buy.faraday.ai).

## How data enrichment actually works

A data enrichment is the process of matching your existing customer records to FIG and appending the datapoints you select. The process is straightforward:

1. **Select your datapoints.** Choose the specific attributes that address your business question — whether that's segmenting your audience, scoring leads, expanding into direct mail, or understanding your ICP. You can select individual fields or use one of Faraday's pre-built essentials packages, curated by vertical for home services, retail and ecommerce, and financial services.  
     
2. **Match against FIG.** Faraday resolves your customer records to verified identities in FIG using a combination of name, email, phone, and address fields. This is deterministic matching — not probabilistic — so you can trust the appended data actually maps to the right person.  
     
3. **Receive enriched records.** Your data comes back in a clean, structured format ready to load into your CRM, CDP, data warehouse, ad platform, or wherever you need it.

[![Data-appends diagram](/images/blog/data-appends-diagram.png)](/images/blog/data-appends-diagram.png)

## How brands use data enrichment

Data enrichment can unlock your entire marketing strategy — [better segmentation](https://faraday.ai/blog/smarter-segmentation-for-banks), [smarter messaging](https://faraday.ai/blog/how-dtc-brands-build-living-personas-faraday-heatseeker), [new channels](https://faraday.ai/blog/match-boost-improves-enrichment) you couldn't reach before. But here are the three ways we see brands put it to work first:

**Fill contact gaps to expand your reach.** A brand that's built its entire customer relationship through email has email addresses — but no physical addresses, no phone numbers. If they want to run direct mail or a call center program, enrichment is what makes that possible. You get the missing contact fields enriched to records you already own.

**Understand your customers beyond the transaction.** Your purchase history tells you what people bought. It doesn't tell you who they are — their financial situation, life stage, household context, or what else they might be in the market for. Enrichment fills that picture in so your segmentation and messaging are based on something real.

**Research your market before you move into it.** If you're evaluating a new geography, a new channel, or a new customer segment, enrichment lets you analyze the landscape before you commit — understanding what the population looks like and where the opportunity actually is.

## Why not just find the data yourself?

Some companies attempt to enhance their marketing efforts by purchasing third-party data directly from various vendors themselves, rather than using packaged and processed services like Faraday’s data enrichment. However, this approach often presents significant challenges:

**Contracting complexity**: Sourcing offline data can involve lengthy negotiations and large upfront commitments.

**Security risks**: Handling sensitive data requires strict compliance frameworks, such as SOC-2 certification. Unlike many vendors, all Faraday data enrichment is privacy-compliant, ethically sourced, and adhere to high standards of data protection, ensuring your customers' personal information remains secure while providing actionable insights.

**Regulatory challenges**: Companies storing offline data must navigate consumer privacy laws, including opt-out and deletion requests. Faraday manages these complexities for you, so you can focus on what matters—leveraging new insights to drive revenue.

**Data science overhead**: Raw offline data often requires significant processing, feature selection, and bias mitigation before it can be effectively used. This can result in costly time and resource investments as companies try to integrate third-party data into their models.

Faraday handles all of this before the data reaches you. The trade-off isn't 'Faraday or DIY' — it's spend your team's time building and maintaining data infrastructure, or spend it running better campaigns.

## How to get started

Getting started with Faraday’s data enrichment is simple. Once you’ve identified the data points you need, we’ll work with you to integrate them into your existing systems. Whether you’re looking for a one-time data boost or ongoing data enrichment, our flexible pricing and easy integration make it simple to scale.

If you're not sure where to start, browse what's available at our new data [catalog](faraday.ai/catalog) — or talk to a [context consultant](/talk-to-sales) who can help you identify which datapoints are most likely to move the needle for your specific business before you commit to anything.

And if you just want the data today, you can find it yourself in our new self service [marketplace](http://buy.faraday.ai).

Faraday's data enrichment appends verified third-party datapoints from FIG to your customer records, filling contact gaps, deepening customer understanding, and powering smarter targeting and expansion decisions.

How data enrichment actually works with Faraday, and what data you can find

If you use Faraday — for predictions, appends, personas, or anything in between — you've been using the [Faraday Identity Graph (FIG)](/catalog) whether you knew it or 
not. FIG is the foundation everything runs on: a continuously refreshed collection of 1,500+ attributes on 240 million US consumers, curated and normalized from leading data vendors. Every lead score, every data append, every audience we help you build draws from it.

Consumer data is the foundation of modern marketing — and what businesses need from that foundation has changed. Models are getting more sophisticated and need richer, time-aware inputs to keep getting better. Marketers want to act on signals the moment they emerge, not weeks later. Data science teams want documentation and metadata they can actually build with. And operations teams want predictable release cadences that fit their change management processes.

FIG v2 is built for where consumer data needs to go next. It's a complete rebuild of our consumer data system that introduces historical data, an expanded and redesigned attribute catalog, continuous data refreshes, and release pinning — so the foundation underneath your marketing keeps pace with what your business is asking of it.

Here's what's new:

## 1. Hundreds of new attributes, with richer signal across the board

We expanded our underlying [attribute set](/catalog) significantly — adding hundreds of new data points to capture consumer signals we couldn't surface before. We also harmonized unwieldy groups of related attributes into cleaner replacements with more expressive, consistent names.

**What this means for you:** richer raw material for every prediction, append, and audience you build. More signal in the data means models can pick up on patterns that were previously invisible, which translates directly to better prospect targeting and higher marketing ROI.

## 2. A redesigned catalog you can actually navigate

We rebuilt our [data catalog](/catalog) from scratch around the expanded attribute set — retiring data points that weren't pulling their weight and reshaping what remains around the attributes that consistently drive better prediction quality. Every attribute now comes with detailed documentation: descriptions, visual data distributions, and technical metadata designed for both marketers and data scientists.

**What this means for you:** less time hunting for the right attribute, faster model iteration for your data science team, and a catalog that both marketers and analysts can actually use without a data science degree. 

## 3. Historical data

FIG v2 carries the [full history of values for every attribute](/blog/fig-v2-historical-data), each one timestamped and annotated with precision and source metadata. Rather than knowing someone's income is \$85,000 today, you now know it was \$62,000 five years ago, \$74,000 three years ago, and \$85,000 today.

This is possible because Faraday has been a careful steward of consumer data for over a decade. We've been collecting, normalizing, and preserving these observations the whole time. FIG v2 is the first time we've been able to surface all of that accumulated history in a structured, ML-ready format — every attribute is now a timeline of events rather than a single point in time.

**What this means for you:** dramatically better predictive models. Models trained on current-state data are actually learning what people look like *after* they convert — not before. Take a home remodeling brand trying to predict who'll buy new windows: after someone buys, there's a contractor transaction, possibly a HELOC, elevated home-improvement spend. Those signals weren't there when they were still a prospect. If your training data captures people as they look *after* converting, the model goes hunting for prospects who already bought.

With FIG v2, we can press rewind and train on what a person looked like just before they became a lead — so models learn what a genuine pre-purchase prospect looks like and get dramatically better at finding more of them.

For data science teams, every historical observation comes with the metadata you need to use it directly: data types, statistical types, directionality guidance, coverage breakdowns. If your team builds internal models and wants rich consumer context that goes back in time, this is the dataset to do it with.

## 4. Fresher data, delivered continuously

Our new infrastructure [ingests fresh data from vendors as soon as we receive it](/blog/fig-v2-freshness-release-pinning) — no more waiting for scheduled monolithic releases.

**What this means for you:** less stale targeting and fewer wasted impressions. This matters most for time-sensitive signals like recent movers, life events, and income changes — the kinds of triggers where a few weeks of lag can mean the difference between reaching someone at the right moment and missing them entirely.

## 5. Release pinning

We've heard from many of you that you love getting the most current data automatically. We've also heard from teams with strict change management requirements who need more control over when their data updates. FIG v2 supports both. You can now [pin your account to a specific FIG release](/blog/fig-v2-freshness-release-pinning), and new updates won't apply until you move your pin date forward.

**What this means for you:** stability when you need it, freshness when you want it. Teams with regulated workflows, locked-in model versions, or careful release calendars get to upgrade on their own schedule — without sacrificing access to the latest data when they're ready.

### **What this means for your account**

* **FIG v2 is live today for all new accounts**  
* **Existing accounts will be migrated over the coming months** — no account will be moved without coordination with your team  
* **Your current account keeps running as-is** throughout the process — there's no disruption until you're ready  
* **Questions about your migration or timeline?** Reach out to your account manager

## Ready for even better customer data?

FIG v2 is a fundamental shift in what consumer data can do inside Faraday — better models, fresher data, hundreds of new attributes, a catalog you can actually navigate, and a release cadence that respects your operations. 

We're working through migrations as quickly as we can and will be in touch when it's your turn. In the meantime, if you want to go deeper on what's changed and why, reach out to your account management team or [read the full FIG v2 series](/blog/series/figv2).

FIG v2 is a complete rebuild of Faraday's consumer data system, introducing historical data, redesigned attributes, fresher updates, and new tools to make customer data easier to discover, deploy, and operationalize.

Now live: FIG v2 - the biggest upgrade to our customer data system, ever

If you've been following our FIG v2 series, you've already read about the [overhauled data catalog](/blog/fig-v2-data-catalog) and the [historical data upgrade](/blog/fig-v2-historical-data) that makes your models dramatically more accurate. Today I want to close out the series with something that might sound like a backend plumbing detail—but has real consequences for anyone running production pipelines on Faraday data: how we handle data freshness, and the new release pinning controls that come with it.

## How most identity graphs handle updates (and why that's a problem)

To understand why this matters, it helps to know how data updates typically work in the identity data industry.

Most identity graphs operate on a monolithic release model. Providers work with many data vendors, each of whom delivers updated data snapshots on their own schedule. The standard approach is to either wait for all vendors to deliver their latest data, or simply release on a fixed schedule—monthly, quarterly—regardless of when the underlying data actually arrived. Either way, the result is the same: a gap between when data is collected in the real world and when it's available to you.

The obvious problem: every day you spend waiting is another day the data sits there getting older. The release cycle itself introduces latency that has nothing to do with how fresh the underlying data actually was. **FIG v2 eliminates that bottleneck entirely.**

## Rolling releases: freshness without the wait

Now with FIG v2, instead of waiting for a full vendor sweep before publishing an update, we ingest each vendor's data the moment it arrives. Each new observation is processed, normalized, and added to the appropriate attribute's history vector—timestamped, sourced, and immediately available—without waiting for anything else.

The result is rolling releases. Every time even a single vendor delivers updated data, that data flows into FIG v2 right away. There's no more monolithic "release event" to wait for. The gap between when data is collected in the real world and when it's available to you in Faraday shrinks dramatically.

For most attributes, this is just a nice quality-of-life improvement—fresher data, more often. But for certain attributes, it's much more significant. Pre-mover signals, for example, are almost entirely time-sensitive. A pre-mover flag that's six weeks stale is considerably less valuable than one that arrived yesterday. With rolling releases, those high-velocity attributes stay sharp.

## Release pinning: you control when freshness flows in

Fresher data is almost always better — but there's a legitimate exception. If you're running a mission-critical pipeline and you've carefully validated the data distributions it depends on, constant rolling updates can feel like a liability. Think an automated risk scoring model at an insurance company, or a lead prioritization pipeline that's deeply integrated into a sales team's daily workflow — cases where an unexpected shift in data distributions could quietly change outputs in ways that are hard to detect and costly to diagnose. If you're in one of those situations (or something similar), you don't want things shifting under your feet without warning.

That's what release pinning is for.

In FIG v2, you can pin any individual pipeline to a specific date. What that means concretely: any new consumer data incorporated into FIG *after* that date is simply ignored for that pipeline. The pipeline keeps running—against exactly the data it was validated against—until you decide you're ready to move it forward.

The workflow this enables is deliberate and controlled. Say you're running a critical pipeline and a new round of vendor data has just been ingested into FIG. Rather than letting it flow in automatically, you:

1. **Make a copy of the pipeline.** The copy runs against the newer data.  
2. **Do your validation.** Check distributions, spot-check outputs, confirm that everything downstream still behaves as expected.  
3. **When you're satisfied, advance the pin date on your production pipeline.** The new data flows in. The copy gets retired.

You can do this on whatever cadence makes sense for your team—monthly, quarterly, or whenever a new vendor refresh comes in that you want to evaluate. You get the benefit of fresher data without ever being surprised by it.

For teams that don't need this level of control, the default behavior is simple: data stays fresh automatically, with no manual steps required.

## Freshness and stability, on your terms

The two ideas here—rolling releases and release pinning—are designed to work together. Rolling releases make FIG v2 the freshest consumer data platform we've ever built. Release pinning makes sure that freshness doesn't come at the cost of predictability for the teams who need it most.

Together, they represent something we think is genuinely new in the market: a consumer data system that treats data freshness not as a fixed property you accept or reject wholesale, but as a dial you can tune per-pipeline based on your actual operational needs.

If you want to learn more or start testing these capabilities in your account, reach out to your account management team. And if you missed the earlier posts in this series, you can catch up on the [data catalog deep dive](/blog/fig-v2-data-catalog), the [historical data deep dive](/blog/fig-v2-historical-data), or the [freshness and release pinning deep dive](/blog/fig-v2-freshness-release-pinning) on our blog.

And if you're new to the platform, reach out to [talk to a context consultant](/contact) to see what FIG v2 can do for your data.

FIG v2 features continuous rolling releases for immediate data freshness, while introducing release pinning to give teams precise control over when new data enters their mission-critical pipelines.

Deep dive: FIG v2 data freshness and release pinning

Most marketing teams aren't starting from zero. You have CRM records, campaign data, transaction history — probably more than you can act on. But having data and having the *right* data are two different things. And the gap between them is where budget gets wasted, campaigns underperform, and your team ends up doing work that better data would have made unnecessary.

The right question isn't "how much data do I have?" It's "do I have the data that lets my team stop compensating and start executing?"

## What goes wrong when the data is wrong

The same patterns show up for consumer brands across industry verticals — sometimes as a campaign that outright fails, more often as the quiet drain of time your team spends manually compensating for gaps the data should have filled. Here’s how that might look for you:

* A brand builds their entire acquisition strategy around a customer profile that turns out not to reflect their most valuable buyers — and spends years optimizing for the wrong person.  
* A paid media campaign misfires because the targeting is built on assumption-based profiles instead of actual purchase signals — the right message reaches the wrong people, and someone spends two weeks in spreadsheets trying to figure out why the numbers look off.  
* An inbound sales team wastes hours on leads that were never going to convert, because there's no signal to distinguish high-value prospects from low-value ones until a rep is already on the phone.  
* An outbound sales team can't reach half their customer list because the records are missing valid emails, phone numbers, or addresses — and somebody has to go fix that before anything else can happen.

In every case, the brand had data. They just didn't have the data that would have prevented the wrong decision.

## What the right data actually looks like

The right data isn't a single thing. It's three layers, each doing distinct work:

**Verified identity data.** Clean, foundational fields — name, phone, email, address. These make sure you can actually reach your customers, eliminate duplicates, and build a complete view of each household. Everything downstream depends on this being right.

**Customer profile datapoints.** Curated context like household income, homeownership, Credit Score Proxy, and industry-specific essentials packages (see our offerings for [home services](/blog/homeservices-essential-data), [retail/ecommerce](/blog/retail-ecom-essential-data), and [financial services](/blog/finservices-essential-data)). Use these to segment audiences, tailor messaging, and prioritize outreach based on signals that have actually been shown to drive value in your vertical. 

**Custom predictive signals.** Predictive scores trained on your specific outcomes and customers — [likelihood to convert](/templates/lead-prioritization), [best product recommendation](/templates/next-best-offer), [persona assignment](/docs/abstractions/persona_sets). Because they're built on your data, not generic benchmarks, they reflect your business and get sharper over time. Use them to tell your team exactly where to focus: which leads to call first, which product to surface, which customers to retain.

The key is knowing which layer solves which problem. You don't throw all of it at your CRM at once and see what sticks. An informed data strategy means identifying which specific categories of data — and which specific datapoints within those categories — actually drive value for your business right now. When you get that right, here's what becomes possible.

## What this actually changes

Once you have the right data foundation, the same problems that were draining time and budget become solvable:

* **Know your real ICP.** Instead of building an acquisition strategy around an assumed customer profile, you have the data to understand who actually buys from you and where to find more of them.  
* **Personalize at scale.** Instead of generic messaging that treats every customer the same, you have the context to tailor offers, timing, and creative to the specific people most likely to respond.  
* **Score and prioritize leads.** Instead of burning hours on leads that were never going to convert, you have predictive signals that tell your team exactly where to focus their time.

## Ready to get the right data?

Most brands we talk to are a few targeted datapoints away from campaigns that actually work. Explore what's available at [buy.faraday.ai](http://buy.faraday.ai) — or [talk to a context consultant](/talk-to-sales) if you want help figuring out exactly which ones move the needle for your business.

Faraday gives B2C brands the verified identity data, customer profile attributes, and predictive signals they need to stop patching data gaps and run campaigns that actually work.

Get the B2C data your brand needs to stop patching gaps and start executing

Even the most powerful model in the world can’t drive real value if it doesn’t know who it’s talking to. As our CEO, [Andy Rossmeisl](https://www.linkedin.com/in/rossmeissl), [put it](/blog/ai-needs-consumer-data-to-give-context):  “Without the specific identity and history of the customer it’s talking to the world's most advanced AI can do little more than offer a polite, generic greeting."

That's not a model problem. It's a context problem.

## The real issue isn't your model — it's what your model knows

Most AI agents start every conversation blind. Without history or clear signal about who's actually on the other end. So they default to generic responses. Not because they're poorly built, but because they have nothing to work with.

The instinct is to throw more data at the problem. More attributes. More integrations. More inputs. But raw data doesn't tell your agent anything useful about this specific person, right now. You can have a thousand attributes about someone and still not know whether they're price-sensitive, in-market, or a high-value prospect worth a personalized offer.

[As Andy puts it](/blog/ai-needs-consumer-data-to-give-context): "Context is the guiding light that turns a cold interaction into a warm conversion. It is the vital distinction between having a list of names and possessing the clarity to know exactly who a prospect is, and how to best engage them."

That shift — from dumping data at your model and hoping it figures things out, to feeding it curated signals — is what separates agents that plateau from agents that keep getting smarter.

## What Faraday does

Faraday enriches incoming identities against the [Faraday Identity Graph](/catalog) — 1,500+ consumer attributes across 240 million U.S. adults. You send us PII. We return a rich consumer profile, appended to your customer record, ready to pass directly to your agent or model.

For you, it’s as simple as a clean API call that turns an anonymous identity into someone your agent actually knows.

## What this means for what you're building

**Your agent personalizes from message one.** No more [cold-start](/stories/dtc-retailer) problems. The moment an identity hits your system, it can be enriched with the context your agent needs to skip the generic greeting and get to the relevant conversation. This is exactly what [Hazel](/stories/hazel) did with our data — their agentic marketing platform went from knowing nothing about a user to having real customer context on every interaction, with a working prototype in a single day.

**Your agent acts on signal, not noise.** Curated traits — household composition, lifestyle indicators, behavioral factors — slot directly into your agent's reasoning. Platforms like [Heatseeker](/blog/how-dtc-brands-build-living-personas-faraday-heatseeker) work with us to turn their customers' flat transactional records into rich consumer profiles overnight to power smarter personas, better targeting, and a shared language across the org.

**Your roadmap moves at API speed.** What would take a multi-year build internally — sourcing data, hosting it, modeling against it, maintaining it — becomes a few API calls. There's no infrastructure overhead, no six-figure procurement cycle, and no contract to renegotiate as you grow. Brands like [Hazel](/stories/hazel) and [SalesRabbit](/blog/salesrabbit) collapsed timelines from months or years down to days — and scaled on the same infrastructure they prototyped with.

## The bottom line

Your agent isn't underperforming because it's poorly built. It's underperforming because it doesn't know who it's talking to.

Faraday fixes that with a single API call — enriching every incoming identity with 1,500+ consumer attributes before your agent says a word. No data sourcing, no infrastructure build, no long procurement cycle. Just context, from day one.

If you’re ready to accelerate your roadmap and get the context your agent needs today, [start here](http://buy.faraday.ai) — or [talk to a context consultant](https://faraday.ai/talk-to-sales) about what enrichment looks like for your use case.

Faraday's API gives AI agents the customer context they need to personalize from day one — no data sourcing, no infrastructure build, no long procurement cycle.

All posts