Changelog

We’ve introduced the Faraday Predictor Leaderboard—a live, filterable view into the most predictive traits across our platform. Updated weekly, it showcases which traits are driving outcomes across industries and use cases, offering insights into the real-world signals powering our models.

You can also sort by industry and use case to zero in on the traits that are most relevant to your team. Take a look for yourself.

Modeling strategies used in each Outcome are now available via the API via (GET /outcomes/:id/analysis) and in the dashboard. Each strategy includes details like name, description, and data providers—making it easier to audit models, trace predictions, and build smarter automations.

Note: Strategy metadata will only appear for newly generated Outcomes. Older Outcomes may not include this information unless reprocessed.

To learn more, review our full blog post on the subject, or visit the Outcome API documentation!

We weren't seeing enough usage of the Recipes builder to justify keeping it around. So, we have removed it from the Faraday dashboard so we can focus on building the features that matter most to you. You can still use all other interfaces or API to build the same resources you could with Recipes.

Now you can confidently run performance analyses on dynamic predictions—even when individuals have already attained the outcome.

When dynamic prediction is enabled, Faraday updates scores in real time based on the latest available data. Previously, if someone had already attained the predicted outcome at the time of processing, their score fields were left empty. This made it difficult to analyze how well the model performed after the fact.

With this update, those individuals now retain their original probability and percentile scores—captured as they were at the moment of attainment. That means cleaner, more complete reporting when comparing predicted vs. actual outcomes over time.

No changes are required to your pipeline setup or outputs; these updated values will be included automatically where applicable.

You can now get a quick snapshot of who’s in your deployment—no spreadsheet required.

Every Target now includes a row count by default. For deeper insights, you can opt into analysis by specifying traits (like age or income) and geographies (like ZIP code or state). Faraday will return a structured breakdown via API, plus a downloadable PDF if requested, which are perfect for internal reporting—or for sharing results with clients and partners.

To enable it, just include analysis_config when creating or updating a Target. More information can be retrieved from our "Retrieve a target's analysis" docs.

Faraday's Lookup API now supports payload fallbacks, a new strategy for resolving partial identity matches. When a full match isn't possible, we aggregate traits at the household, address, or ZIP code level to return meaningful results—along with an update to our match_type field so it now includes a new postcode_only value. This means more complete data, fewer gaps, and better predictions from every lookup.

Want to learn more? Here's a blog from the very engineer who merged the new feature!

When creating a CSV dataset, you can now upload multiple files simultaneously. Previously, you had to upload your first file and then navigate to the Data tab within your dataset to append new files.

Each CSV must have the same headers and data types, but now you can upload them all at once. This feature is particularly beneficial for large datasets that are divided across multiple files.

Additionally, you can upload multiple files at once in the Data tab of your dataset. Other overall user experience issues have also been addressed in this UI.

You can now easily download Faraday Personas as CSV files directly from the dashboard (a feature previously available only to API users).

This update adds a "Download CSV" button on the Persona analysis page and in the Persona set menus, making it easier to access and export Persona data for reporting and analysis. The CSV format mirrors the layout of the Persona set analysis page, providing a more accessible and flexible way for both brands and agencies to integrate Persona data into internal reports, BI tools, and dashboards.

Want to learn more? Check out this blog about how one of our partners, EmpowerFi™, has already started to generate value with the feature!

We've improved persona chart comparisons on the analysis page by ensuring consistency between persona charts and their corresponding column charts.

We've improved data visibility and navigation by adding summary cards for data points to detail pages for cohorts and datasets. These changes mean you no longer have to return to list views to find important counts. The dataset data tab also now includes a sample data table, and connection information links directly to the used connection for easier access. Finally, we've refined summary item styles, making the interface cleaner and easier to read.