As an increasing number of brands look to data science to guide crucial business decisions, predictive capabilities have become popular across marketing and sales platforms. From conversational AI chatbots to website personalization, data-driven predictions let brands improve customer experiences throughout every customer’s journey.
But with so many platforms boasting new predictive capabilities, how do you know which ones your brand should really be investing in? At Faraday, our framework revolves around the business outcomes you want to optimize (e.g. grow your customer base, converting more high-LTV leads, driving average customer LTV).
Next, we consider your channel strategy and historical customer data to determine the types of consumer predictions that are right for your team. The three types of consumer predictions we find the most helpful for brands are:
- Propensity: likelihood to perform a certain action
- Personas: best-fit archetype
- Value: AOV, LTV, etc.
Predicting a consumer's propensity to take action
Machine learning models score consumers on their likelihood of performing a certain behavior (e.g. converting to a customer, buying a certain product, referring a friend) in order to generate propensity predictions. Popular use cases for these predictions are:
- Customer acquisition: who out there is most likely to become a customer?
- Lead scoring: which leads are likely/unlikely to become a customer, high-LTV customer, etc?
- Repeat purchases: which customers are likely to make another purchase, make three or more purchase, etc?
- Churn prevention: which customers are likely to stop buying, cancel a subscription, etc?
Too often, consumers are targeted with ad campaigns that don’t align with how they actually will engage with a brand, wasting precious time and marketing dollars. Predicting the likelihood that an individual will exhibit a certain behavior allows you proactively optimize your campaigns. Rather than going after people who aren’t likely to engage with you a certain way (and finding that out the hard way later down the road), with propensity scores you can target individuals who score high for a desired business outcome.
Consumer trends in 2020 took many brands by surprise, and it wasn’t unusual for marketing teams to feel like they were left in the lurch, spending money targeting people who weren’t likely to engage — or having to cut back on overall ad spend for the same reason. The ability to predict propensity will be a game changer this year.
When you look at predictive platforms to leverage with your customer data, be mindful of which one you choose. Some predictive platforms only work with first-party data, limiting the depth of your predictions to historical data you already have access to. The key to making the most accurate predictions is enriching your data with third-party data — provided by your chosen predictive platform, or licensed separately — so that you understand more about your customers than the limited data they’ve shared with you. After all, we are all more than just our digital footprint!
Predicting which personas consumers fit into
Personas are archetypes of customers that brands can use to guide personalization for customer groups across marketing and sales initiatives.
With the right personalization systems in place, you can assign personas to contacts across all of your marketing channels (digital ads, email, direct mailers, website, etc). Having a centralized persona engine helps ensure consistent personalization at every touchpoint.
It’s important to note that machine learning makes scalability possible. Check out our webinar, Scaling personalization with machine learning, to learn more.
Predicting a consumer's value over time
We find that a significant portion of brands that begin to use data science focus heavily on customer acquisition, with little thought on what those newly acquired customers will contribute to their business in the long run. Failing to recognize — or simply ignoring — the need for customer lifetime value predictions is a disservice to the growth of your brand.
Engagement and retention marketers benefit the greatest from CLV predictions, as it helps them prioritize who to spend their marketing budget on. It’s more important to attract high-value customers who will spend more over their lifetime with the brand, rather than a few one-time purchasers. Predicted longevity of a customer’s lifetime, too, can aid in loyalty and growth efforts, with the opportunity to increase their lifetime value.
Perhaps an overused, but no less relevant, statistic is that it costs anywhere from five to 25 times as much to acquire a new customer as it does to retain an existing one. Focusing on acquisition with customer value in mind is one strategy, but just focusing on acquisition with no regard for how long those customers will stick around doesn’t make for optimized brand loyalty or increased revenue.
Again, it is important to look at your data sources for customer lifetime value; adding in third-party data enriches your view of each customer and allows you to make much more accurate predictions, as you can understand them as whole people, not just a transactional entity.
The era of consumer prediction is already upon us
The world's biggest brands have already integrated AI into their business strategy. Companies without a sense of their customers' propensities, personas, lifetime value, and customer journeys are playing catch-up to those who do. Let these categories be your guide — or a helpful review! — as you consider how consumer prediction can help your business work smarter this year.
Interested in learning more about the types of predictions Faraday provides? Get in touch!