Supercharge Conversions With Predictive Cross-Sell and Upsell Strategies
This post is our second on cross-sell and upsell strategies; if you missed the first, check it out here! Both posts are part of our Predictive Frameworks series, which explores the most effective use cases for predictive analytics.
Strong, sustained customer relationships are invaluable for every business. Usually, that ongoing connection means that not only do customers return to your business often, but they also make repeat and additional purchases of the products or services you offer.
Fortunately, predictive analytics now helps you know in advance whether a customer is likely to be interested in additional products or an upscaled offering of your products or services. You can use your customer data to look for meaningful patterns in what customers have bought when. Then, those patterns can provide predictions about other customers so you can provide them with relevant, effective offers that they’re most likely to find appealing.
And yes, your business can benefit from this predictive approach, no matter what industry you’re in — boosting your bottom line and, when done right, also improving customers’ experience and the value you deliver to them.
- We’ve got more about what that process looks like in our first blog post about cross-sell and upsell predictive modeling.
- If you’re not sure if you have enough data or the right kinds of data to use for this purpose, be sure to read about how your data is probably ready to rock and roll. Generally, if you have data for thousands of customers and thousands of transactions, you’re good to go with predictive analytics.
We’ll take a look here at how businesses in different industries can find success with this strategy. We’ll also go behind the scenes for a quick glimpse into how these predictive models use your data and allow you to time-travel with your customers into the future.
What success looks like with predictive cross-sell and upsell strategies
Retail and e-commerce
- Use sales resources efficiently to contact prioritized customers who are most likely to take advantage of a cross-sell or upsell offer
- See which customers might be more likely to purchase a higher-level model of a product with added features, in lieu of an entry-level model, if presented the option
- Offer personalized invitations in your mobile shopping app to entice likely high-value customers to attend events at brick-and-mortar stores
Financial services and insurance
- Find out which customers who use one of your services are most likely to purchase additional services. For example, which car insurance policyholders might be most interested in home insurance?
- Provide premium account or membership opportunities to customers whose profiles match others who upgraded beyond the basic level
- Increase wallet share by offering complementary account types (for example, a home equity line of credit to a mortgage customer) or suggesting refinancing opportunities
CPG and DTC
- Give customers personalized coupons or promotions for products complementary to what they’ve already purchased
- Build bundles of items frequently purchased together by past customers whose profiles and purchasing behavior are similar to your current customers
- Suggest to likely-to-convert customers the most relevant additional and/or upgraded products and services you offer
Apps and gaming
- Offer upgraded subscription opportunities to entry-level users similar to those who purchased higher-level subscriptions
- Send social proof messages to users with high lifetime value to encourage them to purchase upgraded options
- Showcase new features or game levels to consistent users of the app or game
- Identify customers most likely to add supplemental in-person services — for example, adding a deep conditioning treatment to a haircut and style
- Select subscribers to whom your upgraded streaming media subscription would most likely appeal and extend an exclusive offer
- Send targeted emails to customers who haven’t yet tried an additional service you offer, but who match the profile of customers who used that service in the past
How predictive models determine conversion likelihood
With all these potential applications, it might seem magical that a bit of math can do so much to see into customers’ likely future behaviors. The math is actually quite complex because it analyzes lots and lots of customer data with highly accurate algorithms. So how exactly does that work?
We won’t dive into too much technical detail here, but let’s take a quick look at the overall concept. Pecan’s predictive modeling for cross-sell and upsell conversion is based on the broad idea of propensity modeling. Propensity modeling is a general term that can include a variety of specific modeling techniques, all from the realm of classification models. Classification models try to determine the correct “class” or category for an individual row of data.
In propensity modeling, the mathematical model examines each row of data — in this case, representing a single customer — and runs those data points through a complex mathematical process. The outcome of that process is a prediction of a class. In this case, the class is whether the customer fits the category of “likely to convert with the cross-sell/upsell offer” or “not likely to convert.”
Thanks to the specific way Pecan uses classification models, our platform’s predictions are even more nuanced. Instead of a simplistic judgment of “likely” or “unlikely,” we assign each customer a specific propensity score representing how likely the customer is to buy. That score can guide how you prioritize outreach and even which offers you select for specific customers.
When Pecan builds models, the automated feature engineering and selection process figures out from your historical data which customer data points matter, and how much they matter, in determining a customer’s propensity to buy.
Predictions are generated about future customer behavior by looking for similar patterns in the profiles of customers who are yet to convert, and then seeing exactly how similar they are to those who have already converted. The way every customer characteristic contributed to each prediction is also explained for each customer. In other words, you can see exactly which factors contributed most to a specific customer’s score.
With all this powerful knowledge in hand, you can proactively plan your offers and outreach to those customers, confident that your priorities and resources are correctly aligned. Moreover, ongoing monitoring and updating of your model will ensure you’re always acting on the latest information. These dynamic, accurate predictions will positively impact your ROI across all teams where the predictions are put to work.
Getting customers interested in more products and services will never have been easier, because with these predictions in hand, you’ll focus efforts and your dollars on the customers most likely to be interested. It’s a mutually beneficial arrangement that builds your understanding of your customers — and reinforces that stronger relationship between them and your business.