Product recommendations: personalize your campaigns customer by customer

Product recommendations make it possible: every customer sees a different product selection in your campaigns, generated automatically, with no extra versions to build. Yet your customers’ tastes and buying behaviors have nothing to do with one another. If that gap frustrates you, this use case is for you.

It’s one of the most common use cases for predictive AI in CRM, part of a series breaking down, one by one, the concrete uses of Batch AI Predict. To understand why a predictive score only creates value once activated, start with our framing article.

Why does sending the same product selection to everyone cap your click rate?

Newsletters, push, emails: your campaigns showcase the same product selection for your whole base. It’s the simplest to produce, but it’s also what caps perceived relevance. A customer passionate about one category gets the same window as a customer with radically opposite tastes. At best, one of the two is served; most often, neither feels truly recognized.

The consequence hits the click rate directly. When content doesn’t reflect the recipient’s interests, they don’t click. And on high-volume campaigns, that lack of relevance is costly: in engagement, in conversion, and eventually in deliverability.

Let’s ask the question that matters: how do you know the problem is solved? The primary metric is the click rate on the personalized section (the most immediate effect of relevant content, and the one that best isolates the contribution of personalization). Watch closely alongside it: the conversion rate and the product coverage rate, meaning the ability to surface the long tail rather than just best-sellers. We’ll come back to it at the end.

What does this look like in practice?

Concretely, picture a brand that regularly runs campaigns showcasing products. The same selection goes to the whole base, because manually personalizing content for each segment would be unmanageable at scale.

The problem isn’t the channel or the format, it’s content uniformity. Everyone sees the same products, so the campaign truly speaks only to the fraction of customers whose tastes match the chosen selection. For everyone else, the email is generic. And a generic email gets ignored.

How does Batch AI Predict personalize product blocks for each campaign?

With Batch AI Predict, content personalization is handled right inside the platform, with no manual assembly. It comes down to three steps:

• First, you choose the product catalog level to recommend on: category, family, department, brand.

• Then, the model generates a personalized list of products or brands for each customer, ranked by propensity.

• Finally, that recommendation dynamically feeds the campaign’s conditional blocks or sections, as a display variable.

Concretely, you build a single campaign, but each customer sees a selection of their own. The product block displays their personal top, not a generic window. The CRM team doesn’t hand-build multiple versions: it sets a dynamic variable, and the recommendation engine does the rest.

How does the product recommendation score calculate each customer’s top N?

No black box. The score powering this use case is product recommendation. Its principle: for each customer, it predicts the top N products most likely to interest them.

This recommendation rests on similarity models between customers and between products, computed from all the data available:

• behavioral data (purchases, browsing, engagement)

• product catalog

• socio-demographic data

• and more

The model learns who resembles whom, which product resembles which, and infers for each customer the products with the highest probability of interest, including products they’ve never looked at but that similar customers adopted.

Does this use case combine with others?

This use case personalizes a campaign’s content. It pairs very well with targeting and other forms of personalization:

• First identify the right audience, then personalize what each person sees.

• Personalize on discount level.

• Choose the right repurchase moment.

What results can you expect from product personalization?

Here are the performance levels you can expect on this type of use case, measured by comparing a personalized-content campaign to a generic-template one.

• The click rate rises sharply on the personalized section: +47% to +64%, direct proof that the content is relevant again for each person.

• Conversion follows in the same direction, at +17% to +20%: relevance turns into purchase.

• The third effect is more structural: product coverage widens to up to 80% of the catalog, versus a handful of best-sellers in a generic window. You're no longer pushing the same hero products to everyone. You're activating the full depth of the catalog.

What does personalized product content actually change for your campaigns?

Broadcasting the same product selection to the whole base caps relevance and click rate. Batch AI Predict’s product recommendation score feeds dynamic blocks so each customer sees a selection of their own, right inside the platform. More clicks, more conversion, and a better-exploited catalog.

To activate it on your campaigns, talk to a Batch expert.