Three data sources, recalculated daily

To move away from a purely ‘historical’ approach, FeedMax Predict® cross-references three types of data on a daily basis:

• Google data: search signals, past performance, trends by product and category.
• First-party CRM data: customer value, recency, frequency, average basket size, margin, affinities by universe.
• 397 external prediction factors: weather, seasonality, holidays, and other contextual signals that influence demand.

Every night or every day, the engine recalculates the probability of a sale in the very short term for each SKU, taking into account the latest signals.

Live predictive scoring: Top ↔ Low

FeedMax Predict® does not pigeonhole your products. On the contrary, they automatically move from Top to Low (and vice versa) depending on their current potential:

• A product can become a Top seller when a heatwave, public holiday or sales period approaches.
• A best seller can drop to Low if demand slows down, if the context changes or if other products take over.

Result: your Top/Low lists are always up to date, ready to be used in your campaigns, promotions or CRM scenarios.

An intelligent co-pilot, not a budget autopilot

FeedMax Predict® does not push for more budget here or less budget there. It gives you an up-to-date map of potential by product, which your teams can then use to steer:

• Creation or adjustment of campaigns dedicated to top products on Google Shopping, PMax, Meta, etc.
• Exclusions, priority reductions or background work on low-performing products.
• Alignment of acquisition/CRM/category with the same vision of high-stakes products.

An A/B test to demonstrate value

To prove the impact of daily predictive scoring, FeedMax Predict® comes with an A/B testing feature:

• Group A: campaigns built without using Top/Low labels.
• Group B: campaigns structured around Top FeedMax Predict® products, updated daily.

In practice, FeedMax Predict® becomes your daily compass: every morning, your teams know which products deserve to be promoted and which ones need to be withdrawn or reworked.