FoundIt! and Google AI Commerce Search
Google AI Commerce Search (formerly Vertex AI Search for commerce, and before that Cloud Retail Search) brings Google’s search models onto your own storefront. It works, and a growing number of retailers are adopting it. It also runs on two things you supply: your product catalog and your user events. FoundIt! improves the first, gives you an intent layer you own and can see, and extends the same intelligence offsite, where AI Commerce Search doesn’t reach.
What you supply
Google’s models are only as good as the catalog you hand them
AI Commerce Search applies Google’s web signals, knowledge graphs, and commerce-tuned models to your onsite search. The semantic understanding is genuinely strong. But it still has to map a customer’s query onto your products, and it can only work with the attributes your catalog actually contains. A product described as “sandals, brown, high heel” gives even a very capable model very little to reason with when someone searches for espadrille wedges for a garden party.
There is a second consideration, and it is about control rather than capability. The intelligence doing the ranking sits inside Google’s stack. Retailers get the outcome, and increasingly some diagnostic tooling, but the understanding of their own customers is not something they hold, inspect, or take with them. For a capability this central to trading, that is worth thinking about before it becomes structural.
And it is an onsite product. Your customers are also discovering products offsite, in Shopping, in AI answers, and through agents acting on their behalf. Solving one half well still leaves the other half open.
Side by side
AI Commerce Search alone vs. AI Commerce Search with FoundIt!
| Capability | AI Commerce Search alone | AI Commerce Search + FoundIt! |
|---|---|---|
Onsite search relevance | Google models applied to the catalog and events you supply | The same models applied to an intent-enriched catalog, so more queries find a match |
Attribute completeness | Works with the attributes already in your catalog | Occasion, use case, material, and mission attributes added across the full range |
Cold start & new products | Learns as user events accumulate | External demand signals mean new products are understood before the first click |
Offsite discovery | Onsite search, browse, and recommendations | The same enrichment flows into Shopping feeds, SEO, and AI answer surfaces |
Beyond the search box | Search, browse, and recommendation surfaces | Visual navigation, curated collections, and PDP discovery deployed across the journey |
Your intent layer | Understanding sits within Google’s platform | An intent layer you own, can inspect, and keep if your stack changes |
Setup | Platform you already run or are evaluating | Layers on top; no rip-and-replace, feeds the engine you chose |
Offsite and onsite
Discovery happens offsite and onsite. Most tools pick one.
Offsite is everything that happens before a customer reaches you: your product feed in Shopping, your pages in organic results, your products cited in AI answers and retrieved by agents. Onsite is everything after: search, navigation, collections, and product pages.
AI Commerce Search is a strong answer to the onsite half. Feed and SEO tools address parts of the offsite half. FoundIt! runs on one intent layer that serves both, so the semantic understanding that helps an onsite search return an oversized jumper when the word “oversized” appears nowhere in the product data is the same understanding that makes that product answer-ready in your Google feed.
That matters commercially because it is one piece of work rather than two, and because a customer who found you offsite and then can’t find anything onsite is still a lost customer. The two halves are the same problem viewed from either side of the click.
Proof in the real world
Both sides of the click
Making the whole range findable, onsite and off
Ironmongery Direct carries more than 20,000 products, including a deep long tail that rarely surfaced. Making every product findable in the language customers actually use delivered an 8% year-on-year revenue uplift in their most recent test, with margin holding firm and customer numbers growing. On the offsite side, John Lewis saw an 8.6% lift in organic PLP traffic from intent-led interlinking, independently validated by SearchPilot. Same intent layer, both halves of discovery.
“We used to be a catalog retailer that happened to have a website. Now we’re a fully digital retailer.”
Keep Google’s models. Own the intent behind them.
AI Commerce Search gives you Google-quality search onsite. FoundIt! gives it a better catalog to work with, an intent layer you own, and the offsite half of discovery in the same piece of work. We’ll show you what that looks like on your own range.