Detect finds. Act fixes. Excel compounds.
By the third stage, the value no longer comes from any single capability — it comes from how they work together. The discovery-intent data you have built up ranks your results by what shoppers actually buy, neural retrieval steps in precisely where that data shows keyword search falling short, and every improvement is proven on real traffic before it ships. Each outcome then flows back into the data, so the whole system keeps getting sharper. All of it in one tool your team manages. It runs continuously, and any decision it makes can be overridden.
Excel is one loop, not another list of features.
Everything in Excel runs on everything else. Your data decides where intelligence is worth applying, testing proves what each change actually earned, and those outcomes make the data richer for the next decision. It is this connection between the capabilities — not any one of them on its own — that makes the results compound.
discoveryData condenses every query, click, impression, basket and purchase into demand, trend and ranking signals.
Those signals rank results by purchase intent, and neural ranking steps in exactly where they show keyword retrieval failing.
Every change is A/B-tested against the same intent on your own traffic, and only the winner ships.
so each pass through the loop starts from a smarter place than the last. That is where the compounding comes from.
Rank by what shoppers buy, not what your catalog says.
NeuralInfusion is a Custom Intent API: one interface to everything you want to customize — neural intent prediction, ranking, and raw-data analysis via Learning-to-Rank — all drawn from your own journey data. Clean intent from Act reaches your stack before it ranks; the API re-weights the result set after.
Neural intelligence where it provides value.
Semantic and neural retrieval, added to the discovery stack you already run, applied where it lifts conversion. It maps each query into an intent space, then runs hybrid vector retrieval, served through the searchHub thin client. Your own data decides where: neural targets the zero-result and underperforming searches your journey data exposes, so it stays orders of magnitude cheaper. It keeps all your business rules: filtering, curation, searchandising and ranking — at almost no added latency. And not only cheaper and faster: more importantly, it is more precise and not a black box.
Dense, enriched discovery-intent data. Use it anywhere.
searchHub condenses the complete discovery journey — every query, suggestion, redirect, click, basket and purchase, from shoppers and AI agents — into dense, highly enriched discovery-intent data: behavior signals such as demand, trends and rankings, plus feeds and an API. It is yours, to use wherever it drives your business — in our tool, in your BI, in your content pipeline. First-party data the cookie-deprecation era makes scarce.
Agents are a new inbound channel. You decide what you share with it.
AI assistants and AI search increasingly bring shoppers to your products — an inbound marketing channel, with a channel's economics: cost versus revenue. You cannot control what an agent selects, and no one can. What you do control is what you give it access to. The same discovery-intent data that ranks and tests your results shows which products earn on this channel and which do not — so you share your catalog deliberately, not by default.
AI-referred retail traffic grew 393% year over year and converts 42% better than other channels. (Adobe, 2026)
- — Not a shopping agent. We do not buy for the shopper.
- — Not a replatform. Your engine and catalogue stay where they are.
- — Not control over what agents pick — nobody has that. It is the data to decide what you expose to them, and what you hold back.
Retail never stands still, so the testing never stops. Show two answers to the same intent. Keep the one that wins.
Most product discovery technology can only test what happens after the search — the ranking, the layout. searchHub works one step earlier, on the query itself. That opens a test no engine can run: same shopper intent, two different queries. Does 'running shoes' sell better than 'shoes for running'? Show both, and you see what changed and why. The winner then writes back into the data the whole loop runs on, so the next decision starts from what you already learned.
Change what the same search returns by electing a different master query.
Send that search to a curated, merchandised page instead of a result list.
Blend neural-matched products into the set where they lift conversion. Beta.
You do not guess which answer wins. You test it on your own traffic, and keep the one with more positive user interaction.
"Even a good onsite search benefits from searchHub. Our A/B tests showed an increase in both conversion rate and average order value."
Neural where it counts, runs on Discovery Data, and feeds back everything it ranks, tests and learns. That loop is the shared intelligence layer: one understanding of intent, available to every system.