Onsite search relevance + merchandising for ecommerce. Tunes ranking off real conversion data, surfaces query-level analytics, and gives the merchandising team an override panel that doesn't require a developer.
Every ecom site has search. Most have search that hides results. "T-shirt" returns 12 items when the catalogue has 400; "blue dress" returns dresses that aren't blue. The default platform search is tuned for the wrong thing — it doesn't see your conversion data, so it can't learn from it.
Built into the same FeedPulse Console that runs every other Pulse-suite product — single auth, single data model, single place to look when something goes wrong.
Click-through and add-to-cart signals weight the ranking continuously. Products that convert from a query rise; products that don't fall.
See the queries that return zero results, the queries that return too many, and the queries that drop off without a click. Fix them in one click.
Pin, boost, hide and rule-based reorder per query. The merchandising team owns this without needing a deploy.
Plurals, common typos, brand misspellings handled without a manual list. The model learns from successful conversions.
Filter UI generated from real product attributes, not a hand-curated list. Adding a new product attribute auto-surfaces in the filter.
Empty-search-results pages, no-product-found pages and "people also viewed" all share the same ranking model.
One call. We'll dig into what you've got, where it's friction, and whether we're the right people to help. No deck, no pitch.