Enable AI Personalization in your Algolia plan and configure a Personalization Strategy in the Algolia dashboard: assign facet weights (how strongly signals like category clicks or brand conversions influence the user profile).
Instrument your storefront to send click, conversion, and view events to the Insights API (POST /1/events), including a stable userToken for every event; the same userToken must be sent with all search queries for that user.
In each search request, set enablePersonalization: true and pass the userToken; Algolia will boost results that match the user's inferred preferences.
User profiles take up to 3 hours to generate initially and refresh every 2 hours; personalization only affects returning users whose profiles have been built from prior events.
Debug the profile for a specific user token by calling GET /1/profiles/personalization/{userToken} to inspect the computed affinity scores and verify that events are being attributed correctly.
Known gotchas
Personalization does not affect brand-new users (first session); it activates on subsequent visits once the user's profile has been computed from their past events — set expectations accordingly in product requirements.
The Personalization API rate limit is 40 API calls per second per application; if your application generates high event volume, use the Insights API's batch events endpoint to reduce call count.
Sending inconsistent userTokens (e.g., a new random ID per session) fragments user history and prevents profile build-up; tie userToken to a durable first-party identifier.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp