Troubleshooting
When something breaks after go-live, work from the symptom you’re seeing to its likely cause, then open the one guide page that covers the fix — this page doesn’t repeat content that already lives there.
Diagnose in order
- Open Personalization logs and find the decision, request, or event in question.
- Match what you see to a row in the table below.
- Open the linked page for that cause and follow its steps.
- If the symptom still doesn’t match anything below, check limits in case you’re hitting one.
Symptom, likely cause, where to check
| Symptom | Likely cause | Where to check |
|---|---|---|
| No decision returned | The personalization point has no decision configured, no trained recommender with a fallback, or the call was made without Personalization enabled for that context | Create personalization points and decisions, Call the Decisioning API |
| Decision returned, but recommendations look wrong | The recommender is under-trained, a fallback is serving instead of the trained model, or a targeting rule is excluding the audience you expected | Create a recommender, Create personalization points and decisions |
| Decision returned, but the wrong profile or content is personalized | The data space is using the wrong default data graph, or the profile/item data graph the recommender reads from is stale or misconfigured | Review the default data graph, Build the profile and item data graphs |
| No events arriving / profile not updating | The connector's data stream is inactive, not mapped, or the foundational data kit isn't deployed | Create and map the website data stream, Install a foundational data kit |
| Same person treated as separate individuals across channels | The identity resolution ruleset isn't active, or the identifiers each channel sends don't match the rules it expects | Build the identity resolution ruleset |
| Calculated insight value missing or stale | The calculated insight isn't attached to the data graph, or it hasn't recalculated since the underlying data changed | Add calculated insights to the data graph |
Pages named in the “Where to check” column above:
- Create personalization points and decisions
- Call the Decisioning API
- Create a recommender
- Review the default data graph
- Build the profile and item data graphs
- Create and map the website data stream
- Install a foundational data kit
- Build the identity resolution ruleset
- Add calculated insights to the data graph
Discussion
Comments are not configured for local development. They appear once the site owner connects GitHub Discussions (Giscus).