Setup
1
Open Analytics
Open Analytics in Console.
2
Select a time range
Select a time range, for example last 7 days.
3
Set granularity
Set granularity appropriate to traffic volume.
Common workflows
Evaluate baseline performance
Track at least:- Recommendations Served
- Click events
- Purchase or conversion events
Investigate anomalies
Use context and advanced filters (request_id, session_id, user_id) to debug sudden drops or spikes.
Build an optimization loop
1
Change one variable
Change one variable at a time, such as a context filter, model family, pipeline setting, or ranking rule.
2
Observe engagement deltas
Observe engagement deltas over a fixed window.
3
Keep the winners
Keep changes that improve KPI targets; revert those that regress.
Example app journey: Improve homepage CTR
1
Capture a baseline
Baseline homepage context metrics are captured for 14 days.
2
Adjust context filters
Team adjusts homepage context filters to remove low-engagement categories.
3
Roll out to a sample
New settings roll out to 25% of traffic.
4
Compare against baseline
CTR and downstream purchase events are compared versus baseline.
5
Expand the winning configuration
Winning configuration is rolled out to 100% traffic.