Check site speed
Measure AXO on the pages and devices your visitors use. The tag, configuration requests, images, and rendered variants can all affect page speed; a small script alone does not guarantee an unchanged result.
Understand what loads
The recommended Install snippet includes an anti-flicker step and the tag loader. Anti-flicker temporarily hides content while personalization is prepared, then reveals it. Depending on configuration, it can hide individual zones or the page body.
The Install page describes the selected strategy. A body-hiding configuration can delay visible content until the swap completes or the safety timeout releases it. Review this alongside download size, main-thread work, and the assets used by your variants.
Run a useful comparison
- Use a controlled test environment with representative pages and content. Keep test traffic and external deliveries isolated from production reporting.
- Record a baseline with the same page, viewport, device emulation, network settings, and cache state.
- Repeat with the current AXO snippet and a representative active variant. Include the initial visit and a repeat visit.
- Run several samples. Compare the distribution of results, including content visibility, layout shifts, main-thread work, and failed requests.
- Inspect the network and performance traces to identify the cause of a regression. Record the AXO configuration and variant used with the results.
If the page gets slower
Check large variant images, duplicate tag installs, slow configuration requests, and broad anti-flicker hiding. Reserve space for inserted content and use assets sized for their displayed dimensions. Re-run the same comparison after each material fix.
Lab results describe the test conditions. Field measurements describe actual visits over a reporting period. Keep those labels attached when sharing a result; neither proves zero impact on every page.
Next: review the installation or check zone boundaries.