Skip to content

Replay Has a Different Transaction Count Than Recording

The recording captured 50 transactions, but the replay shows 45 or 55. This is almost always harmless. The server may now load slightly different assets (new tracking pixels, removed ads, an extra fetch from a CDN), JavaScript on the page may have changed which AJAX endpoints it calls, or third-party content (analytics, ad networks, social widgets) may be unreachable from the replay environment.

What matters is whether the critical transactions succeed, not whether the total count matches.

When to Care

Transaction count differences matter only if:

  • ❌ A critical transaction is missing (e.g., the payment POST disappeared)
  • ❌ The error count is high (many transactions failing, not just count-different)

If login succeeds, search returns results, checkout completes, and so on, ignore the count difference.

How to Investigate

  1. Compare the recording and replay side-by-side in the Test Case Editor
  2. Identify which transactions differ: added or removed
  3. Check whether they're non-critical: analytics, tracking pixels, ads, third-party CDNs

For load testing specifically, irrelevant third-party transactions should not be in the test case at all. They contaminate the load profile, add noise to results, and (in some cases) generate test traffic to services that will rate-limit you or treat your test as an attack. The Configure for Replay wizard restricts the test case to your own hostnames in its first page, which is the right time to drop these.

For a hostname cleanup on a test case that's already been configured, see Modifying Test Case Content.