Hi everyone,
Weβre investigating low reCAPTCHA scores affecting login/signup in our Android and iOS apps. Our backend requires a score of 0.7 or higher.
In a sample of 472 assessments:
- 105 scored below 0.7.
- All tokens were valid and actions matched.
- Many low scores had no recorded reason; some reported UNEXPECTED_ENVIRONMENT.
We also found that our backend sends the literal string "JA3" in event.ja3 instead of a genuine TLS fingerprint.
This placeholder appears in both low-scoring and high-scoring assessments. We have identified an integration issue, but have NOT confirmed that it causes the low scores.
Since ja3 is optional, we propose omitting it when a genuine fingerprint is unavailable, then comparing results using fresh tokens under consistent test conditions.
Could anyone advise:
1. How does reCAPTCHA handle an invalid JA3 value?
2. Is omitting the field the correct approach?
3. What additional diagnostics can explain low mobile scores without reason codes?
We can provide assessment IDs privately through an appropriate Google support channel.
