We’re seeing issues with the featheryFormComplete webhook sending to us very inconsistently
in this scenario, i’m on my mobile safari, after i just cleared all my website data from the cache.
I fully fill out the survey, you can see the results come into feathery
but no form complete event happens, and no form complete event gets sent to rudderstack.
This is causing us gaps in data on our end
(Would be happy to hop on a call to live debug if needed)
Hi Rachael, are you able to determine any patterns on when you don’t receive the webhook? E.g. just on mobile safari, only after cache is cleared, etc.
And are you just seeing the issue with the rudderstack event emission? Or are you talking about the actual Feathery webhook integration as well
And if it is just rudderstack, could it be related to the issue trevin flagged here? Intercom proactive messaging interfering with feathery rudderstack
I’m seeing it just on safari, and just after the cache gets cleared (or, its that devices first time to the site)
I’ve blocked intercom altogether for the time being and the issue still persists
Yeah, as rachael has mentioned, we nuked intercom on these specific pages where Feathery is embedded.
and to confirm, is this just the rudderstack completion event that isn’t sending? or are you talking about all completion triggers, such as the webhook integration as well
It was the completion event through rudderstack. We have eliminated rudderstack now and it’s firing quickly and consistently. So there seems to be something awry with when it’s fired through rudderstack. We aren’t doing anything special for that other than enabling that integration (but it’s now disabled on our form to be clear)
I’m guessing it’s because we were attempting to include all of the field data on the rudderstack completion event. Do you know if rudderstack events have any type of size or payload complexity constraints?
but that wouldnt make sense as to why this was only happening on a weird edge case? we’re successfully able to send those size payloads 99% of the time
it could be that specific field payloads with specific data types might break. any sense of pattern on those edge cases would help us narrow down the issue here