I have setup the webhook in n8n to receive data from elevenlabs, but this is working only for the test url, not production. I can confirm that I have set a POST request in elevenlabs webhook configuration and given the correct production url. Additionally, I can see that the executions tab brings in the data correctly using the production url, but this is not the case in the editor, thus I cannot pass this data to the next node in my workflow.
Your workflow is working correctly. The production webhook (/webhook/...) is designed to NOT show data in the editor - that’s expected behavior.
Editor view: Only works with test URL (/webhook-test/...) while “Test workflow” is active
Production URL: Data flows through but only visible in Executions tab
Since your Executions tab shows the data correctly, your workflow is functioning properly. The data is being passed to the next nodes - you just can’t see it in the editor view for production executions. Check your Executions tab to see the full flow through all nodes.
I forgot to add a link to the docs which should also confirm this. @Abhishek_Luthra
@Uros_Zdravkovic thank you - I will run a few tests to check and come back to you - thank you for the information.
Could you also help me with another issue I am facing between the urls. If I set a test url in eleven labs and manually execute the webhook in n8n to “listen for test event”, it seems to time out after a specific duration. The problem is that I haven’t finished my conversation with the ai agent in eleven labs, and since the n8n test webhook has timed out, eleven labs does not send anything back as n8n’s test webhook is not “listening” anymore. Is there a way around this? I guess this applies to any test webhook, irrespective of whether data is coming in from eleven labs for elsewhere?
That is one of the commons issues that are mentioned here, it is listening for about 120 seconds and then stops after if the execution was not made, I believe this is security measure.
You can follow the workaround provided in order to resolve the issue. @Abhishek_Luthra
Thanks @Uros_Zdravkovic . I have gone through the documentation on timeouts, but I don’t think I have managed to resolve this. Each time my conversation with elevenlabs ai agent comes to an agent, the ai agent says that they have run into a technical problem to process the information. Upon some research, I believe this happens because the n8n webhook has timedout, and elevenlabs cannot send the information back to n8n. I have tried both test and production urls, and have run out of steam a bit now as it seems I have tried everything I could over the past 2-3 days. Not sure what else I could try.
Thanks. Upon further research, I read that having two workflows would help resolve the issue, so I did that, but still that did not help. I used the production url, where both workflows were published.
I had connected the two workflows as well, as suggested. I have included both here. Thanks for your help.
Hey @Abhishek_Luthra I was not able to configure it as well, because from my investigation ElevenLabs will send the conversation only when it is over, so the issue here is only with using the testing option.
In production this shouldn’t cause any issues.
I assume you can also add a call in ElevenLabs when the conversation starts, and in that case my logic is that you trigger the workflow extract the conversation_id or whatever it is called and then do pooling until the conversation is over, once the conversation is over you will have the full info.
Hi @Parintele_Damaskin I tried the post-call webhook by copying the production url from n8n into the security tab, and checked that the audio and transcript parameters were added under the advanced tab by default, but I couldn’t get successful executions in n8n, and hence nothing came through in the editor either. I am not sure why this would have happened.
@Uros_Zdravkovic thank you very much for the updated JSON workflow. I made a couple of minor tweaks and it seems like it is now working.
Thank you both very much for your suggestions. Over the next few days, I will keep refining the workflow and let you know if I get stuck again! Thanks again.
Basically 11labs is sending “signals” over http/websocket with everything happens during/after the calls.
But we like to fetch them “n” times to find what you need.
P.S is like they offer you “door to door” the informations you need, but you prefer go yourself and knock at their door to ask “n” times if the call is done or anything you wanna know about that.
That’s the “definition” of a WEBHOOK (get delivered all the informations, just filter them) , but yeah everyone can build as they wish!
