Setting up a Zapier 'Zap' to scan incoming feedback, call transcripts or reviews via Canny Autopilot is a straightforward two step process.
​
Firstly you'll need a trigger - Google Form's "New Form Response" in the above example. Configure your trigger as desired, for example selecting a particular form to source responses from.

Finally you need to feed the data your trigger into Canny Autopilot - via the Scan with Autopilot Action. Refer to the image below for an example of how to setup each field. When unsure of the Feedback Type to select, conversation should be the default. It's important that Feedback ID is a unique ID - re-using the same ID will result in previous feedback scans being overwritten with the new feedback.
​
Additionally, another tip is to ensure that you're using unique Author ID's and Author Emails if feasible. Guaranteeing uniqueness means that multiple votes can be merged together and provide you with an accurate count. If all feedback appears to come from the same user then duplicate votes will only count as the original vote and the score will remain as 1 vote.

Ensure you're using unique Author IDs and Author Emails.

No problem! These fields are used to associate any extracted feedback with a certain user within Canny. If your survey is anonymous, you can simply provide a fixed value for Author Name, e.g., Google Forms Respondent. Email is optional, and we'd advise you not to set it if you don't have a source for that as you will likely run into the wrong user id supplied error. See the next point below.

The Author ID is tied to their email. As such, if you provide Author ID 1234 with email example@test.com and then try to call 'Scan With Autopilot' later with a new ID 5678 and the same email, it will throw this error. Instead, you should use a different email address OR the same ID (1234) as before.

If you can't guarantee that your author will be the same as your user's email address, you have a couple of options:

This issue should only arise when a user is allowed to submit feedback with the same email address multiple times. Consequently, allowing only one response per user is a good way to prevent this. This can depend on the trigger (source). In the case of Google Forms, for example, you’d have to say, “Allow only one response per user” when setting up the form.

If you don't have a unique ID in the payload from your trigger, try to use a timestamp or something similar to generate a unique ID.

Currently, conversation, ticket, and review feedback types are processed using the same AI pipeline. If you're uncertain, pick the closest of these three, and you'll see the same outcome. Call is meant to indicate a call transcript, e.g. from a customer success call or a sales call. You will see the best performance when speakers are labeled in call/conversation transcripts, as the AI will be able to distinguish better between employees and customers.

If you’re using Zoom or Gong, we strongly recommend using one of those native integrations directly to Canny rather than going through Zapier.

Any Canny customer can use the 'Scan With Autopilot' Zapier action regardless of the billing plan. Free plan users can also use this tool. Of course, all 'Scan With Autopilot' actions are billed using the standard Autopilot credits. Learn more about credits here

Usages are billed the same as all other Autopilot sources.

For instance, 'conversation', 'review' and 'ticket' Feedback Types are all billed at 1 credit per usage. Similarly, if you select the Feedback Type as 'call' you'll be billed 5 credits like our Gong and Zoom sources as it uses our proprietary method for extracting feature requests from call transcripts.

These fields are used to render a link to the source in-app. Not setting them will not break or impact performance - it will just make it less obvious where feedback has come from and prevent you from clicking a link to view the source.