There are two sections in Arro where the inputs you include will be sent directly to the AI so it’s important to know how to get the most from these.
The interview context field in the interview setup sets the framing for the entire interview itself.
We recommend that you be as specific as possible here, ideally including the following details:
<aside> ✅ Good example: We are hoping to chat to customers who have just started using the AI DJ feature on Spotify. We are keen to learn about when they use the feature (including specific use cases), how they would rate it overall and what improvements we could make.
</aside>
<aside> ❌ Bad example: We want to talk to our customers about a new feature we just launched
</aside>
The next inputs which directly impact the AI are the Outcomes. Think of Outcomes as what you would include on an interview script if you were talking to a customer face to face. These are the ‘must include’ topics which you want to cover.
It’s important to note though that Outcomes are not questions. Unlike more static structures like surveys, you don’t need to clarify the exact question that you want to ask. The AI actually performs better if you just provide the general guardrails. That allows it to adapt it’s specific questions depending on the topic it is addressing. If you specific the exact wording, this may lead to a lower quality interview as the AI is forced to stick to the pre-determined questions even if the respondent has answered in a way that is unexpected.
It’s also important to note that the AI will raise the interview questions in the order that you list them. You can rearrange the order in our UI as required. The rule of thumb here is to order the topics in a natural way, the same way you would if you were talking to the person in a more casual format.
Taking the above example with the Spotify AI DJ feature, let’s draft the right Outcomes for this interview.
<aside> ✅ Good example:
<aside> ❌ Bad example: