vidIQ - Reducing churn

Understanding the main causes of paid user churn and creating a solution that led to a 15% reduction on paid user churn.

My roles

User Research

Product Strategy

User Research

Understanding what was causing a rise in churn

While working at vidIQ, we’d seen a steady increase in churn over a 6 month period. So we had a high level goal of brining churn back down.

The next step was to get on some calls to get a deeper understanding of what was happening and understand the gap between the expectation when a user paid for vidIQ and what we delivered. So I carried out 10 “jobs to be done, switch interviews” with users who had recently churned within 1 - 3 months of paying for vidIQ.

Uncovering our opportunity

One feature stood out in particular, this was our “Daily ideas” feature, where users would get between 10 & 50 video ideas depending on their plan.

We’d seen in the usage data, that this was a core activation feature and it was usually the first thing new users went to. But we also saw that usage would drop over the first few weeks.

We saw a common pattern in people being really excited about this feature. It promised so much “Ai powered video ideas” harnessing the full power of vidIQ.

A common story we heard was that users expected our Ai to learn about their channel over time. It usually took between 1 - 3 months for a user to realise that the AI wasn’t learning and the results weren’t improving.

Solving the problem

The most common issues people had with our Ai results were that they felt like AI had written them. They weren’t personalised to the users channel, they didn’t match their channel style or tone and often felt corny and click-baity.

Now I had a clear picture of the problem I was able to get to work on a solution.

I started by creating a profile of the users channel using chatGPT. This gave us a foundation to generate more personalised results around. From here I utilised the YouTube recommendation engine to get a better understanding of their audience according to YouTube.

Getting feedback from the right people

Before we spent engineering resources on this new solution, I wanted to make sure the results were worth the bet. So I reached out to the people who expressed the most frustration with the feature on the calls.

I created a new prompt and added their data to chatGPT in order to get personalised results for each user. I then sent an email with a bit more detail on what we were trying to do and asked for their honest feedback.

The response was overwhelmingly positive, with users expressing a desire to use these ideas immediately.

Here’s a quote from one of the people I reached out to


Nailed it, Dave.

The channel description and the details are fantastic. It’s like, you get me.

The ideas generated are so useful. I would be EXCITED to make each of those videos. Especially if those are keywords people are searching for right now. Are they?

It feels highly relevant to me, tailored to my channel, and like someone was actually listening rather than offering a bunch of YouTube bro advice. Lol.

I’m impressed. Did the AI generate this or you?

I’m not clear on that (and that’s a good thing!)

Result: 15% reduction in churn

We knew that the core problem was only around the results, so we decided not to make any design changes in order to speed up development.

By the end of the experiment, we saw a 15% reduction in churn. This helped us understand the real value of the Daily Ideas feature and how we can utilise AI to help users uncover opportunities to create content that will get more views and grow their channel.