How I Lost 50 Leads With Two Simple Mistakes

After a round of sharing engram with friends and family, I decided I wanted to have offline functionality be front and center in the app. I coded up the necessary changes and launched a "Use without account" option.

Excited about the possibilities, I decided to make this the preferred default way to use engram. I updated the sign up page to highlight this and proceeded to run my first ad on Reddit.

A day and $50 later I see a bump in traffic to the product website. After getting over the excitement of new people on the website, I realized that I didn't learn anything from that experience. Some may have tried the app, but if they did it was completely offline and I have no way of contacting those people.

At this point, I decided to switch back to focusing on getting people to sign up so I could follow up with them by email. I quickly re-wired the sign up page and got ready to start sharing content on Reddit, Medium, and Twitter.

Weeks passed and I could see different marketing attempts were bringing people to the website, but nothing led to people signing up. It seemed plausible that people might just look at the product page and churn, so I didn't think too much about it.

I then launched a Zapier Integration and shared via the usual channels. The whole day passed and despite the fact that I could see website traffic, again no one signed up.

The next morning I woke up to a Reddit notification. Someone commented on my post saying that the sign up page was returning an error when attempting to create an account. I quickly verified that this was an issue, and then checked to see how long this might have been happening for.

Turns out, when I switched the sign up page back to focus on online enabled accounts, I made a minor coding error that caused the sign up to fail 100% of the time.

Embarassed and frustrated, I went to my server to check the logs for how many sign ups had failed. Turns out about 12 people had tried to sign up and hit the error. These were only HTTP status error logs, so no information to recover.

Now that I've seen how critical this piece of the process is, I have significantly bumped up the priority of getting monitoring and automated testing in place. Especially at this scale, any kind of an error thrown should be addressed as soon as possible.