Almost nobody writes in to say a product disappointed them. They sign up, try the first screen, and close the tab. For a small team that silence is most of the feedback there is, and it is the most honest kind: nobody is being polite.
So we stopped waiting for emails and started reading what people did.
What we record, and what we refuse to
On the first screen we keep a short record of the visit: how long someone stayed, which boxes they touched, whether they pressed Generate, and whether something failed. We keep which box they typed in, never what they typed. A record that says "wrote a topic, pressed Generate, nothing happened" is enough to find a bug. The words themselves are none of our business.
After that, the posts tell the story. A post that was opened full size and then left without a download is not indifference. Somebody looked at it properly and decided it was not good enough to use. That is the same verdict a tester gives in words, from a person who will never write to you.
Three things the silence told us this week
A button that said nothing. People wrote what their post should be about and pressed Generate, and nothing happened. The button was waiting for one more thing (a website, or one sentence about the business) and the screen never said so. Pressing it now says what is missing and puts the cursor there. A dead button is always the product's fault, never the user's.
One line in the wrong language. A post written in Czech came out with its last line in English. The person had also asked, in their own words, for the post to end with a call to order on their website and to speak to customers informally. The post did neither. The closing line is now written in the language of the post, and when someone asks for a specific call to action or a way of addressing their customers, that is what they get.
Somebody else's colours. A brand whose website is clearly purple got posts in teal. We were choosing the accent colour by how vivid it was, not by how much the brand actually uses it, so a colour from a single button beat the colour on every page. The brand's most-used colour now leads.
None of these people complained. Each one made a post, looked at it, and left. Each one was right.
Ask once, at the right moment
Recording what happened tells you where people stopped. It does not always tell you why. For that you have to ask, and most products ask too often and too early.
We ask "what stopped you?" once, ever, and only at a moment that earns the question: the second time someone opens a post full size and does not take it. If they dismiss it, we store that as a dismissal, because "did not want to answer" and "never asked" are different facts. If someone signs up and makes nothing at all, one email goes 24 hours later. That is the whole list.
And when we fix the thing that stopped someone, we write to that person and tell them. It is the cheapest way there is to get a second chance.
If you run a subscription product
For a subscription business the silent leaver is a customer who cancels. The idea is the same: ask why at the moment they leave, then ask the follow-up question, because the first answer ("too expensive") is rarely the real one. If you want that done for you, WhyLeft emails every customer who cancels, asks the follow-up you would ask if you had the time, and turns the answers into the one thing worth fixing next.
What we would tell another small team
- Record what people did, not what they typed.
- A dead button, a wrong language or a wrong colour is yours to fix, however quiet the user was.
- Ask why once, at a moment that earns it, and take "no" as an answer.
- When you fix it, tell the person who found it.