Your users keep asking for features you already built
Why the most common feature request is for something already shipped, why docs and tooltips rarely fix it, and what does.
Quick answer
When users ask for a feature you already have, the feature is not missing, the path to it is: it sits behind a name they would never search for, inside a screen they never open, after steps they do not know. More docs, tooltips and release emails help the people who read them, which is rarely the person asking. What closes the gap is letting users say what they want in their own words and having the product do it, with a confirmation before anything changes. You also learn the exact words your users use, the best input there is for naming and navigation.
It arrives as a feature request, on a sales call or at the bottom of a feedback form: "it would be great if we could export only the unpaid ones". You can. You shipped it months ago. It is a filter and an export button, two screens apart.
The tempting conclusion is that this user did not look hard enough. The useful one is that if one person asked, many more gave up without asking, and the feature is not missing. The path to it is.
Why it keeps happening
Different words. Users look for things in their own language. You named the feature in yours, and the two rarely match: they search for "unpaid", the filter says "status: outstanding". A search box does not help when the word they type is not the word you used.
Depth. Plenty of real capability lives three levels into settings or behind a menu that only appears once something else is configured. Each level halves the number of people who will ever see it.
It is two features, not one. What the user wants is often a combination, a filter and then a bulk action, or a recurring schedule plus an approval step. Each part exists. Nobody would think to look for them together.
They looked once, at the wrong moment. Someone checks for a capability during a trial, does not find it, and files it as missing for good. When they need it again they ask, or they leave.
Why docs, tooltips and release emails rarely fix it
Each of them helps a little, and none of them helps the person who asked. Documentation is read by people who already suspect the feature exists. Tooltips appear on screens the user never opened. Release emails land weeks before the need. A redesigned navigation is a big project that fixes a handful of names. All of them depend on the user looking in the right place with the right word, which is the exact thing that failed.
Let them ask in their own words
What closes the gap is meeting users at the moment they want something, in the words they already have. An AI agent inside the product lets them type "export the unpaid invoices from March", works out that this is your status filter plus your export, and does it for them, as them. Anything that changes data stops on a confirmation with the real values first, so nothing happens without their yes.
The same move covers the combinations and the deep settings that no navigation will ever make obvious, because the user never has to know where any of it lives. They describe the outcome, and the product carries it out.
The side effect worth more than the feature
Every request becomes a record of how your users actually talk about your product. The requests the agent completed tell you which features were hidden. The ones it could not complete tell you what people expected your product to do and did not find, which is a better roadmap input than any vote count. With Verb, the requests no action could handle are grouped for you on the dashboard, in the users' own words.
Use those words. Rename the filter to what people type. Move the setting they keep asking for up a level. The agent covers the long tail, and what it hears makes the rest of the product easier to find.
If your users get stuck later in a workflow rather than finding it in the first place, product tours versus an AI agent covers that moment. To see what setting one up involves, the docs start with the four steps.
Frequently asked questions
Why do users ask for features that already exist?
Because they look for it in their own words, and the product names it in yours. Add a feature that lives three levels into settings, or one that is really two features combined, and most users will conclude it does not exist long before they find it.
Will better documentation fix it?
It helps the users who read documentation, and those are rarely the ones asking. Docs, tooltips and release emails all depend on the user looking in the right place with the right word, which is the exact thing that failed in the first place.
How do I find out which features users cannot find?
Collect their words, not just their votes: feature requests, sales call notes, feedback forms. If users can ask for things inside the product, the requests it could not complete are the most direct list there is, written in exactly the language your users think in.
Does an AI agent replace good product design?
No. Clear names and a navigation that matches how people think still matter, and what users ask for tells you how to improve both. An agent covers the long tail: the combined, deep or rarely used things no navigation will ever make obvious.
Vishnu, founder of AskVerb @itsvishnups
Building the AI agent that lives inside a SaaS product and does the work its users ask for. Writing down what I learn along the way, including the parts I got wrong first.