Our team says the tool we built is no good. Now what?
Somebody asked me this on a call this week. He had built an AI tool aimed at recruitment firms, put it in front of experienced recruiters, and the feedback came back in one word. Crap. His word, repeated to me the way the recruiters had said it. He wanted to know which part of the tool to rebuild, and my honest answer was that nobody had written down what the tool was supposed to be taking off their desks, so there was no way to tell.
What that word is usually describing
When an experienced person calls a new tool rubbish, they are rarely reviewing the software. They are telling you it has added a step to a day that was already full. They have a way of doing the job that works, built up over years, and you have put something in front of them that asks them to type the same thing in a second place. Of course they hate it.
The other version of this is a tool that does something genuinely useful, in the wrong place. The work happens in an inbox and a notebook, and the tool lives in a browser tab nobody opens. Same outcome, same word.
The questions I would ask before touching the code
My first move is not a feature list. I go back to the people who called it rubbish and ask them about their own week, because their complaints are the only reliable source of what to build. The question I lean on most is the magic wand one. If you could perform actual magic, how would the morning go? What is on the list when you walk in, what happens on its own, what decisions are left for you, and how do you want the result handed to you?
Then I get specific about waste, because that is what turns a moan into something you can cost.
- What do you do more than once a day that feels like typing rather than thinking?
- How long does it take, and how often does it come back?
- Who else is held up while it is happening?
- What happens when you get it wrong, and how often is that?
- Where do you sit waiting for somebody else to send you something before you can start?
Five answers from four people and you have a short list of things that cost real hours. That list is what the tool should have been built against. It is also the part you can put a figure on, which matters when the person paying asks why this is worth doing.
What to do with the thing you already built
Keep the part that maps onto a named waste from those answers. Throw the rest away, properly, not into a backlog where it sits waiting to be defended later. Then show the smallest working version to the person who complained loudest, and ask them whether it removes a step or adds one. If they cannot tell you which, it is not ready to go near the rest of the team.
Going back and asking is slower than another round of building, and it feels like going backwards when you have already spent money. It still beats polishing something nobody asked for. Nobody had asked those recruiters anything, and that is what the word crap was describing.
Got a version of this problem? Twenty minutes minimum, seven questions, and a one-page Bottleneck Blueprint within 48 hours. If it's not a fit, we'll say so.
Book a call → Not ready to talk? Take the eight-question AI assessment → Questions about cost, data and ownership? Thirty-five straight answers →