ICPSignalOpening a feed full of irrelevant posts gets old quickly. The temptation is to rewrite everything or assume the channel can't work. I'd pause long enough to name the problem more precisely.
Wrong people can mean several things. You may be seeing other vendors, people learning the subject, buyers outside your scope, or relevant people whose questions are too early for a sales reply. Each needs a different correction.
Group the results you would reject
Review a manageable batch and give each rejection a concrete reason. Keep the original link beside the reason so you can check your judgment later. You don't need an elaborate reporting system to start.
I'd avoid a single ‘bad lead’ category. It hides the distinction between an incorrect match and a correct match that needs a different response. The next change depends on that distinction.
| Pattern | What to inspect |
|---|---|
| Mostly vendors | Whether the profile describes the buyer’s problem or the seller’s category. |
| Mostly tutorials | Whether you have distinguished learning from looking for help. |
| Requirements you cannot meet | The product or service constraints in the profile. |
| Relevant but early questions | Whether an educational reply fits better than a sales request. |
Make a change you can explain
If vendor promotions dominate, clarify who needs the service and exclude people offering it. If the results describe a different kind of project, specify the scope you handle. Use the pattern in the posts to choose the adjustment.
I'd change one major part at a time. Otherwise, an improved batch won't tell you which revision helped. Keep a copy of the earlier brief in your notes so the decision remains understandable.
Read the source behind the explanation
ICPSignal provides a potential-fit explanation with matching posts. Inspect whether that explanation relies on evidence in the original conversation. A broad phrase may have been interpreted in a way you didn't intend.
Also check the surrounding discussion. A useful question might already be resolved, or a reply may reveal a requirement that changes the fit. Discovery and review both matter. Tuning the profile won't remove the need to read.
Look for a change in the kind of results
After revising the profile, review another batch under comparable conditions. Count relevant posts and inspect the remaining rejection reasons. A smaller feed can be useful if the posts better match the problem you solve; volume alone won't settle that.
If results stay poor, revisit the audience, language, and available sources. Keep the conclusion specific. ‘We aren't finding current hiring requests in these sources’ gives you something to investigate. ‘The leads are bad’ leaves you in the same place.