Open the pipeline view and count. Forty open opportunities, say, spread across a few stages, and it looks like a company with real momentum. Now go through each one and ask a single question: is there a specific next step, with an actual date on it, that both sides agreed to. Do that honestly and the forty usually drops to something like six. The other thirty-four are not deals. They are names of people who replied to something once and never got demoted out of the pipeline afterward.
What actually separates a name from a deal
A real deal has a specific person attached, not a company. It has a problem that person stated out loud, in their own words, not one you inferred from their job title. It has a next step with a date on it, agreed by both sides, not a note that says follow up. And it has at least some visibility into whether they can actually buy: budget, timeline, or the authority to decide, even if that visibility is rough. Anything missing all four of those is a name, however promising the last conversation felt at the time.
Run the count on your own pipeline right now. Of everything sitting in an open stage, how many have a dated next step that another human being agreed to. That number, not the count of open rows, is your actual pipeline. It is usually the smaller, less comfortable one, and it is the one worth building a forecast on.
Why founders inflate their own pipeline without meaning to
Nobody drags a lukewarm reply into an open stage out of dishonesty. It happens because a full pipeline feels like evidence that the business is working, and an accurate one, six real deals instead of forty, can feel like an admission that it isn't. So a reply that says "sure, tell me more" gets logged as an opportunity and stays there for months, because moving it back out feels like giving up on something, even after it has gone quiet for six weeks. No CRM enforces honesty here. The optimism is the founder's, not the software's, and software cannot talk someone out of their own hope.
The actual mechanism: entry criteria with no exit criteria
Almost every pipeline has a clear bar for getting in: someone replied, or took a call, or asked a question. Almost none has an equally clear bar for getting back out. Silence for six weeks should mean something specific happens to that row, and in most pipelines it means nothing at all, because nobody wrote the rule down. The fix is not a better tool. It is a rule applied the same way every week: a deal needs a dated next step to stay open, and if that date passes with no update, it comes out of the pipeline that same day, not at the end of the quarter when someone finally does a cleanup pass. There is no stage in between called hope, even though most pipelines have one, unlabeled, holding most of their volume.
What a tool can do about this, and what it flatly cannot
It is worth being exact about where something like Rocketship actually helps here, because it is a narrower piece than it might sound. Classifying an incoming reply as interested, not now, or not interested is a real service: it stops the raw, unread version of this problem, where a promising reply just sits in an inbox and becomes a phantom opportunity nobody even looked at twice. Reading every reply and sorting it is useful input hygiene.
What it cannot do is apply the judgment above. Deciding that a friendly "tell me more" is not yet a deal, and deciding it again next week when the same account still has not moved, is a call a founder has to make and keep making. No software will demote a row out of politeness toward the person who sent it, and that reluctance is exactly the human instinct causing the inflation in the first place. Once the discipline exists, the pipeline stages, tasks, and custom fields in a CRM are a fine place to encode it. They are not where the discipline comes from.
The cost of the inflated number is not just embarrassment
A pipeline padded with names instead of deals does real damage beyond a forecast that misses. It tells a founder they have more selling capacity than they do, so a hire gets planned around a number that was never real. It tells an investor update a story that quietly falls apart a quarter later. And it costs the founder's own attention, because a dashboard showing forty open rows feels like it needs forty check-ins, when only six of them are actually asking anything of you this week. The other thirty-four were never going to answer differently no matter how often you looked at them.
The same test works on someone else's pipeline too
If you ever inherit a pipeline, from a co-founder, an early hire, or your own past self six months ago, run this test before trusting a single row in it. A pipeline handed to you with forty names and no dates attached is not forty leads you get to skip finding yourself. It is thirty-four names someone else was too optimistic to delete, plus six real conversations worth actually picking up. Treating the whole handoff as usable pipeline, rather than running the same date-and-next-step check on every row, is how the inflation problem quietly passes from one person to the next inside the same company.
Run the six-deal test this week
Go through every open row. For each one, ask whether there is a dated next step that another human agreed to. If yes, it stays. If no, decide right now whether to get one this week or take the row out of the pipeline. Do this once, honestly, and you will know your real number for the first time. It will almost certainly be smaller than the dashboard suggested, and it will be the first pipeline count in a while that you can actually trust when you plan the next month around it.