The setupA team that was good at the part they could see
Picture an owner-led service company a few years past startup. Paid ads and referrals bring in a steady stream of inquiries, there is a CRM, an automated first text goes out on every new lead, and a small sales team handles the conversations. Demand is not the problem. The people are not the problem: when a lead actually reaches an answer, more than one in three becomes a customer, and the ones they win they win fast.
When we open a pipeline like this, we pull every opportunity from a season, call it 1,000, read the message threads and the call timing behind them, and sort each one into won, lost, or still open. Call recordings are usually not stored, so the teardown is of what the system can see. That limit is itself the first finding.
Finding 1The pipeline isn't losing deals. It's holding them.
Ninety-five deals were lost cleanly: the prospect said no. That is healthy. A clear no closes the file and frees attention.
850 opportunities, 85% of the pipeline, are still open with no outcome in either direction. Some have a text or two from a rep; most have the automated greeting as the last event. They look like an active pipeline on the dashboard. Only 150 of the 1,000 ever reached a yes or a no.
Nobody decided to leave them there. Nothing in the operation is responsible for moving them, so they sit. That is the difference between a team and a system: a team does the work it can see, a system decides what has to happen next and makes sure someone sees it.
Finding 2The wins and the losses start the same way and diverge in the first hour
| Illustrative comparison | Wins | Losses |
|---|---|---|
| First human reply after the prospect responded | 1 to 3 minutes | 30 minutes to 7 hours |
| Calls longer than 3 minutes | 2 to 3 | 0 to 1 |
| Touches before an outcome | 20 to 30 | 10 to 15 |
| Prospect called in at least once | about half | under 10% |
The wins share a shape: the prospect replies to the greeting, a person picks up within minutes, asks for a photo or two, gets them within hours, and has one real call of five to fifteen minutes inside the first two days. The estimate goes out during or right after that call.
The losses have the same first step and then a gap. Not because anyone chooses to wait; because nothing tells them a reply is sitting there, and the reps are on other calls. A customer comparing three providers has usually talked to one of them by the time the gap closes.
We're not claiming the minutes cause the wins. We're claiming that nobody is measuring the minutes, so the difference between the two columns stays invisible until someone reads a thousand threads by hand.
Finding 3The estimate is doing a job it can't do
In the wins, a long call overlaps with or immediately follows the estimate. Questions get answered as they come up, and the number lands with a person attached to it.
In the losses, the estimate goes out and the next event in the thread is silence, until an automated message a week or two later finally surfaces the no. Nobody decides to send the estimate and walk away. It is just what happens when the next step after "estimate sent" belongs to no one.
The estimate is not the close. The conversation during the estimate is. A system has to own the hour after the number goes out.
Finding 4A no closes the file. It doesn't teach anything.
The team handles a no well: thanks the prospect, closes it within minutes, moves on. Correct and consistent. The only thing missing is one question before closing, not to rescue the deal, but to learn. The few losses that explain themselves each point at something fixable in the offer. Even one in five answers to "what changed?" is product, pricing, and competitor intelligence a business has no other way to collect.
Finding 5The numbers the owner trusts aren't quite the numbers
A handful of the wins carry data errors when audited against the threads: an inquiry from a reseller marked as a customer, a zero-dollar value, a price that doesn't match anything on the menu. None of it is dishonest. It is the normal drift of a CRM that people update when they remember to.
This is the Belief, Intention, Reality gap in one screen. The owner believes the pipeline is moving and the wins are wins. The intention is a clean record. The reality is whatever the last person to touch the CRM had time for. If your systems don't enforce your intent, your reality will drift.
What gets installedFive rules the operation runs on
None of this is fixed with a pep talk, and none of it needs better people. Each finding becomes a rule that runs whether or not anyone remembers it that day.
| Leak | Rule | Where it lives |
|---|---|---|
| Estimate, then silence | A post-estimate cadence at 30 minutes, 24 hours, 72 hours, and 7 days, so every estimate has a path to a yes or a no within a week; silence at day 7 closes the file as "no response," which clears the pipeline but never counts as an answer. | The workflow, with the rep's personal message at each step |
| The gap after a reply | Under 5 minutes to the first human reply on any inbound response. | An alert on prospect reply; the number sits on the weekly log |
| Opens that never resolve | Every open opportunity gets a path to an answer within 7 days. Won and lost both count as closed. | The closed-with-answer ratio (won plus an explicit no, over everything) is the primary KPI; no-response closures stay in the denominator |
| A no that teaches nothing | One question before every close. The answer goes to the objection log; no chasing. | The playbook, reviewed monthly |
| Records that drift | A five-minute monthly audit of won and lost against the actual thread. | The monthly review, owner-visible |
What changes for the ownerFrom reading threads to reading three numbers
Before the teardown, the owner's picture of the pipeline comes from opening conversations and asking the team how it feels. After it, the operation reports three numbers a week. In the example: closed-with-answer at 15% (target 25% inside 60 days), first human reply at 4 minutes (target under 5), and estimate-to-close at 22% (target 50%).
The owner's job moves from being the follow-up to seeing whether the follow-up is happening. That is the whole point of an operating system: the work leaves people's heads, and the numbers stop being a feeling.
Check your ownThree questions that take ten minutes
Open your CRM and answer these honestly.
1. Of the opportunities older than a week, what share have had no human message in the last 7 days? If it's above a third, you have an 850 problem, and it isn't a people problem.
2. What happened in the thread after your last five estimates went out? If the answer is "nothing until they replied," the estimate is doing a job it can't do.
3. When was the last time a lost deal told you why? If you can't remember, you're paying for market research you never collect.