The setup
I run three funnels. A sales pipeline at my day job. A GTM Engineering course I teach MBA students — which is, on the nose, a course about building funnels. And this blog, which I keep meaning to feed.
All three are measured the same way: by counting events. Deals created. Demos booked. Emails sent. Page views. The dashboard is a bar chart of things that happened.
And all three have the same failure mode — the one the bar chart can't see.
The struggle
Two strangers said the same sentence to me in the same week, in different words.
An indie hacker building a product: "My users didn't complain. They just disappeared. 67% of them at the same step." No support tickets. No bad reviews. Just silence at a specific point in his funnel, invisible on a dashboard that counts signups.
A RevOps lead on Reddit: "A deal that looked healthy two weeks ago is now completely dead — and there was almost no signal in HubSpot. I'm talking about slow, quiet decay. This has burned me multiple times — lost $200K+ last year."
And a pipeline auditor, the bluntest of the three: "Revenue rarely disappears in big dramatic losses. It leaks quietly, in the middle of the funnel."
Then he listed the four leaks he finds in every audit, and I recognized all four from my own job:
- Leads with no owner.
- Deals stuck in a stage for 30+ days with no next step.
- Proposals sent with zero follow-up.
- Lost deals closed with no reason logged.
Read that list again. None of those are bad metrics. They're blank fields. The signal was never something that happened — it was something that stopped, or never got written down at all.
The breaking point
Then I looked at my own blog, the third funnel, and the joke landed on me.
I measure this blog by page views. A reader who arrives, reads nothing, and leaves quietly contributes a page view and zero complaints. I have no idea how many of you left quietly, because I built a dashboard that counts arrivals, not departures. My readers didn't complain. They just left.
Same sentence as the indie hacker. Same sentence as the RevOps lead with the $200K hole.
And here's the part that stings: I'd already written this law down once — for the other silent system I run. My second brain runs cron jobs. When a cron queue runs dry, it doesn't crash. It doesn't error. It doesn't log anything alarming. It just keeps firing against nothing, and every green checkmark says "all healthy" while the content engine quietly stops producing. I named that one drained queues fail silently.
The sales pipeline is just the next queue that fails silently.
The resolution — and the part nobody's selling
Here's the punchline. Clari, Gong, and HubSpot all already know about this silence. HubSpot's own copy says it outright: "Most deals don't end with a 'no.' They end with silence."
And then all three sell you the same fix: a black-box score. Gong feeds 300+ signals into an ML model and hands you a percentile. Clari sells an "Opportunity Score." HubSpot sells "Deal Insights" with a "Predictive Deal Score."
Here's the thing none of them will tell you: the silence was never a pattern to predict. It was a blank field.
A deal with no next step doesn't need a model to predict it's at risk. It needs a human to notice the field is empty and ask why. The four leaks aren't signals — they're missing data. And a query finds them:
WHERE next_step IS NULL OR last_activity < now() - interval '14 days'
That's it. No 300 signals. No seat license. No faith in a black box. I can write that query in an afternoon, and it will name every silent deal in my pipeline this week.
The score asserts the risk. The blank derives it.
The lesson
Dashboards are built to count events, and the funnel's most expensive failure is an absence. So the honest question isn't "what did we do this week?" It's "what stopped, and where's the blank?"
- The step where the drop happens is the diagnosis. "67% disappeared at the same step" and "deals stall at 'proposal' for 30 days" are the same sentence. The where tells you product vs. process vs. follow-through. The that is useless.
- A blank is cheaper to fix than a churn model. Naming "this deal has no next step" and asking why costs a manager ten minutes. A predictive churn model costs six figures and still reads the same blank fields you left empty.
- The empty field isn't a data-quality bug to paper over. It's the exact place the funnel is leaking. Fix the blank, and you've fixed the leak.
The moat was never the fancier score. It was the discipline to read the blank.
What's next
This week I'm building the weekly gap audit — a read-only query over my pipeline that lists the silent deals: no owner, no next step, no activity in 14 days, closed-lost with no reason. Three to five named deals, each with a recommended action. No org buy-in, no new tool, no re-architecture. Just me reading the blank.
So here's my question, and it's a genuine one: which funnel are you measuring by events — and where's the blank in it?
Sales, product, content. Doesn't matter. The funnel doesn't die with a bang. It dies with a blank.