The Meeting I Keep Getting Called Into
I've been called into the same meeting a dozen times. The logo on the wall changes. The meeting doesn't.
It's a growth review at a Series A startup. Twenty people, eighteen months of runway quietly turning into twelve. Signup-to-paid conversion has been stuck near 2% for two quarters. The team is not lazy. They've run thirty-odd A/B tests on the pricing page. They've rewritten onboarding twice. They've retrained the sales reps and rebuilt the demo. The conversion chart is the first slide in every meeting, and every meeting ends with one more experiment to move it.
The head of growth walks me through the funnel step by step, where people land and where they drop. It's a good teardown. Then I ask one question that isn't about the funnel. Who is this for?
The room answers four times. The founder says mid-market ops teams. The account executive says whoever will take the call. The PM says the power users who file the most tickets. Marketing says the persona in the deck, a composite named Alex. Four people in one room, each naming a different customer with a different Job.
So I ask the next one. Which Job does this product do better than anything else these people could use? The room goes quiet. Then the founder says it. The product is great at closing the books fast at month-end, and the finance teams who tried it loved exactly that. But a year earlier the team had repositioned around something grander, the all-in-one operations hub, and started chasing operations leaders. Those buyers carry a dozen Jobs and no burning one. The product was strong. It was strong at a Job the team had stopped selling, for a segment they had stopped pointing at.
That conversion number was never going to move — no pricing page, onboarding rewrite, or sharper demo was going to reach it. The leak wasn't in the funnel. It sat two layers up, in who the product was for and which Job it was being sold to perform. They had spent two quarters and a slice of runway sanding the bottom of a chain that was cracked at the top. The hard part to watch was the team. Smart people were working nights on a number that could not move, because the work that would move it sat nowhere near the number.
The model in the room, the one nobody says out loud, goes like this. The metric that's red is the problem you have. Conversion is low, so conversion is the work. Activation is low, so the onboarding team owns it. Churn is up, so lifecycle ships a winback email. Each number has a dashboard, an owner, and a toolkit, and you fix it where it lives. It feels like rigor. You can point at the number you moved.
A Product Runs on Cause and Effect — and You Have to Respect It
In the last chapter I said a funnel can be perfectly optimized and still fail. This is the diagnosis that follows from it.
A product runs on a strict cause-and-effect chain, and the chain runs in one direction:
- a market with money
- a segment and its Job — one choice, not two
- value — the Job performed above the segment's bar
- three conditions that must hold at the same time:
- the business model and the unit economics work — the math on each customer is positive
- you can create demand — reach these people in words they recognize, at an acquisition cost the math survives, with leads that actually close
- you can scale — operations and support keep up with growth without quality falling apart
- the economics keep closing at scale — people convert, stay, and buy again
- profit
Each step stands on the quality of the step above it. The three middle conditions don't take turns — they have to hold at the same time, and any one of them failing stops the chain. Higher up, two words do quiet work: Job and value.
Value is performing the segment's Job more efficiently than anything else those people could use, measured against the bar they walked in with. Meeting the bar isn't enough — to make anyone switch, you have to beat it.
Both the efficiency and the bar come from the Job. Choose the wrong Job and you build efficiency against criteria nobody in your market holds, and beat a bar nobody is measuring you against. The value can be real and still land on no one.
A weak step can't be rescued from below. It can only be rescued from above. If the segment can't pay enough to cover what it costs to serve them, the economics never close, however high conversion climbs.
This is why the red number lies to you. It sits at the bottom of the chain, so it is a consequence, not a cause. Low conversion is a Problem in the precise sense: the result of a product hired for a Job and then underdelivering against that Job's criteria. Treat the consequence as the cause and you ship fix after fix that the real cause quietly regenerates.
Once the error is at the top, work at the bottom can't reach it. You can run the best pricing-page test of your career and still not move a number that's broken three steps up. The data-driven version of this trap has its own twist: most experiments fail even when you're testing the right thing. Ron Kohavi, who built the experimentation platforms at Microsoft and later ran experimentation at Airbnb, puts numbers on it. Over two-thirds of well-designed ideas fail to move the metric they were built to move, and at Airbnb only 20 of 250 tested ideas had a positive impact. If two of every three tests miss on the right surface, testing harder on the wrong surface is close to pure waste.
So when conversion is low, it's almost always one of three things, or some mix of them. First, the Job or the segment is wrong: you're standing in front of people who don't have an urgent version of the Job you perform. Second, the value is wrong for that segment's Job: you're better at something they don't rank highly. Third, the communication is wrong: you do perform their Job well, but the page describes it in words they don't recognize as their own. Map them back onto the chain: the first is its second link, the second is its third, and the third lives in the demand condition — between the value and the people it was built for. All three live above the funnel. The funnel is where the symptom shows. It's rarely where the cause lives.
The third break deserves a longer look, because teams misread it most often. Communication is the transmission of value that already exists, and it only works in the segment's own words. Fast, simple, powerful are adjectives any competitor can also type; value lands in concrete criteria. DoorDash sells dinner at the door in 35 minutes, not a convenient dinner. And a page that leads with the mechanism — the AI, the workflow engine, the integrations — asks the customer to translate your machinery into their Job. Most customers won't do that work.
The fix is cheap and slightly embarrassing: take the sentences paying customers use to describe what the product did for them, and put those sentences on the page. That cheapness is also diagnostic. Communication is the only one of the three breaks you can fix without touching the product, so it's the first thing to check on the walk up the chain. If the page already speaks the segment's criteria in the segment's words and conversion still doesn't move, the break is higher — in the value or in the segment itself.
The trap doesn't need an org chart. A sales team that isn't closing reaches for a new deck and tighter objection handling — sharpening the last step while the break sits three steps above it, in the segment, the Job, or the risk the offer leaves on the buyer. A solo builder is in the same spot: vibe-code an MVP in a week, then spend a month rewriting the landing page, thirty tweaks and a new headline every Friday, while the people the page points at never had an urgent version of the Job at all. The chain breaks in the same places at any size — and lies through the same bottom number.
Back in that growth review, the diagnosis took about ten minutes once we stopped staring at the funnel. The segment was wrong and the Job was wrong, two layers up. Everything beneath them had been built faithfully for the wrong people: the value, the page, the demo, the conversion number. The team didn't need a better experiment. They needed to climb back up the chain and choose again.
Wes Defends His Dashboard
The pull is identical at every scale, with or without a CRO: fix the thing in front of you. The discipline is the same too — before you touch the number, walk up the chain and ask whether the Job, the segment, the value, or the words are what's actually broken.
Where the Error Gets Made
The most expensive error sits at the top of the chain — so the real question is how a team ends up choosing it without ever noticing.