Most Stage 6 businesses meet the same problem the same way. The headline revenue chart looks healthy. The team is growing. The pipeline is working. And yet, somewhere in the financials, margin is leaking, and nobody can quite point to where. The default diagnosis is usually 'we have a churn problem' or 'CAC is rising'. Both are usually downstream symptoms. The actual problem is upstream.
Stage 6 businesses are at the size where customer mix matters more than customer count. A customer base of 500 contains, almost without exception, three or four cohort types — and one or two of those cohorts are quietly losing the business money. They cost more to onboard than they pay back. They consume more support hours than the spend justifies. They churn at a rate that wipes out the contribution margin before it compounds. They look like revenue from the top of the P&L, but they are loss-making at the unit-economics level.
The Workflow #1 cohort dashboard surfaces this. Once it does, the instinct is to fix retention for the loss-making cohort — better onboarding, more attentive customer success, expanded scope. That instinct is wrong. Or rather, it's slow. The faster lever is the kill-or-keep decision: stop acquiring those customer types at all, and stop trying to save the ones already in the book who shouldn't have been there in the first place.
Why the kill-or-keep decision is the Stage 6 lever
At earlier stages, customer count matters more than customer mix. Stage 2-3 businesses need pipeline, full stop. Stage 4-5 businesses need a premium ICP that holds price. By Stage 6, the business has enough volume that mix is the lever. Adding 10% more revenue to a healthy cohort compounds. Adding 10% more revenue to a loss-making cohort destroys margin further.
The maths is straightforward. If a cohort has a contribution margin of -£500 per customer over its lifetime, and you acquire 200 of them this year, that's -£100,000 of compounding margin damage that will surface across the next 18-24 months. Meanwhile, the right-fit cohort with a contribution margin of +£3,000 per customer is being under-served because the team's hours are being absorbed by the wrong-fit cohort. The opportunity cost is doubled: real loss on the wrong-fit, plus capacity stolen from the right-fit.
The fix is structural. It's not a customer success initiative or a retention programme. It's a kill-or-keep decision applied in three places simultaneously, with discipline.
The three places the kill-or-keep decision applies
1. Funnel top. The customer types that lose you money don't enter the pipeline anymore. Outbound targeting filters get rebuilt to exclude wrong-fit segments — by industry, by company size, by use case, by buying behaviour, by whatever the cohort dashboard surfaced as the disqualifying pattern. Inbound capture forms get updated so the wrong-fit visitor either self-disqualifies or gets routed to a lower-touch motion that's actually profitable. Ad targeting and audience definitions get tightened so spend stops feeding a top-of-funnel that the bottom can't profitably serve.
2. Sales calls. The qualifying script gets a disqualifier section. Specific questions, asked early, that surface the disqualifying pattern in the first 10 minutes of a call. The rep is trained to disqualify cleanly when the answers come back — not to fight for the deal, not to massage scope, not to discount into fit. A clean disqualification is a saved sales hour and a saved support hour later. The team initially resists this because closed-won is the comfort metric. The fix is to comp the team on right-fit closed-won, not on closed-won-anything.
3. Existing customers. This is the part Stage 6 founders skip. The book of business contains, almost certainly, customers from the loss-making cohort that were acquired before the discipline existed. Letting them stay is letting the margin damage compound. The triage runs on a timeline: 90 days to communicate a price increase that brings them closer to break-even, 180 days to downscope to a lower-touch service tier that's actually profitable to deliver, 365 days for graceful sunset where the relationship can't be made profitable on any tier.
Why existing-customer triage is the hardest piece
Three reasons. First, the relationship friction. The team has been serving these customers for years in some cases. There are personal relationships, account managers who care, founders who feel the loyalty. Telling a customer their price is going up by 40%, or that their service tier is changing, is socially expensive. Skipping the conversation feels easier in the moment.
Second, the revenue narrative. Cutting a customer from the book — even a loss-making one — shows up as a revenue dip in the next quarter. For a founder running on a board narrative or a quarterly investor update, that dip is a story to tell. The instinct is to avoid the story by keeping the customer.
Third, the optionality argument. 'We can fix this customer later' is a compelling story until the financials reveal that 'later' has been the answer for two years. Every quarter the loss-maker stays in the book is another quarter of compounding margin damage that doesn't get unwound until the kill-or-keep call is finally made.
The discipline is to run the triage on a timeline, not on a feeling. 90 days for repricing notice (the customer can accept the new price or self-select out). 180 days for downscoping (the customer moves to the lower tier or self-selects out). 365 days for graceful sunset where contracts or relationships require it. The timeline does the emotional work for you. Once it's communicated, the conversation becomes operational, not relational.
What the cohort dashboard reveals that financials don't
Standard P&L reporting is blind to cohort dynamics. Revenue by month, costs by month, margin by month — these capture the headline but lose the upstream signal. A business can have flat margin month-to-month while the underlying customer mix is shifting in ways that will damage margin in 12 months' time. The damage doesn't appear until the cohort matures.
The cohort dashboard from Workflow #1 captures the underlying. Customers grouped by signup quarter, product mix, deal size, and acquisition channel. Each cohort tracked against contribution margin over its lifetime, not just first-year revenue. The patterns become visible. The cohort that signed up via the wrong channel and never activated. The cohort that bought the wrong-tier product because the sales team discounted to close. The cohort that came from a campaign that overpromised and under-delivered.
Each of these cohorts has a story, and the dashboard names it. Once named, the kill-or-keep decision becomes specific and actionable. 'We stop running the campaign that produced cohort X.' 'We retrain the sales team to disqualify cohort Y.' 'We give cohort Z a 180-day window to migrate to the higher tier or sunset.' Each decision is a discrete action with a discrete timeline.
What this looks like in real businesses
A professional services firm we worked with had been growing 15% year-on-year on the headline number, but margin had been compressing for six quarters in a row. The cohort dashboard surfaced that one customer type — a specific industry vertical that had been historically high-volume but low-margin — was responsible for 70% of the margin compression. The firm had been adding 30 new customers a year from this cohort, each of which lost money over its lifetime. The kill-or-keep decision: stop outbounding to that vertical, retrain inbound qualification to disqualify, and give existing accounts in that cohort a 180-day window to migrate to a higher-touch tier or sunset gracefully. Within 6 weeks, margin direction reversed. Within 18 months, the firm was 20% smaller in customer count but 40% larger in margin. The customers who stayed paid back. The customers who left were always going to.
Why founders resist this
The most common resistance is the belief that volume is the lever. 'If we just acquire more, the maths will sort itself out.' At Stage 6 the maths doesn't sort itself out — it compounds in the wrong direction. More volume of a loss-making cohort is more loss, not more revenue.
The second resistance is the optionality argument. 'We don't want to close the door on a market segment.' The kill-or-keep decision isn't closing the door forever; it's stopping the bleed while the offer or the delivery model gets reworked to make that segment profitable. Once the contribution margin maths works for that cohort, the door reopens. Until it does, the door staying open is just the bleed continuing.
The third resistance is the team's identity. Sales reps who've been comped on volume push back on disqualification scripts because they feel like brakes. Account managers who've been trained on retention push back on existing-customer triage because they feel like betrayal. The fix is comp realignment plus structural conversation: this isn't punishment, this is the only way the business gets healthier, and the right-fit customers — the ones the team enjoys serving — are the ones who get more attention as a result.
The bottom line
Stage 6 businesses don't have a churn problem. They have an acquisition problem. The cohort dashboard reveals which customer types lose you money, and the natural instinct is to fix them with better service. The faster, more profitable lever is the kill-or-keep decision: stop acquiring them at the funnel top, disqualify them in sales, and triage the existing book on a 90/180/365-day timeline.
The early signal is uncomfortable — revenue dips temporarily, the team feels the loss of customers they'd grown attached to, the board narrative gets harder for a quarter. The late signal is what compounds: margin direction reverses, the right-fit customer base concentrates, and the cohort retention curve starts to bend at the right shape, not just the average shape. The growth that follows is the growth that pays back.
Frequently Asked Questions
Why is kill-or-keep faster than fixing retention for loss-making cohorts?
Retention fixes take 6-12 months to mature against the cohort curve, and only work if the underlying unit economics can be made positive. Kill-or-keep stops the bleed in 6 weeks. The two aren't mutually exclusive — sometimes a cohort can be made profitable through service redesign, and that work runs in parallel — but the lever that reverses margin direction first is the kill, not the fix.
How do we identify which cohort to kill versus which to fix?
The cohort dashboard surfaces contribution margin per cohort over the lifetime, not just first-year revenue. Cohorts where the lifetime contribution margin is structurally negative (typically because cost-to-serve exceeds price across any plausible scenario) are kill candidates. Cohorts where the contribution margin is negative but fixable through service tier change or pricing change are fix candidates. The dashboard makes the distinction quantitative, not gut-feel.
What happens to revenue when we run the existing-customer triage?
Revenue typically dips 5-10% in the first quarter as loss-makers self-select out. Margin starts moving in the right direction in 6 weeks. Within 12-18 months, total revenue usually recovers to the pre-triage level or higher because the team's capacity is now serving right-fit customers who buy more, refer, and stay longer. The board narrative for the dip quarter needs to be prepared in advance.
How do we handle the team resistance to disqualification scripts?
Comp realignment first. If the comp plan still pays on closed-won-anything, the team will keep closing wrong-fit deals and the disqualifier scripts will be ignored. Restructure comp so right-fit closed-won pays out and wrong-fit closed-won doesn't, and the behaviour follows the maths. The structural conversation — 'this is how we get to a healthier business that's better to work in' — sits on top of the comp restructure, not instead of it.
What if the loss-making cohort represents 30%+ of the customer base?
Then the triage takes longer and the dip is bigger, but the discipline is the same. Run the timeline (90/180/365), absorb the dip, and let the right-fit base concentrate. The alternative is letting the loss-making cohort continue to absorb capacity and compound margin damage indefinitely. There's no version of Stage 6→7 graduation that includes preserving a structurally loss-making customer base. The maths simply doesn't work.