Datasheet · Cohort Model report

How to use the Cohort model

Cohort payback tells you what happened. This report answers "what if" — move three levers (how many clients, what each costs, how much therapy they buy) and read off the profit, the payback month, and whether it covers the bills.

The idea

Cohort margin has exactly three levers

Every month's profit-from-new-clients comes from the same little equation: how many clients we win (volume), what each one costs (CAC), and how much each goes on to spend (the engine). The report lets you pull each lever and watch the answer move.

212
new clients (volume)
×
£120.84 − £82
what each is worth at 12 months, minus what they cost (engine − CAC)
=
+£8,234
projected cohort margin at M12
Those are the report's opening numbers — the real joined-cohort averages. Everything on the page recomputes live as you type.
The levers

What each input does

All five are seeded from real data, so the page opens showing today's machine. Change any of them to ask a question.

Apr 26engine

Engine — whose curve?

Pick a real cohort (or the blend of all joined cohorts). This sets how fast money comes back and what a client is worth: its net-25 per session × its sessions per client.

300new clients

New clients

How many we win in the month. Scales the answer up and down — but can't rescue a per-client loss.

£75CAC

CAC

Ads + placement-team shifts per client won. Comes straight off the margin of every single client.

6.8target SDR

Target SDR (by M12)

Sessions per client by their first birthday. The engine lever — every 0.1 is ~£1.78 of value per client.

£15,000overheads

Monthly overheads

The burn CAC doesn't cover — salaries, tools, support shifts. Ads and rota are already inside CAC, so leave them out here.

The three SDRs

Why "6.8 sessions" isn't oddly high

Three different numbers all get called SDR, and mixing them up makes every report look wrong. They're the same measurement taken at different ages.

3.9so far

SDR so far

Sessions per client to date — the New clients report column. It grows every month a cohort is alive, so young cohorts always look low. It's an age curve, not a quality score.

2.03@M1

SDR @M1, @M2…

Sessions per client at the same age — the fair comparison, in the engines table. April at 2.03 by M1 vs June at 1.85 is a real difference; 5.2 vs 1.4 "so far" is just age.

6.8implied M12

Implied M12 SDR

Where the cohort lands at 12 months if it follows the historical curve. This is the model's seed — a full-year number, so of course it beats every "so far".

Age 0
1.4
July cohort, weeks old
Age 3
5.1
April today
Age 6
6.8
January today
Age 7
7.3
December today
Age 12+
5.5–11.6
where finished cohorts actually landed
That row is the real SDR-so-far column from the New clients report, sorted by cohort age — it climbs almost perfectly with age. Every cohort that's had 6+ months is already at or past 6.8.
The range

"Why does the margin swing from £5k to £15k?"

The value-per-client estimate is actually quite tight. But margin is what's left after subtracting a fixed CAC — a thin slice off a big number — so a small wobble in value becomes a big wobble in profit.

Value per client at M12 — tight

£121
£105£143

−13% / +18% — the best and worst full-year curves real cohorts have followed.

Margin per client — the same band, amplified

£39
£23£61

Subtract the fixed £82 CAC and the modest band becomes −40% / +57%. Multiply by 212 clients → £4.9k to £12.9k.

The range isn't a broken number — it's the business truth. Per-client profit is a thin slice, so profit is hypersensitive to the engine. A small SDR slip (like May–June's) moves the cohort's profit a lot. That's exactly why SDR is a top-line KPI.
Burn & breakeven

One cohort is a one-shot deal. The machine is monthly.

A single cohort pays back once: spend ~£17k up front, get ~£26k back across 12 months, keep the difference. But we run one cohort every month — and once 12 are running at once, each month harvests slices from all of them that add up to exactly one cohort's full margin. So "projected cohort margin" doubles as "steady-state monthly profit before overheads".

+£8,234
one cohort's M12 margin = monthly profit before overheads, at steady state
£15,000
monthly overheads
=
−£6,766
short — today's levers don't cover the bills

So the report answers the breakeven question three ways, each holding the other levers still. At today's engine and CAC, £15k of overheads needs any one of:

387clients

Clients needed

At £82 CAC and SDR 6.8. Volume fixes it only while margin per client is positive.

8.6SDR

SDR needed

At 212 clients and £82 CAC. Ambitious — but finished cohorts have hit 8–11 before.

£50max CAC

Max CAC

At 212 clients and SDR 6.8. The cost ceiling the current volume and engine can afford.

Steady state is a destination, not today. The paid engine started March 2026, so only ~5 cohorts are harvesting — the window fills around Feb 2027. Until then each month earns less than the headline, and each cohort's cash is out the door until its ~M4–M5 payback.
Honest limits

Three things to hold while modelling

The curve is a bet

The cohorts that have proven 6.8+ sessions were the smaller pre-ads ones. The big paid cohorts are tracking the same curve so far — the engines table (SDR @M1, @M2…) is where any slip shows first.

Margin ≠ bank balance

The model talks steady-state profit. Cash lags it: CAC is paid on day one, revenue drips for a year. Growing faster makes profit look better and the bank feel tighter.

Same numbers everywhere

The engine curves, multipliers and CAC are the exact figures from Cohort payback, Cohort LTV and New clients — one source underneath, reconciled to the penny.