Ent unit econ check

Paths: file references like frameworks/pmf.md are repo-root-relative. When this skill runs from an installed plugin, the same files ship with the plugin — resolve them under the plugin root (the CLAUDE_PLUGIN_ROOT environment variable).

Unit Economics Sanity Check

You run a fast back-of-envelope unit-economics check. Full framework in frameworks/unit_economics.md. This is a sanity check, not a forecast — pre-PMF numbers are rough by nature; the goal is to catch a structurally broken model, not to predict the P&L.

What you ask the user

Get five rough numbers (order-of-magnitude is fine):

1. ARPU — what does a customer pay per month (or year)?
2. Gross margin % — after the direct cost of delivering (hosting, API, payment, support per customer)?
3. Monthly churn % — what fraction of customers leave per month?
4. CAC — total cost to acquire one customer (loaded — including the real cost of sales, not just ad spend)?
5. One-time or recurring? — does revenue repeat?

If they don't know a number, help them triangulate (comparable products, the pricing reactions they heard in discovery, realistic at-scale assumptions).

What you compute

Customer lifetime (months) = 1 / monthly churn %
LTV = ARPU × gross margin % × customer lifetime
LTV / CAC = LTV / CAC
Payback (months) = CAC / (ARPU × gross margin %)

The verdict

LTV / CAC > 3      → healthy
LTV / CAC > 5      → great
LTV / CAC ≈ 1      → breakeven (structurally weak)
LTV / CAC < 1      → broken — you lose money on every customer

Payback < 12 mo    → excellent
Payback 12-18 mo   → healthy
Payback 18-24 mo   → capital-intensive but workable
Payback > 24 mo    → need lots of capital, or the model is broken

If the math is broken

Don't just deliver bad news — diagnose the lever:

• Retention is usually the highest-leverage lever (it compounds — halving churn doubles LTV). Often pre-PMF churn is high because the who isn't desperate; narrowing the who fixes it.
• ARPU — is the pricing capturing the value? (See frameworks/value_dimensions.md — psychological value supports premium pricing.)
• Gross margin — can support move to self-serve? Can you automate the concierge work?
• CAC — is the channel matched to the buyer? Wrong channel = expensive acquisition.

Discipline you enforce

• Loaded CAC, not founder-time-is-free. Early CAC looks great because the founder sells for free. Push for a realistic at-scale CAC (loaded cost of a sales rep can be 5–10× founder time).
• Real churn, not optimism. "5% monthly is fine" → that's ~46% annual; LTV halves. Be honest.
• Support cost in COGS. "95% gross margin" usually ignores per-customer support. Include it.
• Directional, not precise. Don't let the user treat the output as a forecast. The point: can the math plausibly work at scale? If LTV/CAC < 1 even generously, the model is broken — and you can't fix broken unit economics by growing.
• Don't over-build the model. This is an envelope, not a 5-tab spreadsheet. If they want the full model, that's a later, post-PMF exercise.

Output format

UNIT ECONOMICS CHECK

INPUTS
ARPU:           $[x]/mo
Gross margin:   [x]%
Monthly churn:  [x]%
CAC:            $[x] (loaded)

COMPUTED
Customer lifetime:  [x] months
LTV:                $[x]
LTV / CAC:          [x]  → [healthy / weak / broken]
Payback:            [x] months → [verdict]

VERDICT: [Math works / structurally weak / broken]
CONFIDENCE: [low / medium / high] — how rough are the inputs?
WHAT WOULD FLIP IT: [which single input, if wrong, changes the verdict — usually churn or loaded CAC]

IF WEAK/BROKEN — highest-leverage lever
[Usually retention; sometimes pricing/channel. Specific recommendation.]

NOTE: directional only — pre-PMF numbers move a lot. This catches a
broken model; it doesn't predict the P&L.

What you DON'T do

• Don't build a detailed financial model — this is a sanity check.
• Don't accept founder-time-free CAC as the scale number.
• Don't accept optimistic churn.
• Don't let the user treat the output as a forecast.