The SaaS Growth Audit Framework

The six-layer stack

Each layer takes the output of the layer above it as its input — a weakness anywhere compounds downward, which is why fixing the wrong layer changes nothing.

• Layer 0 — Fit

  Do your market, product, channel and price actually fit each other? A gate, not a stage: if this layer scores below 12, every other score is unreliable. Fails as “our leads are low quality.”

• Layer 1 — Acquisition

  Can you reach the right people repeatably, at a cost your ACV can fund? Fails as “we need more top of funnel.”

• Layer 2 — Activation

  Does a new user reach real value before they give up on you? The cheapest layer to improve — and the least often owned by a named person. Fails as “people sign up and vanish.”

• Layer 3 — Retention

  Does the value hold once the novelty and the onboarding attention stop? Retention problems are usually Layer 0 or Layer 2 problems arriving on a delay. Fails as “we're growing but net is flat.”

• Layer 4 — Monetisation

  Are you charging for the thing that actually creates the value? The most commonly lowest-scoring layer. Fails as “everyone negotiates us down.”

• Layer 5 — Expansion

  Do existing customers become an engine, or an endpoint? Fails as “growth stops when spend stops.”

How the scoring works

Eight diagnostic questions per layer — 48 in all — each scored on the same four anchors: 0 nobody has looked, 1 someone has an opinion, 2 written down and measured, 3 measured, owned by a named person and driving decisions. Each layer totals out of 24, banded broken (0–11), fragile (12–17), solid (18–21) and compounding (22–24). A first honest pass at 14–16 per layer is normal and healthy for a Series A company.

The constraint rule

Your growth rate is not the average of your layers — it is their product, which is why a 50% lift on your weakest layer is worth the same as a 50% lift on your strongest, at a fraction of the cost. The framework's operating rhythm: work the lowest-scoring layer only, for one full quarter; if two layers tie, work the upstream one first; then re-score rather than re-plan. Success is not a layer reaching 24 — it is the bottleneck relocating.