Methodology

The Autopsify method

A Blueprint is a forensic reconstruction of a store's economics, not a dashboard. This is how the reconstruction is built, how each figure is classified, how the binding constraint is identified, and how a change is proved by the next reading rather than assumed.

Operator-led · Read-only Shopify access · Human-reviewed

The premise

Reporting is not diagnosis

Most ecommerce tooling reports. It shows revenue, orders, average order value and return on ad spend, and it shows them accurately. None of that establishes why margin moved, which of several plausible causes the data actually supports, or what to change first.

Diagnosis is a different job. It starts from the same source data and asks what can be established from it — then says so, including where the answer is that nothing can be established. A figure displayed on a dashboard has been measured. A figure in a Blueprint has been measured, reconciled, classified and, where the evidence does not reach, withheld.

The flow

Eight steps, in order

  1. Data
  2. Reconstruct
  3. Test
  4. Classify evidence
  5. Find constraint
  6. Act
  7. Re-diagnose
  8. Verify

The order matters. Evidence is classified before a constraint is named, and a constraint is named before an action is proposed, so a recommendation can always be traced back to the specific figures that justify it. Where that chain breaks, the Blueprint stops rather than bridging the gap with an assumption.

The parts

What each step involves

Economic reconstruction

The store's own order, product, refund, tax, cost and advertising data is rebuilt into a ladder: gross sales down through tax, refunds and cost of goods to gross profit, then advertising to contribution after advertising, and further only where the remaining costs are actually evidenced. How the ladder is built.

Evidence classification

Every figure is marked known, supported or unknown. The classification travels with the number, so a reader can see which conclusions rest on source data and which rest on a documented derivation. The evidence model.

Coverage assessment

Before a figure is reported, the share of the business it actually covers is measured — how much revenue carries a recorded unit cost, how much is attributable to a source, and whether the periods being compared align. Coverage decides how strong a conclusion the evidence can carry. Measurement coverage.

Finding the constraint

Not every true finding is worth acting on. The Blueprint ranks findings by the money involved and the strength of the evidence behind them, and names the one constraint binding hardest now — rather than a list of twenty observations of equal weight.

Action logic

Each action is tied to the finding that justifies it, with the owner, the effort and the signal that would show it working. Autopsify does not make the change: the actions are the merchant's, in the merchant's store.

Re-diagnosis and verification

A second cycle roughly 30 days later re-reads the same store on the same rules, against the baseline the first cycle recorded, and reports whether the figure moved — or that it did not. Verification & re-diagnosis.

The next constraint

Clearing one constraint usually moves the binding one somewhere else. The next cycle starts from the store as it now is, not from the previous report.

The discipline

Deliberately conservative

The method is built to under-claim. Four rules do most of that work, and they are the reason a Blueprint sometimes returns fewer conclusions than a dashboard would display.

Missing evidence stays missing

A cost that is not in the data is not estimated from a benchmark, a category average or a plausible assumption. It is reported as unavailable.

Unknown never equals zero

An absent figure and a measured zero are different findings. Treating the first as the second silently inflates every figure below it in the ladder.

Attribution is not claimed beyond the evidence

Where only part of revenue carries a recognisable marketing source, channel-level figures are reported as covering that part — not as covering the business.

Conclusions are bounded by source data

Every claim states the scope it holds over. A margin computed on the products that carry a recorded cost is reported as exactly that, not as the margin of the whole catalogue.

Run the method on your own store

£497 for one diagnostic cycle. A short conversation first, then a readiness check on your data — if the evidence cannot support the analysis, we say so before you pay.