HomeAnswersSavantCat: who we are, what we deliver, how to verify itHow can I verify your work? What is a 'process evidence trail'?

How can I verify your work? What is a 'process evidence trail'?

Published 2026-09-22 · Method: same inputs, same outputs

Direct answer

Process evidence is a set of records that show what was done and what resulted, and that you can reproduce: ① regression re-run records (same question set before and after a change) ② a change log for the transfer-to-human rules ③ a fact ledger with the source and collection date of every published fact ④ fixed-frequency monthly observations using the same questions. The point is that you can re-run it; a tool's score cannot be re-run because you do not control the algorithm.

Why we don't hand you a score

Plenty of tools will give you an 'AI visibility score'. The problem with a score is that you cannot see how it was computed and cannot reproduce it — 62 this week, 58 next week, and no way to tell what changed or why.

We do the opposite: we hand over the process, not a number.

What the evidence trail contains

ItemContentHow you re-check it
Regression re-runsSame question set, results before and after a changeRun the same questions again; results should match
Rule change logEvery adjustment to transfer rules: when, why, what it affectedCompare the log against live behaviour
Fact ledgerSource, collection date and reviewer for each published factSample entries and go back to the source
Monthly observationsFixed questions, fixed frequency, stable valuesCompare the same question across weeks for trend

One sampling discipline

A single observation is not a conclusion. Our rule: collect the same question at least three times within a week and record the stable value; read trends week over week, never off a single point.

How this differs from rank monitoring

Commercial rank monitoring gives you a position number while the vendor controls the algorithm; you cannot change or audit it. Process evidence gives you a series of actions and their results that you can re-run yourself at any time. One is a black-box conclusion; the other is an auditable record.

Key facts

kv
kv
kv

Sources

  • Our delivery acceptance criteria
  • T/CAPT 026—2026 chapter 8 (monitoring, traceability and process records)
  • Our fact base and sourcing ledger template (14 columns)

Follow-up questions

Why can't you give me a daily score like a tool does?

We can collect daily, but we refuse to compress it into one weighted score — a score cannot be audited. We give you the raw observations and the method.

Does the evidence trail cost extra?

It is included in project delivery. A standalone ongoing observation service is quoted separately by period and question volume.