Evidence standards · Last reviewed 20 October 2026

How INKSIGHTS turns studio data into a recommendation.

Every useful conclusion should be traceable back to evidence. This page explains how INKSIGHTS collects, labels, compares and validates information before it becomes a finding, recommendation, benchmark or client result.

The core method

Evidence first. Diagnosis second. Intervention last.

INKSIGHTS does not treat an observation as a conclusion. The method keeps each step separate so uncertainty does not quietly turn into a fact.

Observation

Metric

Evidence

Finding

Diagnosis

Opportunity

Recommendation

Intervention

Outcome

Attribution

Learning

Important: missing information is recorded as unknown or missing. It is not automatically replaced with an estimate. When a model or assumption is necessary, it is labelled as such and should not be presented as a verified result.

Evidence labels

A reader should be able to tell what kind of claim they are looking at.

These labels are used to stop measured facts, calculations, estimates and hypotheses from being blended together.

VERIFIED

Confirmed against a primary source, connected first-party system or repeatable direct check.

OBSERVED

Directly visible in a public or supplied source, but not independently verified beyond that observation.

CALCULATED

Derived from disclosed inputs using a reproducible formula or transformation.

MODELLED

An estimate produced from stated assumptions, ranges or scenario inputs rather than a measured outcome.

HYPOTHESIS

A testable explanation or opportunity that requires evidence before it can be treated as a finding.

Source hierarchy

Use the strongest available source, and preserve provenance.

A recommendation is only as trustworthy as the evidence beneath it. Source strength, date, geography and limitations are therefore part of the record.

First-party operating data

Studio CRM, booking, payment, analytics or other permissioned operational records.

Connected platform data

Authenticated data from approved systems such as Search Console, analytics, CRM or payments.

Primary public evidence

The studio website, Google Business Profile, published policies, booking pages and official profiles.

Independent public evidence

Relevant directories, reviews, industry publications and other attributable third-party sources.

Modelled or estimated inputs

Used only when necessary, labelled clearly and kept separate from measured facts.

Provenance

Where the evidence came from, when it was observed, what geography it applies to and which record or URL supports it.

Versioning

Benchmarks, rules and scoring methods should carry a version so a later change does not silently rewrite an earlier conclusion.

Comparability

Studios should only be compared with a cohort that is sufficiently similar for the comparison to be useful.

Confidence

Confidence should fall when the source is incomplete, stale, estimated or difficult to verify.

Benchmark standard

A benchmark must disclose what it actually represents.

A number should not be called a UK tattoo-studio benchmark simply because it looks plausible.

Define the cohort and geography before comparing studios.

Record the observation period and freshness of the underlying data.

Disclose sample size where a sample-based claim is published.

Separate public observations from permissioned first-party operating data.

Do not report a cohort median or percentile until the comparable dataset is large enough to support it.

Label controlled reference bands when they are not proprietary market observations.

The current Pricing Benchmark therefore distinguishes its controlled reference bands from future proprietary studio observations rather than implying that an early reference range is already a statistically representative UK market benchmark.

Case-study standard

A result needs a baseline, an intervention and enough time to measure it.

INKSIGHTS will not treat a screenshot, one unusually good week or an unattributed revenue change as proof of impact.

Baseline

State what was happening before the intervention, including missing or unreliable measures.

Intervention

Describe exactly what changed, who owned it and when it went live.

Measurement window

Use a window long enough to distinguish an outcome from short-term noise.

External factors

Record staffing, seasonality, ad spend, price changes and other material changes.

Attribution

Claim only the part of an outcome that the evidence can reasonably support.

Permission and limitations

Publish identifiable client evidence only with permission and include limitations alongside positive results.

The INKSIGHTS Proof Library keeps system demonstrations separate from verified client outcomes for this reason.

Corrections and updates

Research should be correctable.

Public evidence changes. Software changes. Prices change. A methodology that cannot be updated becomes misleading.

Re-check

Time-sensitive public claims should be rechecked before a material update or new recommendation.

Record

Material changes should update the page date, evidence record or dataset version instead of silently replacing history.

Correct

When a material published claim is shown to be wrong, the preferred action is to correct it rather than defend the old version.

INKSIGHTS does not guarantee search rankings, bookings or revenue. The purpose of the method is to improve decision quality and make claims auditable—not to turn uncertainty into certainty.