Skip to content
Knowledge / Evidence & Trust

Which Product Facts Can a Small Business Verify Itself?

Separate public, internal, supplier-held and not-reliably-measurable product facts, and understand what each evidence source actually proves.

Last verified
2 September 2026
Share
LinkedIn X Email
Navigate this page

Direct Answer

A small business can establish some product facts from its own controlled records. Other facts require supplier access, specialist evidence or a source outside the business. Publicly visible information is useful, but it cannot tell you what the business holds internally.

Use five observability classes instead of treating every blank as “missing”:

  • PUBLICLY_OBSERVABLE
  • PUBLICLY_INFERABLE_WITH_CAVEAT
  • INTERNAL_ACCESS_REQUIRED
  • SUPPLIER_ACCESS_REQUIRED
  • NOT_MEASURABLE_RELIABLY.

The key distinction is access, not confidence theatre. If a fact cannot be established from the current evidence, record what access is needed next.

Why this matters for a smaller business

A small team often has a mixed evidence environment. Some facts are in the storefront, some in a spreadsheet, some in technical files and some sit upstream with suppliers. If those states are compressed into “complete” and “missing”, the business loses the information needed to decide what to do next.

The same is true of public research. A product page can show that a business published a claim. It cannot establish that no internal record exists when the claim is absent.

NOT FOUND PUBLICLY ≠ BUSINESS DOES NOT HAVE IT

That rule prevents a common category error: confusing what an external researcher can observe with what a company can access.

The observability test

For each product fact, ask these questions in order.

1. Is the value directly visible in a controlled public source?

If yes, it may be PUBLICLY_OBSERVABLE.

Examples can include a published product identifier, listed fibre composition or an operator name. Public visibility does not by itself make the source independent evidence. It only establishes that the assertion was published.

2. Can the value only be inferred?

If the source supports a bounded inference but not a direct fact, use PUBLICLY_INFERABLE_WITH_CAVEAT.

Do not convert an inference into a clean product value just because it is convenient for a spreadsheet.

3. Would an internal record answer the question?

If the answer is likely to live in the business’s own non-public records, use INTERNAL_ACCESS_REQUIRED until those records are inspected.

Possible locations include:

  • product master spreadsheets
  • purchase or production records
  • technical files
  • customs records
  • certificates already held by the business
  • internal ecommerce fields not published to customers.

4. Does the answer sit upstream?

If the business needs a supplier, manufacturer, test house or upstream evidence holder to establish the fact, use SUPPLIER_ACCESS_REQUIRED.

This is not the same as saying the fact is absent. It says the next useful action is outside the current internal evidence boundary.

5. Can the fact be established reliably at all?

Use NOT_MEASURABLE_RELIABLY where the available method cannot support a defensible value. This is preferable to manufacturing precision from a weak proxy.

Evidence role is a separate question

Observability tells you where access is needed. Evidence role tells you what a source does.

Keep the three roles separate:

Evidence roleWhat it establishes
ASSERTION_SOURCEThe business or supplier states the value.
CORROBORATING_SOURCEA second source supports the assertion, but is not necessarily independent verification.
INDEPENDENT_VERIFICATIONEvidence independently establishes the underlying fact within its stated scope.

A supplier document can be useful and still be an assertion source. A laboratory report can be independent for one tested batch and irrelevant to another. Evidence role does not remove the need to check scope.

Scope still matters

A fact can be well evidenced and still be attached to the wrong product level.

Record whether the evidence applies to:

  • model or style
  • variant
  • SKU
  • batch or lot
  • individual item or serial.

For example, evidence about one production batch should not silently become a model-level truth.

Practical output: give every uncertain fact a next state

Instead of a blank, record:

  • current value, if any
  • observability class
  • evidence role
  • scope
  • source reference
  • next access requirement
  • current owner.

That structure turns “we do not know” into a governed state without pretending the answer is known.

Canonical Knowledge connections

What not to assume

Do not assume that:

  • public absence means internal absence
  • a structured field is more truthful than an unstructured document
  • a supplier statement is independently verified
  • a certificate covers every product carrying the brand name
  • a model-level document proves a batch-level fact
  • an estimated value is better than an explicit unknown.

Keep exploring

The question a small business usually reaches next.

Does this reach your products?

Give ActivateDigital one product and it works out which obligations apply from the product's own character, and says which it cannot decide.

Worth sharing?

Help someone else make sense of product passports.

LinkedInXEmail

Sources & legal basis

  • ActivateDigital Product Evidence + SME Original Research Programme v1.1: Product Evidence Method v1.1.
  • ActivateDigital Product Evidence + SME Original Research Programme v1.1: Observability Matrix v1.1.
  • ActivateDigital Product Evidence + SME Original Research Programme v1.1: Evidence Role Codebook.
  • ActivateDigital Knowledge: How We Know, and What a Blank Means.

Last verified: 2 September 2026.