Diagnostic

Entity Coherence Product Truth and Claims

The system may find the facts and still connect them to the wrong thing. · 4 min read

Entity coherence, Product truth and claims in AI search

Two Products share a similar name. A founder’s biography is confused with the company profile. A regional service is described as nationally available. A discontinued offer remains associated with the Brand.

The facts may exist. The entity relationships are unclear.

The model at a glance

Entity coherence, Product truth and claims in AI search: the connected entity map, claims attached to the correct entity, sources and context, the AI observation layer and the risk of incoherence.

What entity coherence means

Entity coherence is the degree to which important organisations, Brands, Products, people, services and locations are consistently identified and correctly related across sources.

Common failure patterns

Parent And Subsidiary Blended Product Names Change Without Supersession Inconsistent Adviser Identities Regional Availability Omitted Claims Without Their Conditions Similar Services Merged Legacy Pages Still Accessible And Referenced

Product truth requires more than SEO metadata

Metadata is not Product truth.

What the organisation needs
Approved Names And Descriptions Clear Relationships Current Fees Conditions And Limitations Owner And Effective Date Canonical Sources Accessible Supporting Evidence Correction Routes For Partner Sources

An entity-coherence example

The wealth service is attributed to the retail bank.

A financial group has a parent Brand, a retail bank, a wealth division and several advisory Products. Public sources use the names inconsistently, so AI answers merge divisions and combine adviser biographies.

What the organisation may need
A Clear Entity Hierarchy Consistent Names And Descriptions Relationship Statements Distinct Canonical Pages Current People And Role Information Structured Data Reflecting Real Relationships Correction Of High-Value Third-Party Sources Clear Regional And Product Boundaries
Claims need context

True in isolation. Misleading in a summary.

“No monthly account fee” may require a condition about account type or eligibility. When the condition repeatedly disappears in summaries, the source architecture needs a clearer public explanation — not a hidden disclaimer.

Narrative Integrity examines whether the answer preserves the context the customer needs, not only whether isolated words are accurate.
What entity confusion leaves unclear
Who Provides The Service Who Is Accountable Whether The Product Is Available To Them Which Adviser Or Company They Engage Which Terms And Reputation Evidence Apply

That uncertainty weakens Trust even when individual facts are mostly correct.

A useful governance artefact

The Entity and Product Truth Register

Approved Names Relationships Descriptions Markets Owners Effective Dates Claims Qualifications Canonical Sources

It does not need to be public.

It gives Content, SEO, Data, Product and assurance teams one governed reference whenever a source is created or corrected.
What teams should review together

Eight things one review has to cover.

01 Product names and approved descriptions
02 Parent, Brand and subsidiary relationships
03 Adviser, author and leadership identity
04 Locations and markets
05 Availability and eligibility
06 Claims, evidence and qualifications
07 Legacy names and discontinued Products
08 Canonical source and supersession
Evidence boundary

Clear entity architecture improves the source environment. It cannot guarantee that every external system interprets the relationship correctly — record the changed source condition and re-observe the representation.

Protect the relationship between fact and entity.

Explore AI Narrative Integrity, or return to the cornerstone framework.