What you will achieve
Create a small approved record that helps reviewers identify the correct product in campaign evidence, draft content and sampled AI answers. You will define names, contextual aliases, distinguishing details and exclusions, then test the record against real material.
The record is an editorial aid. Keep it in your normal approved reference document or a suitable current product feature where available. This guide does not assume automatic entity matching, a new identity dashboard or a guarantee that external AI systems will use your preferred name.
Before you start
Choose one product, plan or range with an identifiable owner. Gather current authorised product references, relevant historical records, campaign videos and examples of names people actually use. If AI visibility is in scope, retain an available sampled answer with its question, provider and date.
Agree the purpose of the first record. A range-level overview and an exact variant check need different levels of detail. The product identity article explains why a familiar name does not always establish a reliable match.
1. Define the identity and its boundaries
Record the approved public name, owning brand and relationship to any range or parent product. Add the exact variant, generation, market and relevant dates where these affect the answer.
Write a short boundary statement: what this record covers and what it excludes. For a telecoms plan, it might cover the current UK offer while excluding an earlier bundle with the same public name. For a vehicle, it might identify a model while requiring a separate configuration record for trim-specific claims.
Attach the authoritative source that establishes the identity and name an owner who can resolve disputes. Keep the source's own date or version when available; do not invent a verification date because the template has an empty field.
Check: a reviewer can explain what the record identifies and where it stops being sufficient.
2. List aliases with their conditions
Collect shortened names, model codes and common references from approved materials and actual examples. Record whether each is an accepted alias, a broad range label or an ambiguous phrase needing context.
For every alias, add the condition that makes the match useful. A short plan name may need a market and offer period. A model code may identify a configuration more precisely than the public range name.
Also list exclusions: similar names, old versions or related products that should not automatically merge. Do not turn spelling resemblance into identity. The aim is a dependable review aid, not a family reunion for every product with the word Plus in its name.
Check: each alias has a reason and boundary, and ambiguous terms remain visibly ambiguous.
Fictional Connect Plus identity record
A name is not enough

This is a human review aid, not an automatic entity-matching configuration.
3. Add distinguishing facts and source references
Choose a few details that help distinguish this identity from its closest alternatives. These might be the generation, service type, configuration, market or authoritative destination URL. Use approved facts rather than campaign impressions.
Record the supporting reference for each detail and any conditions on its use. A feature shared by every product in the range may provide little identifying value. A current URL can be helpful, but its content still needs checking when used as evidence.
Use the following structure as an ordinary editorial table. The Connect Plus entries illustrate the decisions to make; they do not describe a live customer record.
| Field | Example treatment |
|---|---|
| Approved name | Connect Plus, with the approved market and offer context |
| Parent identity | The relevant brand and plan family |
| Accepted short form | Connect Plus when the surrounding context identifies this offer |
| Ambiguous usage | Connect Plus without enough detail to distinguish historical versions |
| Exclusion | Earlier bundle carrying the same name |
| Distinguishing reference | Current authorised plan page and applicable offer record |
| Owner and review trigger | Named offer owner; review when the plan or public name changes |
Check: the record provides useful discriminators without accumulating unsupported product claims.
4. Test the record against campaign evidence
Open the relevant ads in AdCyte and inspect the findings at their supporting moments. Check spoken names, on-screen text, depicted products and campaign context against the identity record.
Confirm clear matches and preserve uncertainty where the evidence is insufficient. A product can appear without enough detail to establish its exact variant. If several configurations occur in one film, map the relevant findings separately rather than assigning every claim to one convenient identity.
Record any discrepancy between the campaign and the approved reference. Resolve it with the source owner before using the identity in public copy. Update the record if the test reveals a missing legitimate alias or an important exclusion.
Check: each intended campaign match has a stated basis, and uncertain variant details remain unresolved until reviewed.
Constructed identity-review record
Preserve uncertainty in the review

No real AI response is depicted or attributed to a provider.
5. Review one sampled answer without losing the original
If visibility monitoring is part of the project, open an actual available sample and preserve its question, provider, date and response. Record other relevant check conditions where available. Use the product's current observation view, rather than assuming a particular label or layout.
Read the full relevant passage. Compare the name and distinguishing details with the identity record. Inspect cited sources where present, then record whether the answer appears to refer to the intended product, a different version, a mixture or an uncertain identity.
Assess important claims separately from identity. An answer can identify the right plan and give outdated inclusions. Keep both findings visible and avoid changing the original response to make the review easier.
Check: another reviewer can see the exact observation and understand the reason for the identity judgement.
6. Resolve disagreements and assign the next action
Have a second reviewer check difficult cases where practical. If the evidence remains insufficient, retain an uncertain outcome. Consistency comes from shared reasoning, not from forcing agreement on a guess.
Use the issue to choose a specific task: clarify a current page, label historical material, correct a draft or obtain a missing reference. Do not assume the cause of an external answer without inspecting the available sources.
| Review problem | Useful response |
|---|---|
| Correct brand, no product detail | Record a brand mention with product identity unconfirmed |
| Current name, historical feature | Investigate a possible version or fact mismatch |
| Two aliases map to different offers | Add conditions and exclusions to the record |
| Citation opens the right page but supports a narrower claim | Keep the identity finding and review the claim separately |
Check: each resolved issue has a reason, and each unresolved issue has an owner or explicit limitation.
7. Approve, share and maintain the record
Have the appropriate owner approve the record for its defined use. Place it where campaign reviewers and writers can access the current version. If the current AdCyte reference workflow supports the needed context, use it after checking the controls; otherwise retain the approved document in the established handoff.
Set review triggers for renamed products, new generations, changed offers or new markets. Keep meaningful historical distinctions when updating the current record. Replacing the latest name everywhere can erase the context needed to understand old campaign material.
Use the record in future briefs and sample reviews, while keeping external outcomes separate. Clear source identity supports a better workflow; it does not control how every AI provider interprets the brand.
Check: the current record, owner and review triggers are easy to find, and superseded context remains understandable.
Completion checklist
- The identity has an approved name, scope and authoritative reference.
- Aliases include conditions, and lookalikes or historical versions have exclusions.
- Distinguishing facts have sources and relevant qualifications.
- Campaign findings have been checked at supporting moments.
- Any sampled answer retains its original wording and review context.
- The approved record has an owner and change triggers.
Next step
Use AdCyte's AI visibility workflow to inspect available observations, or book a demo to discuss the process. If the review identifies misleading public information, continue with the correction investigation guide.
FAQs
Is this an automatic matching configuration?
No. It is an approved reference for human review and writing. Use native controls where their purpose and availability have been confirmed, without assuming the record automatically changes matching across the product or external systems.
Should one record cover an entire range?
It can describe the range, but variant-specific questions may need more precise records. Keep the relationship clear and do not apply a feature of one configuration to every product under the range name.
What should I do when identity remains uncertain?
Preserve the original evidence and record the uncertainty. Ask for the missing context or assign a source check. A reasoned uncertain outcome is more useful than an unsupported positive match.



