Short answer: a useful dashboard does not need an "entity salience score". It can track ten concrete indicators about identity, consistency and ownership: first-party conflicts, external profiles, owner coverage, naming stability, duplicate locations and factual accuracy. Google documents LocalBusiness and Organization structured data, but does not publish a universal saliency score for local businesses.

1. First-party conflict rates

Numerator: first-party attributes with material conflict. Denominator: audited eligible attributes.

Separate address, program, phone, services and canonical. Do not include `not applicable' in the denominator.

2. Critical-profile consistency

Preselect important external profiles. Numerator: profiles without material conflict. Denominator: eligible priority profiles.

Do not add directories after seeing the result.

3. Owner coverage

Numerator: critical attributes with owner. Denominator: total critical attributes.

Data can be correct today and stale tomorrow if no one is responsible for it.

4. Naming stability

It measures canonical name usage versus legacy aliases in first-party, external profiles, and monitored outputs.

After the rebrand, this metric may be more useful than an aggregate score.

5. Location duplication count

Count pages that describe the same location or service area without a distinct role. Includes redirects and accidentally active history pages.

Local duplication increases the maintenance cost and can produce contradictions.

6. Service-location consistency

For each location, check that the list of services corresponds to reality. A service only available in premises A should not be automatically copied to B.

The denominator is the eligible location-service pairs.

7. Structured-data alignment

Measure pages where LocalBusiness/Organization markup reflects visible content. A syntactically valid markup is not sufficient if it describes another address or name.

8. Time-to-resolution

It measures the time from finding to rechecking the corrected surface. Ticket created' is notresolved'.

For external sources, keep `external unresolved' when you have no control.

9. Factual accuracy in query set

For program, address, service and coverage area, classify correct',incomplete', wrong',unverifiable'.

This is an external series and should not be mixed directly with first-party health.

10. Change-event coverage

It measures how many relevant operational events triggered a complete update: relocation, seasonal schedule, phone change, service launch, rebrand.

This metric shows whether the process prevents drift.

Baselines

On first run it saves the active locations, attributes, profiles, owners and query set. Version the snapshot.

If population changes, report new locations and closed locations separately.

The denominators

Each indicator has its own denominator. Conflict rate uses eligible attributes. Profile consistency uses priority profiles. Service-location consistency uses eligible pairs. Factual accuracy uses verifiable claims.

Don't compress everything into a single percentage.

Observation window

First-party health can be checked immediately. External profiles and outputs may take time. Change-event coverage is evaluated over a longer period.

Use windows appropriate for each layer.

False-attribution risks

If traffic, calls or leads increase after an identity cleanup project, don't automatically attribute the result to the dashboard. Seasonality, reviews, campaigns, PR and competition can influence outcomes.

Keep business metrics separate.

How do you segment

For many locations, segment by type of service, region and operational complexity. Do not hide P0 conflicts in a good global average.

Displays the list of material findings along with the percentages.

How do you set thresholds

Bad address or program can have zero tolerance. Cosmetic variations may have low priority. Thresholds must be related to the impact on the user's decision.

What a useful finding looks like

Location B, first-party conflict program: location page M-S 09:00-13:00, footer M-F; severity P0; owner Operations; global component update action; recheck after deploy.

This finding is more useful than "entity score 72".

When the dashboard is enough

Stop scaling when existing indicators cover operational decisions. If a new chart doesn't change any action, don't add it.

How to check the quality of data from the dashboard

A dashboard can look flawless and still be powered by stale sources. For each indicator, it stores the date of the last collection and the number of excluded records. If a crawl failed or a platform didn't respond, mark `not verified' instead of reusing the last value as if it were current.

How to avoid the hidden score

If management asks for a summary, use material states and findings, not an average that combines address, schedule, and external profiles. One wrong address P0 should not be compensated by ten solved cosmetic P3s.

Re-audit threshold

Resume full audit after relocation, rebrand or major service expansion. Between these events, use incremental monitoring and watch for changes that may cause drift.

How do you handle missing data?

If an external source does not respond or a crawl fails, do not report the last value as if it were current. Use NOT_VERIFIED and remove the observation from the denominator if the methodology requires it. This rule prevents the dashboard from appearing more stable than it is.

How to check the robustness of indicators

Once a quarter, it takes a sample and manually reconstructs the values ​​from the raw sources. If the dashboard and independent verification differ, investigate the pipeline before interpreting the trends.

Claim ledger

  • FACT/EVIDENCE: Google documents LocalBusiness and Organization structured data.
  • PRACTITIONER GUIDANCE: entity health can be measured by conflicts, ownership, consistency and time-to-resolution.
  • INFERENCE: processes based on change events can reduce drift.
  • NOT PROVEN: a universal entity salience score or a guaranteed direct relationship with ranking/citations.

Conclusion

A good local service dashboard shows what information is wrong, who owns it, and how long it takes to fix it. If the indicators do not lead to action, they do not measure entity salience, but only produce analytical decorum.

Sources reviewed

Related insights