Module 05 · Lesson 1 of 4
Entity Consistency Across the Web
An entity is a distinct thing a search or AI system can identify and relate to other things: your company, your product, your founder, your city. Machines do not infer identity from vibes. They match names, facts and relationships across sources, and they trust what agrees.
Consistency is the signal
Imagine your company appears as "Northside Plumbing Ltd" on its site, "Northside Plumbing" on Google's business profile, "NorthSide Plumbing & Heating" on a directory, and "northside-plumbing.co.uk" as the only identifier on a forum. A human reads four spellings of one firm. A system weighs the possibility of one, two or four related entities, and association weakens with every mismatch. The same applies to facts: founding dates, service areas, phone numbers, founder names, pricing tiers. Where independent sources agree with your own site, confidence rises. Where they conflict, it falls.
Write the canonical record first
Before auditing anything, write your canonical record: one document holding the exact legal and trading name, the one-sentence description you use everywhere, founder and key people names, locations, contact facts, and your key entity relationships (brand to product to parent company). Every profile, footer and bio is then filled from this document, not from memory. This sounds bureaucratic. It is the difference between one clear entity and four vague ones.
| Surface | What must match | Common failure |
|---|---|---|
| Your site (about, footer, contact) | Name, description, people, facts | Marketing tagline instead of plain identity |
| Business profiles and maps | Name, address, phone, hours, category | Old address live on one profile |
| Social accounts | Handle, display name, bio link | Three different one-line descriptions |
| Directories and listings | Canonical name and URL | Abbreviations and tracking URLs |
| Author pages and bylines | Person name, role, same bio | Byline names that match nobody's page |
- Collect every surface you control or claim: your pages, profiles, listings, old domains.
- For each, record the exact name string, the URL, and the description used.
- Compare each row against your canonical record. Mark match, minor mismatch or contradiction.
- Fix contradictions first (they split the entity), then minor mismatches, then cosmetic differences.
- Note any surface you cannot fix (a directory that ignores requests): document it, move on, and let your own consistency carry the weight.
- Write your canonical record document. Keep it short enough to actually use.
- Run the audit across at least eight surfaces, including your site, one business profile, two social accounts and two directories.
- Fix the two worst contradictions you find, and put a date in your log for the rest.
- Add a line to your quarterly calendar: re-run the audit. Entities drift as profiles get edited.
Why does a misspelled directory listing matter if nobody visits it?
What is a canonical record?
Should you fix cosmetic differences or contradictions first?
Sources used in this lesson
Google Search Central: structured data
Google Search Central: helpful, reliable, people-first content