How to audit your business's entity consistency

Run the check before you buy the fix

INFERRED An answer composed from several sources reflects what those sources agree on. A business whose name, address, services and people are stated differently across the web gives a system nothing to agree on. This check finds the disagreements. It takes about an hour, needs no software, and is the same inventory any competent provider would build before charging for entity work, so doing it first tells you whether there is anything to pay for.

Step 1: Write down what you think is true

Before looking at any external source, open a document and record the facts as they are supposed to be: legal name, trading name, the exact address form (unit, street, suburb, state, postcode), phone format, founding year, a one-sentence description, the names and roles of key people, and the exact names of the services offered.

This baseline stops you accepting a widespread error as fact. A five-year-old typo that appears on three directories looks canonical unless you wrote the correct version down first.

Step 2: Check your own website

Look for every fact in every place it appears: homepage, footer, about and contact pages, service pages, the copyright line, and the JSON-LD structured data read separately from the visible text. DOCUMENTED Google’s guidance for its AI features is that structured data must match the visible text, so markup that says one thing while the page says another is a conflict the site has created for itself. Sites contradict themselves more often than owners expect, usually because a developer updated the homepage and not the global footer.

Step 3: Check your official records

DOCUMENTED For an Australian business the official records are ABN Lookup (the Australian Business Register) and, for companies, the ASIC register, plus any licensing or professional body the trade requires. Check the entity name, the registered business name, the address and the status. A registered address the business vacated years ago is a live contradiction in the highest-authority source there is. If the registered name and the trading name differ, both need to be stated correctly on the site (registered form as legal name, trading form as name) rather than used interchangeably.

Step 4: Check your profiles and directories

Start with the Google Business Profile, then Apple’s business listing for Maps, then the national and industry directories and any association listing. Then search:

• the business name in quotes plus the suburb;
• old versions of the name, if it has been through a rebrand;
• the phone number on its own.

Each query surfaces a different set of stale records. The phone-number search in particular finds listings created by a previous owner, an old agency or a directory scraper that nobody remembers.

Step 5: Record what you find

One row per source, one column per fact, and a notes column for which variant appeared. Do not fix anything during discovery. Recording first shows the pattern: a single wrong address copied across six directories is one root problem with one fix, not six.

The five common inconsistency patterns

1. Name variants. Legal name, registered business name and a shortened colloquial name used as if interchangeable; a Pty Ltd suffix on registries and dropped everywhere else.
2. Address drift. Old premises still listed; a unit number on the site and missing from the profile; “Street” and “St” mixed.
3. Unclear service naming. The same service under three names across the site and directories, so a system cannot tell whether the business offers one thing or three.
4. Missing or unlinked people. Practitioners listed on an about page with no stated role, no connection to the services they deliver, and no matching Person markup.
5. Stale facts and credentials. Founding year, team size and accreditations that were true in 2021 and are asserted unchanged in 2026.

Prioritise by authority, not by effort

INFERRED Fix the highest-authority sources first: the site, the ASIC and ABN records, the Google Business Profile. Many smaller directories copy from larger aggregators, so correcting upstream often propagates without further work. Correcting three major sources does more for agreement than editing thirty obscure ones, and bulk directory submission adds low-authority duplicates with slightly wrong details, each one a new conflict.

Fixing without creating new drift

Publish the canonical record internally: one page with the approved name, address form, description and service names, stored where anyone filling out a form can find it. Most future drift starts with a well-meaning staff member completing a directory from memory.

Attach a check to events. A rebrand, an office move or a new senior practitioner triggers a fresh pass. When correcting, change one variable at a time on the high-authority sources and note the date, so any odd movement in local rankings afterwards can be traced.

What this check does not cover

It finds disagreements between visible sources. It does not tell you whether the facts are modelled well enough for a system to identify the business: twenty sources can agree on a description so generic it fits three competitors. It does not catch structured data that validates cleanly while describing the wrong entity. And it does not measure whether any AI system currently resolves the business correctly; for that, ask an assistant in fresh sessions what the business does, where it is and who runs it, more than once, and record what comes back.

What is documented and what is inferred

• Documented: the Australian registry sources; Google’s instruction that structured data match the visible page and that Business Profile data be kept current.
• Inferred: that agreement between sources affects how confidently a system names a business; the authority ordering; propagation from aggregators.
• Not established: any measured relationship between a business’s consistency score and its appearance in AI answers for Perth prompts, pending the observation programme.