Here is a pattern we have now seen enough times to call it a rule.

Ask an AI assistant about a specific location by name and it answers well. It knows the address, it knows the hours, it often cites the company's own website directly. Ask the same assistant a discovery question, something like "best options near me" with no brand name in it, and that same company does not appear at all.

We confirmed this across a six-property shopping centre portfolio. Branded queries returned the client's own site as a cited source. Discovery queries returned none of the six properties, routing instead through Wikipedia and mall aggregator directories that none of them had established a presence on.

Both things were true at once. Findable if you already knew the name, invisible if you did not.

That gap is where multi-location businesses lose customers, and here is what causes it.

1. Your locations exist as a list, not as places

Many multi-location sites handle locations with a single page carrying a store locator, or with location pages that are the same 200 words with the city swapped.

Neither gives an answer engine anything to work with. A location needs to read as a distinct place with its own facts, which means its own address, hours, services offered there, staff, parking, transit access, and the questions people actually ask about that specific site.

Ten near-identical pages do not make ten entities. They make one entity mentioned ten times.

2. You have optimized for people who already know your name

Branded search is comfortable. It converts well, it looks good in reports, and it is the easiest thing to rank for.

It is also the half of the funnel AI left alone. Discovery is the half that changed, and discovery questions get answered from sources that compare and rank options rather than from any single company's website.

If every page you have written assumes the reader already knows who you are, you have built for the queries you were already winning.

3. Your location data lives in three systems and agrees in none

At scale, this is almost universal. Head office maintains a spreadsheet, each location updates its own Google Business Profile, and the website pulls from a CMS that was last reconciled during a redesign.

Holiday hours are where it shows first. One source says open, another says closed, a directory says nothing. Answer engines cross-check across sources, and inconsistency lowers confidence in all of them, including yours.

One canonical source of location truth, pushed everywhere else automatically, is the fix. It is a systems problem more than a marketing problem, which is exactly why it stays broken for years.

4. Each location is a stranger in its own city

A national brand's authority does not automatically transfer to a specific address. For local discovery questions, answer engines lean on locally established sources, which means regional directories, local news, community organizations, chambers of commerce, and the aggregator sites that dominate your category.

In the portfolio we measured, discovery answers routed through Wikipedia and mall directories. The properties themselves had never established a presence in either.

Pick the three or four sources that keep appearing in AI answers for your category and city, then earn a real listing on each one, per location. This is slow work and it is the part competitors rarely bother to do.

5. Location pages answer navigation questions and nothing else

Address, hours, map, phone number. Then nothing.

Customers ask far more than that. Do you have parking. Is it accessible. Do you take walk-ins. What time is quietest. Do you have someone who speaks Punjabi. Is there a family washroom.

Every one of those is a question an assistant would love to answer and currently cannot, because the answer is not written anywhere. Sitting on your reception phone log for twenty minutes will hand you the twenty questions your location pages should already answer.

6. Nobody measures a single location

Portfolio reporting hides everything that matters. A total can look stable while individual sites move in opposite directions.

Across our six properties, organic search rose 0.6% while total sessions across the portfolio fell 1.2%. Organic was the only major channel to grow, and its share of portfolio traffic rose from 71.5% to 72.8%. Underneath that flat-looking total, individual properties ranged from 10.6% organic growth to outright declines caused by a rebrand and a paid-search wind-down.

One property makes the argument on its own. It lost 100% of its Google paid search traffic, down to zero sessions, and total traffic still grew 1.4%. Organic absorbed the entire loss.

That is the case for organic and AI visibility in one line. It is insurance against media cuts, not just a growth channel. And you only see it if you measure per location.

Where to start if you run more than five locations

Do not start with content. Start with a single discovery question, the one a customer would actually ask with no brand name in it, and run it against ChatGPT, Perplexity, and Copilot for each of your markets.

Write down which businesses get recommended and which sources the assistants cite. That list is your real competitive set and your real link-building plan, and it usually looks nothing like the one in your current SEO deck.

Frequently Asked Questions

**Does multi-location SEO still matter if we rank well on Google?**

Yes, and the two can diverge sharply. Ranking well for branded and navigational searches is compatible with being completely absent from AI discovery answers, because those answers are assembled from comparative and third-party sources rather than from ranked results.

**Should each location have its own page?**

Each location should have its own page with genuinely distinct content, not a template with the city name substituted. Answer engines need enough unique, factual detail to treat the location as a specific place rather than a repeated mention of the parent brand.

**Why do AI assistants cite directories instead of our own website?**

Because directories present comparable, structured information across many businesses, which is exactly what a discovery question requires. Being present and accurate on the directories that keep appearing in answers is part of the work, alongside improving your own pages.

**How many locations before this becomes a real problem?**

The data inconsistency problem starts around three to five locations, when no single person can hold every listing in their head any more. The discovery visibility problem exists at any size.

Want to see where your locations stand?

We run a free AI Visibility Check on the discovery questions your customers ask, market by market, and show you who the assistants recommend instead of you.

Start at [emdigital.ca/ai-visibility-check].