Devstars
Blog
Date: 29/04/2026
Stuart WatkinsShould national and international businesses use Google Business Profile in 2026? Yes. GBP has shifted from a local directory to one of the most important entity signals feeding AI search engines. 86% of AI citations now come from brand-managed sources, with listings accounting for 42% of those citations (Yext, 2025). National brands without optimised GBP profiles are absent from a significant portion of the AI answer layer that ChatGPT, Gemini, Perplexity, Apple Intelligence and Copilot now serve to billions of users monthly.
If you run a national business, you’ve probably been told Google Business Profile is a local thing. Plumbers. Hair salons. The pizza place at the end of the road. Not for you.
The reality is, that advice is about three years out of date.
We’ve been digging into this for a couple of clients recently, both national B2B brands operating across multiple countries. What we’re seeing is that GBP has quietly become one of the most important entity signals feeding the AI search layer. Not just Google’s local pack. ChatGPT, Gemini, Perplexity, Apple Intelligence, Copilot. All of them.
Yext ran the numbers last October across 6.8 million AI citations. 86% came from brand-managed sources. 42% specifically from listings. So if you’re a national business that’s ignored GBP because “we don’t have a shop”, you’re effectively invisible in roughly 42% of the citations AI is now using to answer questions about your category.
Pretty stark really. Let me explain what’s changed and what to actually do about it.
The numbers tell the story:
When someone asks ChatGPT “best access equipment supplier for warehouses” or “experiential marketing agencies in London”, the engine grounds its answer against listings data, LinkedIn, review platforms, Wikipedia, and the brand’s own site. GBP is feeding the listings layer.
No GBP, no entity anchor, no consistent NAP (name, address, phone), and the AI either picks your competitor or makes something up.
The mental model that helps: GBP is no longer a fancy phone book entry. It’s a structured-data feed.
Until about 18 months ago, Google used GBP to populate the local pack and Maps. That was it.
What’s happened since is that GBP data has been pulled into Google’s Knowledge Graph, which then feeds Gemini, AI Overviews and AI Mode. It’s the same foundation we build on in our digital marketing GEO work — entity consistency is what determines whether AI engines mention you or ignore you. Here’s how the major engines compare:
| AI engine | GBP data accuracy | Why it matters |
|---|---|---|
| Google Gemini | 100% | Grounds directly against Google Maps |
| Google AI Mode | 100% | Same direct ingest as Gemini |
| Google AI Overviews | 100% | Pulls from Knowledge Graph |
| Microsoft Copilot | High (via Bing Places) | Sources Bing Places + Bing index |
| ChatGPT | 68% | Indirect — Bing index + listings aggregators |
| Perplexity | Mixed | Reddit-heavy plus listings data |
| Apple Intelligence | High (via Apple Business) | Apple Business is the primary signal |
The gap between Gemini’s 100% and ChatGPT’s 68% is the gap between being cited and being misquoted.
The thing is, this works in your favour if you set it up properly. A complete, consistent GBP becomes a verification anchor. Every other AI engine cross-references it. That’s why brand-managed listings are now responsible for 42% of citations across the board.
Several important policy shifts national businesses need to know about:
Service area limits tightened. Google banned countries and whole states as service areas in June 2025. The maximum is 20 service areas per profile, all specified as cities or postal areas. “United Kingdom” or “California” are no longer valid. This change was specifically aimed at agencies and B2B firms claiming the world.
Reviews policy expanded April 2026. No more review gating, no shared tablets, no incentivised reviews including loyalty points. Google removed 292 million policy-violating reviews in 2025, roughly 22% of all reviews submitted globally. Gemini now scans suggested edits before they go live.
Apple Business Connect became Apple Business on 14 April 2026. It’s now a single platform across 200+ countries. With iPhone share at 58.1% in the US, 60.5% in Canada, and 52.3% in the UK, this isn’t optional anymore.
Bing Places relaunched in October 2025 at bing.com/forbusiness. Bing’s organic traffic isn’t huge, but it’s the grounding layer for Microsoft Copilot and feeds ChatGPT’s live retrieval.
Most national businesses do one of three things, all wrong:
The correct structure: one GBP per country, all managed under a single Location Group.
If you’ve got real depots or offices in multiple cities within a country, and each one is genuinely staffed with non-overlapping service areas, you can run a profile per location. If they share a service territory, you can’t.
Sterling Sky’s Joy Hawkins, who’s the closest thing local SEO has to a referee, has been clear on this: “If you have different locations for your service business, with separate service areas and separate staff at each location, you’re allowed one profile for each location.” Anything else and you’ll get suspended.
Here’s the ranked influence of each GBP field on AI citation likelihood:
| Field | Influence | Why it matters |
|---|---|---|
| Primary category | High | Single biggest ranking lever; wrong choice eliminates you from the candidate pool |
| NAP consistency + schema | High | Ambiguous entities get filtered out by AI engines |
| Review velocity (volume × recency) | High | Reviews now weight ~20% of Local Pack ranking |
| Review response rate ≥ 80% | High | Whitespark 2026 confirms measurable ranking lift |
| GBP completeness | High | Complete profiles are 2.7× more likely to be considered reputable (Google) |
| sameAs schema (Wikipedia, Wikidata, LinkedIn) | High | Brands with Wikipedia get cited in ChatGPT in 28 days vs 52 days without (Status Labs) |
| Photos (recency + volume) | Medium | +42% direction requests, +35% website clicks vs none |
| Posts/Updates cadence | Medium | Zero direct ranking impact, but feed AI summaries |
| Products/Services tabs | Medium | Maps AI summaries pull product entries |
| Booking/Order attributes | Medium | Drive transactional AIOs (69% CTR) |
| Additional categories (4+) | Medium | BrightLocal: 4 additional categories = best avg map rank |
| GBP attributes | Low–Medium | Filter “best X for Y” prompts |
The top five fields above account for most of the AI citation impact. Get those right before optimising the rest.
Here’s the honest bit. AI search visibility is wildly volatile.
AirOps tracked it across thousands of brands and found only 30% stay visible from one answer to the next, and only 20% across five consecutive runs of the same query. AIO content changes about 70% of the time for the same query.
So if you’re optimising for AI citation, you can’t just check once and call it done. You need to run prompts at least 10 times per platform to get a meaningful baseline.
Tools that handle this well:
We’re using a combination of Local Falcon plus Otterly.ai for AI prompt tracking. Not perfect. But you can iteratively improve it over time.
Quick example. We’ve got a B2B equipment client operating in five countries. Started with one UK profile that had a few inconsistencies. Three different addresses across their own assets. Wrong primary category. Service area listed as “United Kingdom” which is now non-compliant. Photos hadn’t been touched in 18 months.
The remediation looks something like this:
sameAs clusterIt’s a fair bit of work. But the foundation matters. You can’t optimise for AI search if your basic entity data doesn’t agree with itself. This is the structural layer our technical GEO specialists handle — entity data, schema architecture, and the sameAs cluster that ties it all together.
If you’re a national or international business and your GBP strategy is “we don’t really do local stuff”, you’ve misunderstood what GBP is for in 2026. It’s not a directory entry anymore. It’s the verified entity record that AI search engines cross-reference when deciding whether to mention your brand.
Get it wrong, you’re invisible. Get it right, you’re the answer.
The thing is, the brands who get this sorted in 2026 will compound the visibility for years. The ones who don’t, won’t.
Stuart Watkins is the founder of Devstars and London Web Design Agency, where we help national and international B2B brands navigate the shift from search to AI search. We’ve been building digital infrastructure since 2003, with a particular focus on the entity signals that AI engines now use to decide who’s worth citing.
No, not in 2026. Google banned countries and whole states as service areas in June 2025. Each GBP must specify cities or postal areas as its service area, with a maximum of 20 areas per profile. National businesses need separate profiles per legitimate physical location, all managed under a single Location Group.
Yes, if they have face-to-face customer contact. B2B firms that visit clients on site (deliveries, demos, training) qualify as service-area businesses with hidden addresses. Pure online-only businesses with no in-person customer contact are ineligible for GBP under Google’s 2026 guidelines.
GBP affects ChatGPT indirectly. ChatGPT grounds against Bing’s index plus listings aggregators, so a complete Bing Places profile (which can be imported from GBP in one click) is essential. Without consistent GBP and Bing Places data, ChatGPT’s accuracy on business contact information drops to 68% versus 100% for Gemini.
Primary category is the single biggest lever. Choose it wrong and the AI eliminates you from the candidate pool entirely. Use tools like PlePer or GMBspy to see what your top three competitors are using before deciding.
60 to 90 days for measurable lift in most cases. Brands with Wikipedia presence get cited in ChatGPT in around 28 days; those without take roughly 52 days (Status Labs research). The volatility means you’ll see fluctuation, so always measure across at least 10 sessions per platform per prompt.
Yes, both. Apple Business covers 200+ countries and feeds Apple Intelligence and the Gemini-powered Siri rolling out in iOS 26.4 (Spring 2026) and iOS 27 (September 2026). Bing Places feeds Microsoft Copilot and ChatGPT’s live retrieval. Together with GBP, these three platforms cover the vast majority of the AI search layer.
A storefront GBP has a publicly visible address customers can visit. A service-area business (SAB) has the address hidden because customers don’t visit; the business goes to them. National B2B brands typically need SAB configuration with city/postcode-level service areas. Failure to hide a residential or virtual address now triggers near-certain suspension.
Tell me what you’re trying to fix. Half an hour, no pitch, no slide deck.
If we’re the right fit we’ll talk about what’s next. If we’re not, I’ll point you to someone who is.
