A dental practice can run a thirty-page website and still take zero calls from search, while the clinic two streets away — with barely any website at all — collects enquiries from the map pack. Local results run on their own algorithm: backlinks and content volume matter far less than distance, profile completeness and reviews. Here is what Google actually weighs in local search, the order in which to fix it, and the numbers that tell you whether the work paid off.

How the map pack is actually assembled

Google runs local results on a separate algorithm and names three groups of signals itself. They are not equally weighted, and misreading that balance is what sends most local budgets to the wrong place.

• Relevance — how closely your category and description match the query. "Dental clinic" and "medical centre" pull completely different query sets, and switching between them is the cheapest fix you have.
• Distance — from the searcher to your address. Without a city name in the query Google uses the phone’s location, so one address realistically influences a radius of 2–5 km in dense urban areas.
• Prominence — review count and freshness, directory mentions, site traffic, calls from the profile. The only block that scales without physically relocating.

The business profile: what to fill in, and in what order

1. 1

   Verify ownership first

   An unverified profile ranks poorly and anyone can suggest edits to it. Verification by postcard or video takes three days to two weeks — start there, not with the description field.

2. 2

   Pick the narrowest possible primary category

   The primary category outweighs secondary ones by a wide margin. "Coffee shop" and "restaurant" mean different queries and different competitors. Add two to four secondary categories at most — beyond that you dilute relevance for no gain.

3. 3

   List services as individual items

   Not one block of prose — a list, each service with its own name, a price or range, and a short description. Google matches those names against queries like "teeth whitening near me", so wording beats formatting here.

4. 4

   Add real photos of the interior and the entrance

   A street-level photo of your door lifts call-through, because people can see they will find you. Fifteen images is a sensible floor, refreshed quarterly; a mediocre phone photo of the real place beats polished stock.

5. 5

   Keep opening hours synced with holidays

   Stale hours are the single most common cause of the "drove over, it was shut" review. Set special hours ahead of public holidays — that field is evaluated separately from the regular schedule.

NAP consistency and directory citations

NAP means name, address, phone. Google cross-checks them across every source that mentions your business, and mismatches erode trust in the profile. The classic case: "22 High Street, Suite 4" on the website, "High Street 22/4" in a directory, and a pre-move address still sitting on a social profile.

Local keyword sets and neighbourhood pages

The profile wins you the map pack. Organic results for geo-modified queries are won by the website — and that is where most local businesses simply stop. The harvesting logic is straightforward: take the service, multiply by geography, discard whatever nobody actually searches.

1. 1Write out services the way customers name them, not the way your price list does. "Teeth cleaning" is searched far more than "professional oral hygiene procedure".
2. 2Multiply each by three levels of geography: city, district, landmark ("near the central station"). Landmark queries are low volume but convert several times better than your site average.
3. 3Check the demand exists. If a "service + district" query returns zero impressions for months, it does not need its own page — a mention in body copy is enough.
4. 4Build a neighbourhood page only when you have something different to say: real directions, parking, separate hours, photos of that branch. Twenty pages with a swapped district name is a direct route to a thin-content problem.

Underneath sits LocalBusiness markup with address, coordinates and hours, plus a distinct URL per branch — we covered how to get that right in our guide to Schema.org structured data. The wider technical requirements are in the SEO checklist for 2026.

Reviews, and measuring whether any of it worked

Reviews are the only one of the three signals you can influence daily. What counts is pace and responsiveness, not the absolute total. A profile with 40 reviews where the newest is a year old reads worse than one with 18 and three from this month.

• Ask at the moment of satisfaction — a short link by text right after the visit yields three to five times more responses than a next-day email.
• Reply to every review, five-star ones included, within 48 hours. Replies are visible in the profile and shape the next customer’s decision.
• Answer negatives with facts and an offer to resolve it offline. Honest criticism cannot be deleted, and trying costs more than the review itself.
• Never buy reviews in batches. A spike of similar texts from new accounts triggers removal of the whole batch — usually taking genuine reviews with it.

Measuring local work by "rankings" barely works — two people at opposite ends of the same street see different results. Watch profile views by query, direction requests, calls from the listing, and on the site side, impressions and clicks on geo-modified queries. Reading those reports is covered in the Google Search Console guide.

Frequently asked questions about local SEO

In short

• Local results run on three signals: relevance, distance, prominence. Distance cannot be optimised, so set goals around the address you actually have.
• The primary category in your business profile outweighs every description field combined. Choose it as narrowly as possible.
• NAP mismatches across site, listings and directories quietly erode trust. Keep one canonical address format and fill every platform from it.
• Neighbourhood pages only work with genuinely unique content: directions, parking, photos, hours. Cloned pages damage the whole domain.
• Judge results by profile views, direction requests and calls — not by a "position" that differs for every searcher.