Your rank tracker says you are third in the local pack. A customer two miles away searches the same phrase and does not see you at all. Both observations are correct, and the contradiction is the single most important thing to understand about local search: the pack is a different ranking system with a different unit of ranking, and a searcher-dependent input baked into its core.
Treating local rankings like organic rankings — one number per keyword — produces data that is not wrong so much as meaningless.
The three documented signals
Google's own documentation for local results names three factors: relevance, distance, and prominence.
- Relevance is how well a Business Profile matches what the searcher asked for.
- Distance is how far each candidate is from the location term in the query — or, absent one, from what Google knows about the searcher's location.
- Prominence is how well known the business is, which Google says draws on information from across the web — links, articles, directories — along with review count and review score. Google's documentation also states that position in web results is a factor, so organic strength feeds the pack rather than being separate from it.
Note what the ranked unit is: the business entity, the Business Profile, not a URL. Your website contributes signals, but it is not the thing being ranked.
Distance makes rank a property of the searcher
Because distance is a first-class input, there is no such thing as "our rank for emergency plumber." There is only your rank for that query from a particular point on the map. Every searcher stands at a different point, so every searcher gets a materially different pack.
How fast rank decays with distance varies by category density. In a downtown full of coffee shops, a few blocks can swap the entire pack; for a specialized practice with three competitors in the metro, the radius is far wider. The decay rate is itself competitive information: it tells you how far your effective market extends, and it is different for every query.
This is also why "near me" queries are not special. They make the location intent explicit, but Google applies location to local-intent queries regardless of whether the searcher types it.
What this does to rank tracking
A single tracked position is one sample from a surface that varies across the map — and worse, a sample from wherever your tool's queries appear to originate, which is typically not where your customers stand.
The workable approach is grid tracking: sampling the same query from a lattice of simulated coordinates across your service area, producing a map of positions rather than a number. Two honest caveats belong next to any grid report:
- Averages hide the shape. An "average rank of 5" can mean uniformly fifth everywhere, or first near the shop and invisible past two miles. Those are different businesses with different problems. Look at the surface, not the mean of it.
- It is still simulation. Grid tools ask from spoofed coordinates at a moment in time. They are a far better sample than one data-center query, but they remain a sample — treat small week-to-week wobbles at individual grid points as noise.
The metric that survives all this is share of visibility: across the grid, in what fraction of sampled locations do you appear in the pack at all? It is more stable than any single position and maps more directly to how many nearby searchers can actually find you.
Prominence is built mostly off your website
Because prominence aggregates reputation from across the web, the levers sit largely outside your site: review volume and rating on your profile, consistent citations in the directories that matter for your category, local press and links, and the organic strength of the site itself. This is the slow signal — it moves on the timescale of months of accumulated reviews and mentions, not on the timescale of a profile edit.
Reviews deserve particular discipline because they are the prominence input you can systematize: a consistent ask built into the post-service workflow compounds; a one-time push does not.
Relevance is the cheapest signal to fix
Relevance responds to profile completeness: the primary category (the strongest lever you directly control — the reasonable inference from widespread testing, though Google does not publish weightings), secondary categories, services, attributes, and description. Most profiles are simply incomplete here, which makes relevance the fastest audit win.
One contested area deserves naming: stuffing keywords into the business name is widely observed to correlate with better pack visibility, and it violates Google's profile guidelines, risking suspension. That is the decision frame — an observed edge purchased with policy risk on the asset your whole local presence depends on. Most businesses should decline.
What to do
- Replace single-point tracking with a grid for your handful of money queries, sized to your real service area, and report share of visibility across the grid rather than an average position.
- Audit the profile against the relevance inputs: correct primary category, complete secondary categories and services, accurate hours and attributes.
- Systematize the review ask so volume accrues continuously, and respond to what arrives — the goal is a durable prominence signal, not a burst.
- Keep doing organic SEO. Documented behavior says web-result position feeds pack ranking, so the local and organic programs are one program.
- Measure outcomes, not just rank: Business Profile insights on calls and direction requests tell you whether visibility is converting, and they are immune to the sampling problems that make local rank data slippery.
The pack rewards the same thing the map does: actually being the prominent, relevant option near the searcher. The data work is mostly about seeing that honestly instead of through one misleading number.