You can rank first from your office and twentieth from where your customers actually search. Both numbers are correct. That single fact explains more local "volatility" than every algorithm update of 2026 combined.
You don't have a local ranking
Start here, because it reframes the question.
Proximity is a ranking input in local search. Which means your position isn't a property of your business — it's a property of your business and the point the searcher is standing on. You have a different ranking for every location in your service area. Thousands of them.
A rank tracker reports one number. It samples from somewhere, and that somewhere is frequently your own address, because that's the sensible default. So the figure you're watching is one draw from a large distribution — and it can move for reasons that have nothing to do with your business changing.
The measurement problem in one line A single local ranking number isn't a summary of your visibility. It's a sample of size one, reported as though it were a fact.
This produces two failures at once. Teams panic over swings that are sampling noise. And teams feel safe on a number that flatters them — ranking first from the office while being invisible three miles away, where most of the demand is.
It's worth noting that some tracking tools compound this by scraping Maps rather than the local pack itself, which are different surfaces with different results. A comfortable Maps position can coexist with poor pack visibility for the same query.
Why three slots amplify everything
The structural reason local feels more turbulent than organic, and it's arithmetic rather than algorithmic.
The map pack has three positions. Organic has ten. A signal change that moves an organic result from position six to eight is barely perceptible — you were on page one, you're still on page one. The same magnitude of change moves a local business from third to fourth, which is the difference between being seen and not existing.
So identical underlying stability produces wildly different perceived stability. Local isn't necessarily moving more; the threshold it's moving across is far sharper.
The pack is also assembled differently. Organic ranks documents on content, links and authority. The pack is closer to a database lookup — profile data, reviews, and the searcher's location, cross-checked against the wider web. Those inputs change faster and more often than a webpage does, and they can be changed by parties other than you.
Mobile screen space compounds it further: local pack advertising has been reported occupying a meaningful share of the mobile results area, which compresses what's left for organic pack results on the device most local searches happen on.
Filtering isn't ranking
A mechanism that surprises people because it doesn't behave like a ranking factor at all.
If your listed hours show you closed at the moment someone searches, you can be removed from results entirely for queries with immediate intent — not demoted, removed. The logic is straightforward: sending someone to a closed business is unhelpful.
Which means a business with unusual hours, or seasonal closures, or hours that were never updated after a change, experiences apparent volatility that follows a daily and weekly rhythm. Check at 10am and you're visible; check at 7pm and you've "disappeared."
Two consequences. Accurate hours are a visibility asset rather than an administrative detail — including special hours for holidays. And falsifying them to stay visible is a documented suspension trigger, which trades a small visibility gain for a large risk, as covered in optimising your Business Profile safely.
GeoTargetly
Redirects website visitors to location-specific pages based on their geographic location
Best for: Geo Redirect & Targeting
What genuinely changed in 2026
Having separated the illusions, here's the real algorithmic movement.
| What shifted | Effect |
|---|---|
| Profile completeness | Became a more direct input; incomplete listings saw disproportionate drops |
| Review recency and response rate | Gained weight relative to raw review counts |
| Proximity in competitive categories | Reportedly tightened, shrinking the radius non-proximate businesses can rank within |
| Peak volatility window | Around 12–18 March, with a secondary wave near 22–24 March |
The characterisation that matters: this functioned as a recalibration of how much each signal contributes, not a penalty for particular tactics. Which is why so many businesses dropped without having done anything wrong — the weighting moved underneath them.
If your instability clusters in those March windows, that's the likely cause and there's no mystery to solve. Verticals with fewer competitors or well-managed profiles reportedly saw less disruption, which is itself informative.
An honest conflict in the reporting
Worth flagging rather than smoothing over, because you'll encounter both claims.
Some 2026 analysis reports proximity being tightened in competitive categories. Other analysis reports proximity losing weight relative to relevance and prominence, with businesses further away able to outrank closer competitors on the strength of engagement.
These are likely both true in different contexts — proximity dominating where competitor density is high and many equally-relevant options exist nearby, while prominence carries further in thinner markets. But nobody outside Google can confirm that, and anyone stating either version as settled fact is overreaching. Treat category-level and market-level variation as the default assumption rather than looking for one universal rule.
Volatility you cause and volatility caused around you
Two categories that get blamed on algorithms and shouldn't be.
Enforcement churn. Profile suspensions and reinstatements have run at elevated rates through 2026, including coordinated waves. When competitors in your area are suspended, you rise. When they're reinstated, you fall. Neither movement had anything to do with your own signals, and both look exactly like an algorithm update from inside your dashboard.
Competitor profile changes you can't see. A competitor changing their primary category, adding services, or accumulating a run of recent reviews can move you without any visible cause. The pack is a relative ranking of a small set — your position is as much about them as about you.
Your own edits. Reputation and ranking respond to changes you made, on a lag. A category change or a batch of profile edits three weeks ago is a plausible cause of movement today, and it's rarely the first thing anyone checks because it doesn't feel recent.
Going quiet. Reporting suggests profiles that go inactive for several weeks can lose visible pack position even with a stable review count. Absence is itself a signal.
What to actually do about it
Given that some volatility is illusory, some is structural and only some is actionable — here's how to allocate effort.
Fix your measurement first. Grid-based tracking across your service area, not a single point. Track the trend in your visible area over months rather than reacting to weekly position changes. This alone eliminates most false alarms.
Watch outcomes alongside positions. Calls, direction requests, enquiries. If those are stable while your tracked position swings, the position was noise. If those decline, you have a real problem regardless of what the tracker says. This is a version of the broader point about not over-trusting single metrics.
Complete the profile properly, since completeness became a more direct input. Hours, services, attributes, photos — the unglamorous fields.
Build steady review velocity with fast responses, given recency and response rate gained weight over raw counts. A consistent trickle beats a burst, and responses appear to matter independently.
Accept your realistic radius. If proximity tightened in your category, the honest response is to concentrate where you can genuinely rank rather than chasing visibility across an area you were never going to hold. That may mean location pages, or a second location, or narrowing the target.
Don't react to single weeks. The strongest practical discipline available. Most local movement reverses, and changes made in response to noise become the cause of the next problem.
The layer above the pack
One structural shift worth planning around separately.
AI-generated answers increasingly sit above or alongside local results, and reporting suggests selection into AI local answers follows entity confidence rather than proximity in the same way — meaning a business can be absent from an AI answer while ranking well in the pack, or vice versa.
The practical implication is that consistency and structured data matter for a surface that doesn't behave like the pack at all. Accurate structured data and location-specific pages feed that layer, which is one reason they keep appearing in local recommendations for reasons unconnected to traditional ranking — the wider picture being covered in how AI Overviews reshaped search.
It also adds a third thing to monitor. Pack position, organic position, and whether you appear when the question is asked conversationally — three surfaces that can move independently.
A diagnostic order
- Is this measurement? Run a grid scan. If your visible area is stable and only your tracked point moved, stop here.
- Is this filtering? Check hours, including special hours. Are you being excluded at certain times rather than ranked lower?
- Is this timing? Does the instability cluster in a known update window, such as mid-to-late March 2026?
- Is this you? What changed on the profile in the last month, including edits that felt minor?
- Is this them? Check whether competitors appeared, disappeared, changed categories, or gained reviews.
- Is this the outcome? Are calls and enquiries actually down, or only the number in the tracker?
Working in that order prevents the common failure: rebuilding a profile in response to sampling noise, thereby triggering the edit-related scrutiny described in the profile guide and creating a genuine problem where there was only a measurement one.
For the broader surrounding work — citations, local pages, on-page signals — the ground is covered in ranking in your service area, and there's a practical local SEO checklist for small businesses that steps through the fundamentals. For platform-level changes since this was written, our running notes on GBP updates track them.
The short version
You don't have a local ranking — you have one for every point a searcher could stand on, so a single tracked number is a sample reported as a fact, and much of the volatility people chase is their tracker sampling a distribution they didn't know existed. Three pack slots rather than ten make small changes look dramatic, and hours-based filtering can remove you entirely without any ranking change at all. Genuine movement did occur in 2026 — profile completeness became a more direct input and review recency overtook volume, with the heaviest local turbulence in mid-to-late March. Fix your measurement with grid tracking before diagnosing anything, watch calls alongside positions, and resist reacting to single weeks, since most movement reverses and the reaction frequently causes the next problem.
Local rankings jumping around and nobody can say why?
We measure local visibility properly across your service area and fix what's genuinely moving.
Explore SEO Services →Frequently asked questions
Why do local rankings change so much day to day?
Much of what looks like volatility is measurement rather than movement. A local business does not have one ranking — it has a different position for every point a searcher could be standing, because proximity is a ranking input. A tracker sampling from a single location reports one number from a distribution of thousands, so small shifts in where or how it samples appear as dramatic swings. Genuine algorithmic change exists too, but it is routinely blamed for variation that was always there and simply was not visible.
Why does the map pack fluctuate more than organic results?
Because it has three slots rather than ten, which makes visibility close to binary. A small change in signal strength that would move an organic result from position six to eight — barely noticeable — moves a local business from third to fourth, which is the difference between being seen and being invisible. The pack is also assembled from profile data, reviews and searcher location rather than from page content, so it responds to a different and faster-moving set of inputs than organic rankings do.
Can your business be removed from local results without ranking lower?
Yes, and this catches people out because it is filtering rather than ranking. If your listed hours show you closed at the moment someone searches, you can be excluded from results entirely for queries with immediate intent, on the reasoning that sending someone to a closed business is unhelpful. You have not dropped in position — you have been removed from that view. This makes accurate hours unusually consequential, and it also means falsifying them to stay visible is a known suspension trigger.
What changed for local search in the March 2026 core update?
Reporting suggests it functioned as a recalibration of how much each signal contributes rather than a penalty for particular tactics. Profile completeness became a more direct input, with incomplete listings seeing disproportionate drops, and review recency and owner response rate gained weight relative to raw review counts. Tracking tools recorded the heaviest local volatility around 12 to 18 March with a secondary wave near 22 to 24 March, so instability in those windows most likely traces to that update rather than to anything you did.
How should you track local rankings properly?
Use a grid of sample points across your actual service area rather than a single location, because one number cannot represent a position that varies with the searcher's location. Grid-based tracking shows where you are genuinely visible and where you are not, which converts an unreadable fluctuating figure into a map you can act on. Track the trend of your visible area over months rather than reacting to weekly movement, and check calls and enquiries alongside position, since those reflect the outcome the ranking exists to produce.
