Local SEO Insights · The manifesto · No. 001

The game didn’t change.
It got more fierce.

A field manual for local search in 2026, written from our own citywide sweeps: 412 real searches across every quadrant of Winnipeg and Edmonton, plus a 146-trade scan of the packs that actually hold them. What Google ranks, where the map sits empty, and how a local business takes ground and holds it.

I.

The man comes home to a different corner

In the third season of The Wire, Dennis “Cutty” Wise walks out of prison after fourteen years and goes back to the only trade he knew. Nothing works the way it did. The crews are younger and colder, the product moves through phones, the old codes are gone. He says as much to Slim Charles, and Slim gives him the truest line the show ever produced: the game’s the same, it just got more fierce.

Every week we talk to a business owner who is Cutty. He learned his Google in 2016 or 2019. He had a guy who did the website, he got some reviews, the phone rang. Then he opens Google in 2026 and the street looks alien: an AI paragraph squatting where the links used to be, a map that behaves like the whole page, review counts that moved overnight, and some competitor he’s never heard of sitting in the three-pack like they own the deed.

Here is the thing the panic merchants won’t tell you, because calm doesn’t sell retainers. Google’s job has not changed by one degree. For a local query, Google is still answering exactly one question: of all the businesses near this person, which one most deserves the customer standing on that corner? That is the game. It was the game in 2016. The corners look different. The question is the same. It is simply asked and answered with far less mercy than it used to be.

This manual is about the fierce part: what the terrain looks like now, where it is defended, where it is inexplicably empty, and what a disciplined operator does about it. Everything quantitative in here is first-party: we ran live grid sweeps across every quadrant of two real cities, Winnipeg and Edmonton, 412 measured searches in all, and scanned 146 Edmonton trades to see what the winning packs are actually made of. Where we make a claim about how Google works, we cite Google. Nobody else gets a vote.

II.

The three levers Google will admit to

Google has published the answer key for years, and it is three words long: local results are ranked on relevance, distance, and prominence.[1] The entire local SEO industry is commentary on those three words. Most of it is bad commentary, because two of the three words are unglamorous and none of them can be gamed by buying a bigger dashboard.

Relevance is entity clarity, not keyword volume

Relevance means Google can confidently connect your business to what the searcher wants. Google’s own guidance is almost embarrassingly plain: complete and accurate business information, correct categories, real services, so that the system “can better understand your business”.[1] In 2026 the operative concept is the entity. Google does not rank a website; it ranks its understanding of a thing in the world. Your profile, your website, your reviews, and every mention of you across the web are being resolved into one identity. Where those signals agree, the entity is crisp and rankable. Where they disagree, a name spelled two ways, an old address alive on some directory, a category chosen in 2017, the entity is blurry, and blurry loses tie-breaks you never see.

Distance is the lever you cannot buy

Distance is exactly what it sounds like, and no one sells it, which is why you never hear about it in a sales call.[1] But it has a consequence almost everyone misses: because the searcher’s location is a ranking input, your rank is not a number, it is a surface. You are first at one intersection and seventh at another, eleven minutes away. Any tool, or any agency report, that hands you a single position for a keyword is averaging away the only geometry that matters. This is why serious measurement is done on a grid of fixed points across a city, and why we built our instruments that way.

Prominence is everything the web says while you aren’t in the room

Prominence, in Google’s words, covers how well known a business is, drawn from “across the web”, including links, articles, directories, review count and review score, all of it feeding local rank.[1] Reviews are the part you can influence fastest, and the part most owners run like an afterthought. We will get to the correct way to run them. The point to hold onto here is that prominence is an aggregate. It is built the way reputations are built, in small deposits, over time, and it is spent the same way.

Google published the answer key years ago. It is three words long. Everything the industry sells you is commentary.

III.

What actually got fiercer

The answer engine ate the question business

Google’s systems now answer a large share of informational questions directly on the results page, assembling responses from their index in place.[6] If your 2019 playbook was “blog about how to fix a garage door and the customers will come”, that traffic is being consumed upstream, before anyone scrolls. Mourn it briefly. Then notice what did not get eaten: the person who searches garage door repair near me at 9pm is not looking for an essay. They are standing on the corner with a broken spring. That query still resolves through the map and the listings under it, and those are still ranked by the three levers. The question business died. The buyer business is intact and, if anything, more concentrated.

The pack is the page

For money queries with local intent, the map results are no longer a feature above the real results. They are the results. A handful of pack positions absorb the taps, the calls, and the direction requests, and everything below them is consolation. This concentration is the fierce part in its purest form: the reward curve got steeper, so the marginal value of one position got enormous, and so did the cost of losing one.

Experience became a ranking language

Google’s quality guidance now leads with experience: content should demonstrate that it was made by someone who has actually done the thing.[5] Their content guidelines ask, in effect, would someone leave your page feeling they’d learned something from a person who knows the trade.[3] For a local business this is a gift with terrible wrapping. You cannot fake having re-piped four hundred basements, but if you have, you are sitting on ranking material your competitors literally cannot write: your jobs, your photos, your before-and-afters, your numbers. First-party proof is the one content category an answer engine cannot summarize away, because it does not exist anywhere else.

The referee reads the fine print now

The guidelines for representing your business are enforced with more appetite than they used to be: real name, real address, real categories, no keyword costumes in your business title.[2] The stuffed profile that survived 2019 is now a liability with a suspension attached. Fierce, again: the same rulebook, read seriously.

IV.

The asymmetry doctrine

Here is the intellectual core of this manual, and the part you can take to the bank. Because the pack is winner-take-most, effort in local search has convex payoff: moving from seventh to fourth is worth little, moving from fourth to third can double your phone. A rational operator therefore does not spread effort evenly. They hunt the spots where a small, well-aimed push crosses a threshold. We call these asymmetric positions: bounded downside, a few pages and a few weeks, against uncapped upside, a pack position that pays every day it holds.

The prevailing assumption is that there are no such spots left, that the big names own the map and everyone else fights for scraps. We tested that assumption instead of repeating it. We put a measurement grid over two entire cities, one point every couple of kilometres, and ran a real search at every point: garage door repair across 179 points of Winnipeg, junk removal across 233 points of Edmonton. Then we scanned 146 Edmonton trades to see how deep the winning packs actually run.

Exhibit A · 179-point grid · garage door repair · Winnipeg

Nobody owns the city

The most-reviewed name (927 reviews): share of #1s7 of 113 · 6%
Largest single operator’s share of #1s11 of 113 · 10%
Quadrants won by a business with under 50 reviews71 of 113 · 63%

First-party data: our live citywide sweep, 179 grid points, one real search each, 2026. 113 quadrants had a ranked winner; 27 different businesses hold a #1 somewhere in the city, and the largest of those footprints belongs to a shop with two reviews. The median ranked competitor carries 22.

Read that again, because it is the whole thesis in one table. The name every Winnipegger would guess owns garage doors holds six percent of the city. The other ninety-four percent is a patchwork of micro-monopolies: twenty-seven different businesses, each king of a few blocks, most holding their ground with fewer reviews than you collected last quarter. Attention piles onto the famous corners while whole districts sit essentially unclaimed. That is not a Winnipeg quirk. It is what the grid shows everywhere we point it: defended peaks, empty valleys, and a crowd convinced the valleys must be defended too, because why else would they be empty?

They are empty because everyone is measuring with the same broken instrument: one search, from one point, filtered through a brand-name assumption. Which brings us to the second finding.

Exhibit B · 233-point grid · junk removal · Edmonton

The pin lies. The brand name lies harder.

The assumptionWhat 233 measured points foundVerdict
The 2,989-review national franchise owns Edmonton#1 in 8 of 172 decided quadrants, about five percent of the cityMyth
“Our rank for junk removal” is one number38 different businesses are #1 somewhere; every one of them “ranks in Edmonton”It’s a surface
You need a giant review count to win anywhere31% of quadrant wins belong to businesses with under 50 reviewsOpen door

First-party data: our live citywide sweep, 2026, on a keyword with roughly 1,600 searches a month in Edmonton. A single rank check from one point would have told every one of those 38 businesses a different, equally flattering story. The grid is the only instrument that shows where you fade.

And if you want to know what an opening is worth before you spend a month taking it, the advertising market will tell you for free. Advertisers are many things, but they are not sentimental. The price they bid for one click is a public, continuously updated appraisal of what a customer is worth in that vein.

Exhibit C · 146-trade scan · Edmonton

What one click costs, next to who is actually guarding it

Search termBid per clickMedian reviews of the pack holding it
ac repair edmonton$47.85695
hvac contractor edmonton$44.13591
property maintenance edmonton$38.7821
garage door repair edmonton$38.08170
parking lot line painting edmonton$23.0116
carpenter edmonton$19.9420

First-party data: our 146-trade Edmonton scan, 2026. Read the columns against each other. AC repair costs $48 a click and the pack guarding it carries seven hundred reviews: a defended peak. Property maintenance costs nearly the same per click and the median incumbent holds 21 reviews: a $39 door held shut by a paperweight. Across the full scan, half of all trades are guarded by packs with about fifty reviews or fewer.

A brand name is a rumor. A grid read is a fact. Bid prices are a confession.

One more piece completes the doctrine. In an era when an answer engine can paraphrase every how-to guide ever written, generic knowledge has a market price of roughly zero. What cannot be paraphrased is a measurement that only you possess. Your grid reads, your review evidence, your before-and-after data: these are the only content assets with a moat, because they cannot be regenerated by anything that was not standing where you stood.[3] The asymmetry doctrine, compressed: fight where the field is thin, verify the field with your own eyes, price the prize off the ad market, and arm yourself with data nobody else owns.

V.

The playbook: rank, hold, be seen

Field orders. Seven of them, in the order a sane operator executes, each traceable to one of the three levers.

1. Claim the entity and complete it to the last field

Your Business Profile is your homepage now; the website is the supporting document. Every empty field is a question Google answers by guessing. Primary category chosen like it matters, because it does. Services enumerated. Hours true, including holidays. Photos that were actually taken at your business, recently.[1] And the name field carries your name, nothing else; the keyword-stuffed title is a 2019 costume that now draws suspensions.[2]

2. Make the web agree about who you are

Mark your site up with LocalBusiness structured data so machines read your identity instead of inferring it,[4] and enforce one canonical name, address, and phone across every directory that mentions you. This is entity hygiene: unglamorous, measurable, and the cheapest relevance you will ever buy. A knowledge graph is just a machine’s dossier on you. Hand it a clean one.

Exhibit D · The dossier

The entity graph Google resolves when it ranks you

IS THE SAME ENTITY ASPROMINENCEPROMINENCERELEVANCEEVIDENCECOMPLETENESSTHE SEARCHERlocation = distance, the unbuyable inputGOOGLE BUSINESS PROFILEcategories · services · hours · photosREVIEWSvelocity · specificity · repliesWEB MENTIONScitations · links · pressYOUR WEBSITELocalBusiness schema · NAP

Every arrow is a place signals can agree or contradict. Ranking work in 2026 is mostly making this graph consistent, complete, and independently corroborated. Distance sits at the top because it is the input you cannot edit.

3. Run reviews as an evidence program, not a vanity metric

Prominence counts reviews, but relevance reads them. A review that says “fixed our furnace on a Sunday in Riverdale” is testimony, with a service, a time, and a neighborhood in it. Fifty of those beat two hundred bare five-stars. So: ask everyone, steadily, so velocity never flatlines. Gate no one; soliciting only happy customers is the kind of trick the rulebook exists for.[2] Reply to everything, because replies are the cheapest public proof that someone is home. And read your competitors’ reviews the way we do professionally: every complaint about them is a positioning instruction addressed to you.

4. Measure like Google ranks: a grid, not a pin

Since distance is a ranking input, a single rank number is the average of a surface, and averages hide exactly the thing you need to see: the block where you fade. Read your keyword from fixed points across the city and you get a surface, top-three here, seventh there, invisible past the river. And never scout from your own logged-in browser. Personalized search is a mirror that flatters; it shows you your shop the way your mother describes you to her friends. Neutral snapshots from fixed coordinates or it didn’t happen.

5. Publish proof, not prose

The blog post that explains what a sump pump is: eaten, and good riddance. The page showing four hundred installs across your county with photos, dates, and outcomes: uneatable, because it is made of experience only you had.[5] Every job generates ranking material. Most businesses throw it away daily.

6. Hunt the thin fields

Apply Exhibit A to your own trade. Your market has defended peaks, usually the city head term, and empty valleys: neighborhoods, adjacent services, comparison queries, emergencies, seasons. Verify with your own eyes who holds each page, then take ground where the field is thin and the ad market says the prize is real. One well-made page for an unclaimed vein beats ten shouting into a defended one.

7. Hold: position decays socially

A ranking is not a possession, it is a lead in a race where everyone else is still running. It decays not by expiring but by being out-hustled. Holding is a cadence, not a heroic act: review velocity that never flatlines, fresh proof pages each season, profile updates that track reality, and a quarterly grid read with one standing order: find the weakest square and reinforce it before it flips. Cheaper to defend a corner than retake it. Ask anyone from Baltimore.

VI.

Why fierce is good news

Fierce cuts both ways. The volatility that terrifies owners is the same force that makes positions takeable at all. A placid market rewards whoever arrived in 2009 and froze. A fierce one re-litigates the rankings constantly, asking every week who deserves the corner, and pays whoever shows up with the best evidence. Meanwhile most of your competitors are still running the 2019 playbook, watching one flattering pin, stuffing keywords into their business name, and blogging essays into the answer engine’s mouth. Their confusion is your margin.

Cutty, for what it’s worth, never did go back to the corners. He opened a gym and taught the next generation how to fight properly, with discipline, footwork, and honest sparring. That is more or less our business model too. The game is the same as it ever was: be the most provably deserving business near the person searching. It just got more fierce, and fierce favors the prepared.

If you want to know where you actually stand before deciding where to push, that measurement is free: run a rank check. It reads the live results from fixed points around your business and shows you your surface, not your reflection.

VII.

Questions owners actually ask

Is local SEO dead in 2026?

No. The question business died; the buyer business consolidated. Transactional local intent still resolves through the map and the listings, ranked on relevance, distance, and prominence.[1] The work moved. It did not disappear.

What matters most: reviews, links, or proximity?

Wrong question, and the honest answer is a measurement. You cannot change proximity, so the real contest is entity clarity against review-and-mention prominence, and which one pays depends on where your grid is weak. Find the weak square first; the lever picks itself.

Do AI Overviews take my customers?

They take your readers, and readers were never reliably customers. A person with a burst pipe does not want an essay. Money queries still end in a call, directions, or a booking, and that surface remains ranked, contested, and winnable.

How long until something moves?

Entity corrections register in weeks. Prominence compounds over months, on the timetable of reputations everywhere. The honest way to know is a before-and-after grid read from the same fixed points, which is precisely the instrument we built.

Sources

Claims about how Google ranks are cited to Google. Market numbers are first-party: our 2026 citywide grid sweeps of Winnipeg (179 points) and Edmonton (233 points), one live search per point, plus a 146-trade Edmonton demand and review-depth scan.

  1. Google, “How to improve your local ranking on Google.” Google Business Profile Help. https://support.google.com/business/answer/7091
  2. Google, “Guidelines for representing your business on Google.” Google Business Profile Help. https://support.google.com/business/answer/3038177
  3. Google Search Central, “Creating helpful, reliable, people-first content.” https://developers.google.com/search/docs/fundamentals/creating-helpful-content
  4. Google Search Central, “Local business structured data.” https://developers.google.com/search/docs/appearance/structured-data/local-business
  5. Google, “Search Quality Evaluator Guidelines” (PDF). https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf
  6. Google, “How Search Works.” https://www.google.com/search/howsearchworks/
Run the free rank check