Of the 153,515 verified stores in ELDB, 44,132 carry a matched Google Business profile. Most people treat that as a curiosity — a way to get a phone number, maybe a city for a filter. It deserves more attention than that. For a large slice of ecommerce, the Google profile is the second storefront, and its condition is one of the fastest reads available on how the business is actually run.
The matched block carries placeId, name, category, rating, reviewsCount, address, city, zip, latitude, longitude, isClaimed and totalPhotos. Twelve fields. Each one answers a question you would otherwise burn half a discovery call asking, and several of them describe defects the merchant can verify in ten seconds and did not know about.
A matched profile is a claim about physical reality
A match means the business has an address somebody verified. That rules out most of the pure bedroom operations and tells you there is stock somewhere, probably staff, probably a lease, and a local customer base that the online catalog may or may not be serving. It changes the shape of every assumption you make about budget and decision-making.
It also means there is a second demand channel that most online-first operators ignore completely. A retailer with a shop in Lyon and a PrestaShop store is running two businesses that barely talk to each other: the local searcher never sees the catalog, and the online buyer never learns there is a place they could walk into. That gap is a service you can sell, and it is visible in the data before you make contact.
The 109,000-odd stores with no matched profile are not therefore bad leads. Many are deliberate online-only plays that never created a listing. That absence is its own segment — no local surface, no walk-in channel, nothing to audit locally — and it needs a completely different pitch. Do not send the local email to that group.
isClaimed is the most underrated boolean in the dataset
An unclaimed profile means Google generated or aggregated the listing and nobody at the business ever took control of it. The consequences are concrete: anyone can suggest an edit, hours can be wrong, the website link can be missing or pointing somewhere stale, and the owner has no ability to respond to reviews. It is a live, non-hypothetical problem with a small, demonstrable fix.
For anyone selling local visibility, reputation work, or general marketing to retailers, this is the single highest-intent filter in the whole database. The defect is real, the merchant usually does not know about it, and the demonstration takes one screenshot. Compare that to any pitch that starts by asking the prospect to imagine a problem.
The filter combination that produces a callable list: isClaimed false, reviewsCount above a floor so you know the location is real and getting attention, plus your country and platform constraints. That query typically returns a few hundred stores, which is a month of genuinely warm outbound rather than a spreadsheet you will never work through.
Read rating and review count as a pair, never separately
A 4.9 with seven reviews says almost nothing — it is three friends and a good week. A 4.3 with nine hundred reviews is a well-run business showing normal variance. A 3.6 with two hundred reviews is a business with a service problem it can feel in revenue but may not have quantified. The pair is a diagnosis; either number in isolation is noise.
Review volume also works as a rough proxy for operating history and footfall. Cross it against productCount and traffic and you can separate the shop that opened eighteen months ago from the family business that bolted a website onto a thirty-year operation. Those two merchants have opposite anxieties and respond to opposite framings.
The commercially interesting band is somewhere between 3.4 and 4.0 with meaningful volume. Ratings in that range suppress clicks in local results, and the merchant is losing walk-ins they will never see in any analytics dashboard. That is an urgent problem with an emotional charge, which makes it a conversation rather than a pitch — provided you open it carefully, which is a point I will come back to.
Category mismatch is a free audit you can run at scale
The profile category comes from a controlled vocabulary, and it largely determines which queries the listing is eligible to appear on. People who work on local search rank the primary category alongside review volume as the highest-leverage fields on the entire profile, and they consistently note that secondary categories are the thing competitors forget to fill in. That makes category error both common and cheap to fix.
So compare the category against what the catalog actually sells. A store whose inventory is technical running shoes, filed under a generic clothing category, is invisible on every query that matters to it. That is a two-line audit finding with an obvious remedy, and it is computable in bulk: join the category field against a catalog or keyword signal, flag the disagreements, and you have a segment of hundreds of stores each carrying a specific named defect.
Mismatch work has a second advantage. It is not a criticism of the merchant’s taste, effort or results. It is a configuration error, which is emotionally safe to point out and therefore easy to reply to.
Photos, geography and the abandoned-profile pattern
totalPhotos near zero on a claimed profile that has accumulated reviews is a specific signature: somebody claimed the listing once, uploaded nothing, and never came back. If reviewsCount is still growing, you have a business receiving local attention it is not converting. That is a tidy, self-contained offer with a visible before and after.
Latitude and longitude let you do things a city string cannot. Radius queries around your own office, if you pitch in person. Density clustering, to find the cities where forty unclaimed stores sit close enough together to work in a single week. And distance-based reasoning, which matters because proximity to the implied centre of a local search genuinely affects which listings surface — a well-known constraint among local operators, and one that tells you which prospects have a structural disadvantage they can only fix with better optimisation.
A practical territory build: draw a thirty-kilometre circle, pull every store inside it, sort by unclaimed first, then by rating below 4.2, then by review count descending. That is a week of local outreach where every single conversation has a concrete reason to exist.
Segments worth building, each with one offer
Five that earn their keep. Unclaimed with more than twenty reviews: the claim-and-optimise offer. Rating under 4.0 with more than a hundred reviews: the reputation offer. Claimed, fewer than five photos, more than fifty reviews: the abandoned-profile offer. Profile present but monthly traffic low: a local business with a weak web channel, which is an ecommerce build or rescue. Profile absent but traffic high: pure online, no local surface, and you should be selling them something else entirely.
The discipline is one segment, one offer, one artifact. The moment you send the same email to all five, the specificity that made the data valuable evaporates and you are back to volume. It is worth writing five short sequences rather than one long one.
Writing the email without sounding like a stalker
The rule that keeps this on the right side of the line: reference only what the merchant can verify in ten seconds and would agree is a problem. “Your Google profile is unclaimed” is verifiable and true. “I noticed your business and love what you are doing” is neither, and it is why most of these emails get deleted unread.
Lead with the artifact, not the argument. A screenshot of the profile with three defects circled, or a one-page audit, does the persuading. The email is a delivery vehicle with a low-friction question at the end. The merchants who reply are the ones who looked at the attachment first.
And do not open with a bad rating. Telling a stranger their reviews are poor, in the first sentence of an unsolicited email, is an insult wearing a business suit. Open with the mechanism instead — how local results decide what to show, which queries the listing is losing, what a competitor two streets away is doing differently. Same information, entirely different reception.
Where the matching fails, and what to do about it
Matching a domain to a physical place is inherently fuzzy. Chains and franchises produce multiple plausible matches. Shared addresses in business centres create false links. Agencies sometimes appear on a client’s profile. Businesses move and the listing lags. Assume a small error rate, and check the individual record before you send anything that depends on it being right.
Absence of a profile is also not proof of absence of a location. Some categories are not eligible, some merchants deliberately keep their address private, and some simply never got round to it — which, incidentally, is itself an offer.
Treat every claim you derive from this block as falsifiable by the recipient in a single click, because it is. That constraint is not a weakness of the data. It is exactly why the approach works when you get it right: you are saying something true and checkable to somebody who is used to receiving neither.