The proxy is reasonable, which is why it survives
It comes up in most first calls, usually as a gotcha. If you're this good at SEO, why aren't you number one for "SEO agency in Mumbai"?
The logic looks sound. SEO is one of the few services where the supplier can demonstrate the product on itself — a dentist can't show you their own root canal. It also has the two properties buyers want in a test: it costs nothing to run and it needs no domain knowledge.
The conclusion is close to worthless, and the reason is mechanical rather than defensive. The query you just typed isn't really an SEO query. It's a local business query, and Google resolves it with a separate system whose inputs have very little to do with whether an agency can grow your revenue.
What SEO company in Mumbai is actually a query for
Type it and look at what sits above the fold. Two or three ads, then a map with three business listings, and only then the ordinary blue links. That map block is the local pack, and it's the part most buyers unconsciously read as "the ranking".
Google's guidance for that block names three inputs: relevance, distance and prominence. Relevance is how well the profile matches the search. Distance is how far the business sits from the person searching. Prominence is how well known it is — and Google says plainly that more reviews and positive ratings can help your local ranking.
Two of those three are facts about real estate and reputation admin. Neither is evidence that an agency can fix your faceted navigation or earn a link from a publication your customers read.
There's a gate in front of all of it. A profile needs a physical location customers can visit, or a business that travels to customers. A rented mailing address the business doesn't operate from — a virtual office — isn't eligible, and nor is a PO box. Then it has to be verified, by video call, postcard, phone or email.
So the agency at the top of your city's map pack has premises there, a verified profile, a steady flow of reviews, and the luck of sitting near you when you searched. Those are real achievements. They are front-office achievements.
The blue links underneath are a different game, and still not yours
Below the map, the organic results for "SEO agency in [city]" are a genuinely hard-fought commercial term. Whoever holds them has spent years aiming content, links and digital PR at exactly one audience: people shopping for an agency.
That is real skill at precisely one thing — ranking a lead-generation page for people who buy SEO. It is not the same problem as ranking a D2C category page against three marketplaces, or a SaaS comparison page against a competitor with eight years of domain history.
There's an incentive underneath it nobody says out loud. Winning your own acquisition term competes directly with client delivery for the same senior people. An agency with spare senior capacity can point it inward; an agency whose seniors are all on client accounts can't. Neither fact tells you which one delivers, and the first is as consistent with a large sales operation feeding churn as it is with a good delivery team.
Some of what looks like ranking on that query isn't ranking at all. Ads sit above it, and several of the "top 10 agencies in India" pages occupying the rest are sold by the slot rather than earned.
Where the test does carry information
It isn't worthless in every direction. The test is weak as a positive signal and meaningfully stronger as a negative one — with one caveat you should let them use.
Their own site, checked by you rather than described by them
Their site is the one property where their hands are the only hands. Ten minutes, no tools.
- Open three of their blog posts on your phone. If the largest element takes five seconds on a good connection, that's a fact about their standards, not their priorities.
- Check the title tags across a few pages. Forty pages sharing one title is a five-minute fix nobody has made.
- Look for the ordinary mistakes: a
noindexleft on a live template, a canonical pointing at the homepage, a staging subdomain still in the index.
The specialism claim, checked against the right SERP
If they pitch you as ecommerce or SaaS specialists, don't check "SEO agency in Pune". Check the queries somebody in that category actually types — the operational questions, the tooling comparisons, the pricing terms.
If they surface for nothing in your category, that tells you more than any city term, because it's the exact skill you're buying.
The caveat, which is real
The cobbler's-children defence is legitimate. Test it by asking when they last shipped a change to their own site, and what it was.
"Eight months ago, a redirect map after we killed the old services pages" is a fine answer. "We've been meaning to get to it" is a different answer, and you now know how they treat a backlog.
Six tests, ordered by how hard they are to fake
The useful way to rank a diligence test is by what it costs the agency to produce a false pass. Anything cheap to fake carries no information, however good it looks in a deck.
The line that matters sits between rows three and four. Everything above it is produced by the agency. Everything below it is produced by somebody else and merely shown to you. A structured vetting call is built almost entirely out of the bottom two rows.
| Test | What it actually proves | Cost to fake |
|---|---|---|
| Their position for SEO agency in [your city] | Premises there, a verified profile, review volume | Low — a serviced office and a few dozen reviews |
| Awards, badges and partner logos | That somebody filled in an entry form, often paying for it | Low |
| Case studies with percentage uplifts | That one account did well once, from a baseline you can't see | Low to medium |
| A named client who left, and why | That they can describe their own failure in checkable detail | Medium — three follow-ups break an invented story |
| The CV and client load of the person doing the work | Who is on your account and how thinly they're spread | High — a staffing fact, and half of it is public |
| Read-only Search Console on a live client, screen-shared | Query, click and impression data they didn't produce | Very high — the numbers are Google's, not theirs |
What this domain ranks for, and what it doesn't
We'd be running the same trick as everyone else if we made that argument and then declined to apply it to ourselves.
This site published its first few hundred pages in late July 2026. By the standards of the query we've been discussing it's a new domain with no history, and new domains wait months before Google trusts them with anything competitive. Whatever it ranks for on the day you read this, the honest reading is "not much yet".
We're also not in the map pack for most of the cities we write about, and we aren't going to be. A profile needs premises in that city and a verification step that proves it. Renting a desk purely to qualify would be exactly the behaviour this page is telling you not to reward.
What we're playing for instead is a few hundred specific question pages — the queries a founder types at eleven at night after a bad reporting call. Slower, far less impressive-looking, and the bet a domain this young can win honestly.
We won't publish a screenshot of our own Search Console either, because a screenshot is the easiest artefact on that table to fake. Ask on the call and we'll share the screen — ours, and with permission a client's.
Why we cut city pages, and what that cost us
The plan for this site called for more city pages than exist on it. We cut the ones that failed a test we'd already published: three sentences about that city that would be false about the next one. Not colour. Statements a local would nod at.
For some cities we had them. Mumbai's local pack genuinely behaves like several suburb-level markets because the city is long and slow, and its finance sector genuinely makes compliance review the bottleneck rather than writing. For others we had a place name and nothing else, so we didn't write the page.
That cost us. Each page dropped is an entry point we don't have on a domain that needs entry points, and it cost a sales argument too — it's easier to sell to a founder in a tier-2 city when you can send them a page with their city in the title.
It also answers a question clients ask: no, we won't build you forty of them. If your business genuinely operates in forty cities, there are forty true pages to write. If it doesn't, those pages are a bet that Google won't notice sameness, and copied city pages fail for reasons that have nothing to do with the copy.
What to do with an agency's SERP position, in fifteen minutes
None of this means ignore the search. It means spend the right amount of time on it and weight it correctly.
- Search the term from your office in a private window, then from home. If the top three change, you've learnt the result is a map rather than a verdict.
- Ignore the map pack unless you're hiring a local-only agency for local-only work — in which case it becomes one of the more relevant tests available.
- Strip out the ads, the directories and the paid lists before you conclude anything about who is ranking.
- Check their own site the way you'd check a supplier's factory: mobile rendering speed, whether posts link to anything, whether titles are distinct.
- Ask what they rank for in your category rather than in theirs, then verify two of those queries yourself.
- Weight the whole exercise as a tiebreak between two agencies you already like on the evidence that costs money to fake.