Why "best SEO agency" is a question with no answer
Consider what a genuine ranking would need. You'd have to compare agencies on outcomes, which means seeing every client's baseline and result — including the ones that failed. None of that is public. Agencies publish their wins, never their losses, and no independent body audits either.
Then there's the sample problem. An agency with a brilliant record in Indian SaaS may be genuinely mediocre at multi-location dental clinics, because the work isn't the same work. One is topical authority and product-led content. The other is map pack, reviews and location pages. "Best" without "at what, for whom" is a category error.
And the practical problem: the agency that's best for you may be at capacity, or may not take accounts your size, or may already work with your competitor. A ranking that ignores availability ranks a menu you can't order from.
So the question has to change. Not "who is the best?" but "which of these three is most likely to work for my site, at my budget, in my timeframe?" That question has an answer, and you can compute it in an afternoon.
What "best" means at each stage of a site
Before scoring anyone, work out which problem you actually have. The right agency for each of these is a different agency, and the pitch decks look nearly identical from the outside.
| Your situation | What "best" means here | What you're really buying |
|---|---|---|
| New site, under 12 months old, few links | Patience, foundations and honest timelines | Architecture, a content plan you can sustain, and someone who'll tell you SEO isn't your first channel |
| Established site, traffic flat for a year | Diagnosis before activity | An audit that finds the actual constraint — usually cannibalisation, thin coverage or a technical ceiling — not another content calendar |
| Site that dropped after an update or migration | Forensic technical work | Someone who can read a traffic chart against update dates and rebuild what broke. A content team can't fix this |
| Ecommerce with a large catalogue | Scale, templates and faceted navigation control | Crawl budget management, category page strategy, and someone who has argued with a Shopify or Magento dev before |
| Multi-market or multi-brand | Coordination more than craft | Hreflang, governance, and the ability to get 40 stakeholders to agree on a URL structure |
The scorecard: nine criteria, weighted
Score each criterion 1 to 5 for each agency, multiply by the weight, and divide the total by 5 for a percentage. The weights below reflect what actually predicts an engagement working, which is not what the pitch spends its time on.
One rule makes this whole exercise honest: a 5 requires evidence you can inspect. A claim is a 3 at most. Apply that consistently and most shortlists separate within twenty minutes.
| Criterion | Weight | What a 5 looks like |
|---|---|---|
| Fit with your stage and site type | 15 | They've done your specific problem — recovery, catalogue scale, local multi-location — and can describe it without you prompting. |
| Who actually does the work | 15 | Named people, named roles, honest about anything subcontracted, and those people are on the call. |
| Baseline definition and measurement | 15 | A business number frozen before work starts — organic qualified leads or organic revenue, dated and signed off. |
| Evidence, including a failure | 12 | A real account that didn't work, with a named cause they take responsibility for. Plus wins with the starting numbers shown, not just the ending ones. |
| Link sourcing and content process | 12 | Live placement URLs from last month, a describable pitch behind each, and a written refusal to buy links. |
| Contract terms and exit | 10 | One quarter committed, then month-to-month on 30 days' notice, with every asset staying yours. |
| Technical depth | 8 | Someone in the room who reads log files, has run a migration, and can talk to your developers without a translator. |
| Reporting clarity and cadence | 8 | A sample report that leads with the business number, explains why it moved, and states what changes next month. |
| Price against scope realism | 5 | The scope is arithmetically deliverable at the price quoted. They'll walk you through the hours if you ask. |
How to score three agencies without kidding yourself
The scorecard fails in one predictable way: you fill it in after you've already decided, and the numbers obediently agree with you. Six rules stop that.
- Score within an hour of each call, before the next one. Memory of a likeable person outlasts memory of a weak answer, every time.
- Use whole numbers only. 3.5 is a way of avoiding a decision. Round down when you're torn — being torn is information.
- Never score the deck. Score the answers to questions you asked. If a criterion never came up, it scores 1, and that's your fault to fix in a follow-up email, not theirs to be credited for.
- Apply the evidence rule strictly. "We're very strong technically" is a 3. A named migration they ran, with what broke and how they caught it, is a 5.
- Score fit against your stage first, using the table above. If an agency scores 2 on fit, the other 85 points are decorating a mismatch.
- Compare percentages, not gut feel. Then, if the gut disagrees with the number, work out which criterion the gut is reacting to and re-score that one honestly. Usually it's criterion two.
Five answers that zero the whole score
Some answers aren't a low score on one line. They end the evaluation, because they reveal a business model rather than a weakness. An agency can be superb on eight criteria and still be un-hireable on the ninth.
- A guaranteed ranking position. Nobody controls Google's index and Google's own guidelines warn against exactly this. A guarantee against *your own baseline* is a real commitment; a promised #1 is a sales tactic. More on why ranking guarantees don't hold up.
- "Our private network" for links. That's a private blog network, it violates Google's link spam policy, and the consequence lands on your domain rather than their invoice.
- No baseline, or a baseline made of traffic. If nobody records where you started, no report can ever prove you moved — and traffic is trivially inflatable with irrelevant queries.
- Refusing to name who does the work. "You'll have a dedicated account manager" describes the person who forwards email. It's not an answer to the question.
- A 12-month lock-in with a 90-day notice period. That contract is engineered around their revenue risk, not your result. Length of commitment is the clearest confidence signal an agency emits.
Why directories, awards and "top 10" lists aren't evidence
Most shortlists start with a list somebody else wrote, so it's worth knowing how those lists are made.
Agency directories commonly sell sponsored placement, and rank partly on review volume — reviews the agencies themselves solicit from clients they choose to ask. That's a measure of sales effort and client-management skill. Both are real skills. Neither is SEO.
"Top 10 SEO agencies in [city]" blog posts are a further step down. They're usually written by an agency that appears on them, by a directory monetising affiliate links, or by a content farm targeting the keyword. Check who owns the domain before you treat the ranking as research — we've written about how these lists actually get built.
Awards are worth roughly what the entry fee suggests. Most are judged on submitted case studies, which means they measure the quality of an agency's award-writing, which is a genuine craft and not the one you're buying.
Why we don't publish a list either
We could rank the Indian agency market. It would rank well, it would bring traffic, and every page like it we've read has been quietly self-serving.
The honest version of that page is this one: a method you can run yourself, on whatever three names you've collected, with us scored on the same nine criteria as everyone else. If we lose on fit, we should lose — a four-page local site does not need what we build.
What we'll say plainly is what we sell and at what price, so you can score us. SEO from ₹75,000 a month, smaller sites from ₹40,000, ex-GST. Month-to-month after the first quarter, 30 days' notice, and you keep every asset. We freeze your trailing-90-day qualified organic leads on day one and if we haven't beaten that number in 90 days we keep working free until we do. Never a promised position — movement against your own number, which is the only thing an agency can honestly stand behind.
That's also why we take three clients a month. You can't carry that risk at volume, and any agency claiming to would be pricing a guarantee it never intends to pay out. See how the engagement works.