How much can SEO really grow a business? Here's the honest ceiling

SEO captures demand. It does not create it.

This is the sentence that should be on the first slide of every SEO pitch and never is. Search is a harvesting channel. Somebody has already decided they have a problem, put words to it, and typed those words into a box. All SEO does is make sure you're what they find.

Which means the size of the opportunity was fixed before you hired anyone. Your agency's skill decides what share of the existing demand you capture. It does not decide how much demand exists. A brilliant team in a category nobody searches for will lose to a mediocre team in a category with 200,000 monthly searches, every time, and neither outcome says much about either team.

So the first question isn't "which agency". It's "how big is the pool". Founders skip it because it's arithmetic and the pitch was more fun, then spend eighteen months learning the answer expensively.

Sizing your non-brand demand in an afternoon

You don't need an agency for this and you shouldn't outsource it. Anyone who's going to quote you a retainer has an incentive to find the pool large.

The output is one number: total monthly searches, in your market, for queries a buyer would actually type. Not your industry's total search volume. Not queries typed by students, job-seekers or competitors.

1. Pull the organic keywords of your three closest competitors. Any of the standard tools will do it. If you can't afford one, a month's trial is enough — this is a one-off exercise.
2. Strip out every branded query. Their brand, your brand, anybody's brand. Brand demand isn't available to you and including it inflates the pool badly.
3. Strip out the non-buyers. "Salary", "course", "jobs", "free", "template", "login", "how to become". In India these can be a third of the raw list, and none of it buys anything.
4. Group what's left by intent. Problem-aware, solution-comparison, and ready-to-buy. Keep the three buckets separate — they convert at wildly different rates and averaging them hides the answer.
5. Sum the volumes, then cut the total by a third. Keyword tools in India report bucketed, rounded estimates and over-count long-tail variants of the same query. The cut is a planning haircut, not a measurement.
6. Sanity-check against Search Console. If you already rank for anything, compare the tool's estimate against your actual impressions. The gap tells you how much to trust the rest of the file.

The ceiling model, in one line

Non-brand monthly searches × realistic click share × visit-to-lead rate × close rate × average order value = your monthly organic revenue ceiling.

Every term in that chain is a range, so run it twice: once pessimistic, once optimistic. The gap between the two answers tells you how much you're actually betting.

The term people get most wrong is click share. Ranking first doesn't mean getting the click any more. AI Overviews, People Also Ask, map packs, shopping units and site links all sit above or beside you, and the effect is heaviest on exactly the informational queries agencies like to report on. Published click-through studies disagree with each other because they sample different query mixes — so treat any single figure as a planning assumption, not a fact.

• You will not rank first for everything. Model capturing 20–40% of your filtered pool within 18 months, not 100%.
• Use your actual site conversion rate from analytics, not an industry benchmark. Benchmarks are averages across companies that aren't you.
• Run the model on qualified leads if your sales cycle is long. Revenue eighteen months out is a guess wearing a suit.
• If the pessimistic case doesn't cover the retainer plus your time, the answer is no. That's the model earning its keep in one afternoon.

Worked example one: compliance software with real demand

A B2B SaaS in Bengaluru selling GST and e-invoicing software to Indian finance teams. The demand here is genuine and structural — compliance rules change, deadlines arrive, and finance teams go looking. Queries like "e-invoicing software India" and "GST return filing software" are typed by people with a budget and a deadline.

The numbers below are illustrative arithmetic, not measured results. Swap in your own from the exercise above.

Say the filtered non-brand pool comes to 40,000 monthly searches after the haircut. Assume you reach top-three positions on the commercial third of that pool within 18 months, and page-one-but-lower positions on the rest. That's roughly 40,000 × a blended 8% click share ≈ 3,200 visits a month. At a 2% visit-to-demo rate that's 64 demos, and at a 20% close rate, 13 new customers a month.

At that point the question answers itself. If a customer is worth ₹60,000 a year, thirteen a month is worth building a company around. If a customer is worth ₹6,000, it's a supplement to a paid programme. Same traffic, opposite decisions — which is why traffic on its own is a useless target.

Worked example two: the D2C brand nobody is searching for

Now a D2C brand launching a new-format snack — a protein-forward version of a traditional Indian savoury, in a category that didn't exist eighteen months ago.

Run the same exercise and the file comes back nearly empty. There's demand for the traditional product, mostly satisfied by kirana shops and marketplaces. There's demand for "protein snacks", already owned by large brands and by Amazon and Flipkart listings that will out-rank a new domain for years. And there's zero demand for your format, because you invented the name last quarter.

The ceiling model returns close to nothing, and that's the correct answer, not a failure of ambition. Demand for a new format has to be manufactured — on Meta, on creator content, on shelf. People search for it after they've met it somewhere else. Buying SEO first is paying to harvest a field nobody planted.

What we'd actually tell this founder: spend on performance marketing and creator content for twelve to eighteen months, and keep Search Console running the whole time. The month you see people searching your brand and your format by name, the demand exists and SEO becomes the cheapest way to serve it. Indian D2C brands and the ad-spend problem is the other half of this argument.

Categories where SEO can't be the growth channel

Not a list of hopeless businesses. A list of businesses where search is a support act and something else is the show.

• Genuinely new categories. No query exists yet. You'd be optimising for a search box nobody is typing into.
• Impulse and discovery purchases. Fashion, trend snacks, novelty gifting. The decision is made in a feed, not a search bar. Search picks up the brand query afterwards.
• Marketplace-dominated product queries. For a lot of physical products in India, the first page is Amazon, Flipkart, Myntra and three aggregators. You can win the informational queries around the product and still lose every transactional one.
• Relationship-sourced enterprise deals. If deals come from twenty named accounts and a conference, there's no search volume to capture. Content still helps you close — it just isn't a lead source.
• Single-location businesses with a tiny catchment. The map pack and a well-kept Google Business Profile do most of the work. A ₹75,000 retainer is bigger than the opportunity.
• Categories where you can't legally say the thing people search for. Parts of health, finance and pharma. The query exists and the compliant answer to it doesn't rank.

Why year two beats year one, and it isn't magic

The compounding claim gets made so loosely that it sounds like faith. It isn't. It's arithmetic, and you can check every term.

Publish 60 pages in year one. In year two you publish 60 more — so you now have 120 pages earning, not 60. Meanwhile the year-one pages have aged: they've accumulated internal links from everything published since, picked up external links, and gathered enough click data for Google to have a settled view of them. A page that entered at position 14 in month four is frequently sitting at 6 by month sixteen without anyone touching it.

So three things multiply at once — page count, page age, and link count — and none of them was available to you in year one. That's why the curve looks flat and then steep, and why judging SEO at month six is like judging a term deposit at week three.

It runs the other way too. Stop paying and the asset doesn't vanish, but it stops growing and starts decaying — competitors refresh, results change shape, your pages age out of accuracy.

When to walk away and spend the money on ads instead

A simple decision rule, and we apply it to ourselves before quoting anyone.

If the pessimistic ceiling model returns less than roughly three times the annual cost of the retainer, don't buy SEO. Three times, because you're paying for eighteen months before the asset matures, and a channel that only just covers its own cost isn't worth the wait or the risk.

Below that line, paid search and paid social do the job better, faster and more reversibly. You'll pay more per lead forever, and in exchange you get answers in three weeks instead of nine months, and you can switch it off on a Tuesday. For a company that needs to know whether the market wants the product at all, that trade is obviously correct.

The version nobody says out loud: run performance first and let it tell you which queries convert, then bring SEO in to rank for the exact terms your ad account has already proved make money. That sequence turns SEO from a bet into a cost-reduction exercise.

And when SEO is right, the commitment should be measurable. We freeze your trailing-90-day count of qualified leads from organic search on day one, and if we haven't beaten it in 90 days we keep working free until we do. Not a ranking position — nobody controls Google's index — but movement on your own number. Our pricing is published, so you can run the three-times test against a real figure.