The four conditions. Fail one and don't sign.
We turn down more work than we take. Not out of principle — out of arithmetic. An SEO retainer is a fixed monthly cost against a return that arrives late and arrives unevenly, and there are whole categories of business where that trade never comes good. Selling a retainer to one of those founders is how agencies earn the reputation the whole industry now carries.
So here's the qualifying test we run before a call becomes a proposal. Four conditions. All four have to be true. When one fails we say so on the call, and usually we name the cheaper thing to do instead.
| Condition | How we test it | If it fails |
|---|---|---|
| Demand already exists | Real non-brand search volume for the problem you solve, not just for your company name | You have a demand problem. Buy attention first, harvest it later. |
| Contribution above ₹3,000 per lead | Gross margin on a closed deal × the rate at which inbound leads close | The arithmetic never clears a retainer. Freelancer, or do it yourself. |
| You can change the site | A CMS you control and someone who can ship a template change inside two weeks | Fix that first. Otherwise you're paying us to write tickets nobody actions. |
| Nine funded months | Cash to cover the retainer for nine months without it becoming a board conversation in month four | Buy demand you can switch off instead. |
Condition one — somebody is already typing this into a search box
SEO harvests demand. It doesn't create it. If nobody searches for the problem you solve, the best technical work in the country wins you position one on a query with eleven searches a month, and eleven searches a month doesn't pay a retainer.
The check takes twenty minutes and you can do it without an agency. You're looking for evidence that strangers describe your problem in words, unprompted, to a search box.
- Strip the brand out of Search Console. If your site has any history, set the range to 12 months, sort queries by impressions, and remove everything containing your company name. What's left is your real non-brand demand today.
- Search the problem, not the product. In Keyword Planner, people type *gst software for small business*, not the name of your feature. If the problem phrasing has volume and your product phrasing doesn't, that's fine — that's the whole game.
- Count who's already answering. Three competitors with dedicated pages means demand exists and is being served. Zero pages anywhere usually means there's nothing there, not that you've found a gap. The gap theory is right maybe one time in ten.
- Read the SERP shape. If the top results are forum threads and Reddit, that's demand with no good commercial answer yet. That's the best sign you'll ever get.
- Check the seasonality. Admissions, tax, wedding, festive — if 70% of your demand lands in eight weeks, the programme has to be built backwards from those eight weeks, and a January start for an October peak is the only sane sequence.
Condition two — one extra lead has to be worth more than ₹3,000
This is the number that disqualifies most people, and almost nobody runs it before taking meetings.
Contribution per lead isn't your deal size. It's gross margin on a closed deal multiplied by the rate at which inbound leads actually close. Sell a ₹40,000 service at 60% margin and close one inbound in five, and each lead is worth ₹4,800. Sell a ₹1,200 product at 30% margin with a 4% conversion rate, and each visitor is worth about ₹14.
Now hold that against the retainer. Our SEO starts at ₹75,000/mo, with smaller sites from ₹40,000/mo, both before 18% GST. Here's how many extra leads a month each has to produce to break even — not to profit. To break even.
| Contribution per lead | To cover ₹40,000/mo | To cover ₹75,000/mo | Honest verdict |
|---|---|---|---|
| ₹500 | 80 | 150 | Only at genuine ecommerce volume |
| ₹1,500 | 27 | 50 | Hard inside year one for most B2B |
| ₹3,000 | 14 | 25 | The floor where a retainer starts to work |
| ₹8,000 | 5 | 10 | Comfortable, and the maths survives a bad quarter |
| ₹25,000 | 2 | 3 | This is where SEO looks like a pricing error in your favour |
Why ₹3,000 is the line, and what sits under it
Twenty-five extra leads a month is a number a founder can see without squinting at a dashboard, and it's a volume a competent team can reach on a mid-size site inside nine months. That's the whole reason ₹3,000 is our threshold — it's the point where the required lead count stops being fantasy.
Under ₹1,500 per lead you need volumes that only ecommerce, marketplaces and high-frequency consumer categories produce. At that point you're not really buying SEO strategy, you're buying merchandising, feed hygiene, faceted-navigation control and a content operation — a different service with different people in it, and it should be priced and staffed as one.
There's an escape hatch worth naming. If your contribution per lead is low but your repeat rate is high, use lifetime contribution instead of first-order contribution. A ₹400 first order at 30% margin looks hopeless until you notice the average buyer orders four times in eighteen months. Just be honest about the repeat rate — use your actual cohort data, not the number in the pitch deck. The full range of what the work costs sits in SEO cost for a small business.
Condition three — you can change your own website this week
Half of SEO is edits. Title tags, internal links, a template change, a redirect map, schema, two pages that need to become one. If those edits take six weeks to ship, nobody is running an SEO programme. They're running a queue.
The failure modes we meet, roughly in order of frequency:
- The site was built by an agency you stopped paying, and nobody in your building has the login.
- It's on a page builder where every template change costs money, breaks the layout, or both.
- There's one developer, they're building the product, and marketing tickets sit behind three sprints — permanently.
- Legal reviews every word. Genuinely unavoidable in finance and healthcare, and it means content velocity gets planned around review capacity rather than promised in a deck.
- The blog lives on a subdomain owned by a marketing tool, and moving it is a migration nobody has budgeted.
Condition four — nine funded months, and the nerve to leave it alone
Ninety days tells you whether it's working. Nine months tells you whether it paid. Those are different questions, and conflating them kills more programmes than bad execution does.
By day 90 you should see impressions climbing on your target queries, pages moving into the index, average position improving on the specific terms in the plan, and on short sales cycles the first leads. Months five to nine is where compounding starts showing up in the bank rather than the dashboard.
₹75,000 a month for nine months is ₹6,75,000 before GST. If writing that number down makes you uneasy, the unease will resurface in month four as a request to pause and re-evaluate. A paused programme doesn't hold its position — it stops, while everyone who kept publishing carries on. Pausing at month four is the most expensive way to spend ₹3,00,000 in Indian marketing.
You may have a demand problem, not an SEO problem
This is the diagnosis founders least want and most often need. Five tests, cheapest first.
- Buy the clicks first. Put ₹30,000–₹50,000 across two weeks into exact-match versions of your five best keywords. If those clicks don't convert, ranking for them for free won't convert either — you'll simply get the same non-conversion at a lower cost. It's the cheapest diagnostic in marketing and almost nobody runs it before signing a twelve-month SEO contract.
- Look at the organic traffic you already have. If 500 organic sessions a month produce zero enquiries, the problem isn't volume. Doubling zero is still zero, and you'll have paid ₹9,00,000 for the privilege.
- Count your brand searches. If people search your name and convert but nobody searches your category, you're in a referral-led market. SEO's job there is defensive — own your name, own comparisons — and that's a much smaller brief than the one you were about to buy.
- Ask your last ten customers how they found you. Not the last ten leads. The last ten who paid. If none of them came through search, the channel isn't where your category buys, and no agency changes that.
- Check whether the category has a search habit at all. Some things get bought on Instagram, some in WhatsApp groups, some through a distributor's rep who visits on Tuesdays. If your buyers don't use a search box, organic is a secondary channel for you at best and a vanity project at worst.
What to do instead, in three cheaper shapes
Failing the test doesn't mean doing nothing. It means buying the right size of the thing.
The local business that needs a profile and six pages
If you're a single-location clinic, salon, studio, restaurant, gym or trade, you probably don't need us and may not need anyone. Local results are decided largely by proximity, relevance and prominence, and none of those three are bought with a retainer.
- A complete, verified Google Business Profile — correct category, real hours, service list, twenty photos of the actual premises, and posts you update monthly rather than never.
- A homepage that names the city and the service in plain words, because the map pack and the local pack read your site too.
- One page for each service you actually sell. Two or three pages, not fourteen. Every page a person could have written.
- An area page only if you genuinely serve more than one locality, with real detail about each. Twelve near-identical locality pages is the fastest way to look like spam to Google in 2026.
- An about page with real names and faces, and a contact page with the address, hours and phone number as selectable text, not baked into an image.
- A standing habit of asking every happy customer for a review. Steady and honest beats twelve reviews in one afternoon, which reads as bought and often gets filtered.
The ₹20,000-a-month freelancer who'll beat us on your site
At Indian rates of roughly ₹1,000–₹1,500 an hour for someone competent, ₹20,000 buys 13 to 20 hours of attention a month. That's genuinely enough when your site is under forty pages, in one language, with one service line and no migration in the plan. One good person doing on-page work and technical hygiene captures most of the available return on a site that size.
Where it stops working is breadth: link acquisition at any scale, multi-market or multi-language rollouts, faceted ecommerce navigation, a replatform, or anything needing four specialisms in the same month. We laid out the full split in agency against freelancer.
The junior hire plus ₹10,000 of tools
If you have volume but not budget, a sharp junior marketer with a paid crawler and a rank tracker will out-produce a cheap retainer, because a cheap retainer is one junior person split across nine accounts anyway. The gap is judgement: they won't know what to do first.
Buy that separately. A few hours a month of a senior person setting the queue and reviewing the work costs a fraction of a full retainer, and it's the arrangement we'd recommend to our own siblings before we'd recommend ₹25,000-a-month SEO from anybody.
What happens when we say no
We take three clients a month. That isn't scarcity marketing, it's the only way to carry what we sell: 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. Nobody can underwrite that across twenty accounts, and nobody sane underwrites it for a business that fails the four conditions above.
So a fail gets a no, a reason, and a cheaper recommendation — sometimes a name, sometimes a two-page plan you can hand to a freelancer. We lose the deal. We also avoid spending nine months explaining to a founder why a number won't move, which turns out to be the better trade for both of us.
The uncomfortable version, since this is our site and we're the ones losing revenue by writing it: the single biggest predictor of whether SEO works for you is decided before you hire anyone. It's your margin, your market's search habit, and your patience. An agency can only add competence on top of those three. It can't supply them.