Price is a contract about incentives
Everyone shops pricing models as if the only variable is what they cost. The more useful question is what each model pays the agency to do on a Tuesday in month seven, when your account is one of eleven and nobody senior is watching.
That's not cynicism about agencies. It's how every professional service works, and it's why the model matters more than the rate.
The five models at a glance
The ranges are what the Indian market actually charges, not what we charge. Our own numbers are on the pricing page and we run a retainer with a baseline guarantee, for reasons that get clearer further down.
| Model | How it's quoted | Typical India range | What it pays the agency to do |
|---|---|---|---|
| Hourly | Per hour logged | ₹1,500–₹8,000 / hr | Log hours. Aligned with almost nothing you care about |
| Monthly retainer | Fixed fee for a block of capacity | ₹25,000–₹3,00,000 / mo | Earn the renewal. Good when honest, expensive when they coast |
| Fixed project | One scope, one price, one deadline | ₹50,000–₹5,00,000 | Ship and leave. No stake in what happens afterwards |
| Per-deliverable | Per article, per link, per page | ₹2,000–₹15,000/article · ₹3,000–₹25,000/link | Produce units. Volume beats fit, every time |
| Performance-based | Base fee plus bonus on an agreed metric | Base ₹25,000–₹75,000 + bonus | Hit the metric — useful only if the metric is honest |
Hourly, retainer and project — what each one quietly rewards
Hourly: honest for diagnosis, corrosive for retained work
Hourly is the cleanest model when the work has an end and a defined question. A migration review, a second opinion on a traffic drop, a technical audit by someone who's seen a hundred of them — pay ₹4,000 an hour for twelve hours and you've bought something specific.
It falls apart on ongoing SEO, and it fails in both directions. You start policing timesheets. The agency starts avoiding work that takes eight hours and produces nothing visible this month — which describes most of the technical and research work that actually decides your outcome. Always ask for a cap.
Retainer: buys capacity, not results, and that's the whole story
This is the dominant model and structurally it's fine. A retainer buys a defined block of monthly capacity across four disciplines. It does not buy an outcome, and if the outcome doesn't arrive, the invoice arrives anyway.
There is one mild built-in alignment: an agency that wants month thirteen has to make months one to twelve defensible. Long lock-ins remove even that, which is why a twelve-month minimum term is a worse deal than a higher fee on monthly terms.
Fixed project: right when the work genuinely ends
A replatform, a set of 40 category pages, a penalty recovery, an audit with implementation. All good candidates — the scope is knowable and so is done.
The incentive problem is stark: the agency is paid on delivery, so they have precisely zero stake in whether the pages rank. Mitigate it with acceptance criteria and a support window. "Indexed within 30 days and no crawl errors introduced" is a testable condition. "Delivered" isn't.
Per-deliverable pricing and the backlink counting problem
Pay ₹6,000 an article and you've told the agency that shorter is more profitable, that a topic's merit is irrelevant, and that the twelfth article on a subject is worth the same as the first. Nobody intends this. Everybody responds to it.
Per-link pricing is the sharper version of the same problem, because a link has no natural quality floor. Do the arithmetic: a genuinely earned placement takes 3–8 hours of research, pitching and follow-up, at a hit rate that's often 5–15% on cold outreach. At ₹5,000 a link, that's ₹600–₹1,600 an hour of revenue before any margin at all. Nobody in India delivers senior outreach at that rate, so at volume the only way to hit the number is a paid placement or a network. Which is what you bought, whether or not that's what you asked for.
There's a legitimate use. When you already have in-house SEO and just need capacity — "rewrite these 40 product pages to this brief" — per-unit pricing is efficient and inspectable. The condition is that you can reject units. Without that right, you're buying volume.
Why performance-based pricing bends toward the easy keywords
Performance pricing sounds like the answer to everything: pay for results, not effort. It's the model with the most attractive pitch and the most predictable failure, and the failure is always in how the metric is defined.
The mechanism is simple. Whatever you make the bonus, that becomes the job.
- Bonus on ranking positions. The agency chooses winnable terms. Branded queries you already rank first for. Near-branded variants. Long-tail phrases with twenty searches a month. You pay a bonus for movement that would have happened without them.
- Bonus on organic traffic. Volume is the cheapest thing on earth to manufacture. Expect informational content that brings visitors with no intention of buying anything, and a chart that goes up while the pipeline doesn't.
- Bonus on organic leads. Better — this is close to the right metric. The fight moves to what counts as a lead, so define it before signing, in your CRM, with a stage name.
- Revenue share. The most aligned in theory and the least workable in practice, for reasons below.
Why revenue share rarely works for SEO
Two structural problems, and neither is anyone's bad faith.
First, attribution. Is branded organic search — someone who saw your ad, then googled your name — organic revenue? Is a returning customer new? Do you count last-click or assisted? Every one of those is a negotiation you'll be having while trying to run a company, and both sides will be sincerely convinced they're right.
Second, timing. SEO pays out six to twelve months after the work. An agency starting in January collects revenue share in months one to six for rankings the previous agency built, then does the work in months seven to twelve that pays out after the contract ends. Both parties get the wrong end of it, just at different times. It's why revenue share so rarely reaches a second year — the longer argument is here.
Where it does work: a single-conversion lead-gen or affiliate property, one channel, clean attribution, and an agency with real control over the site.
What a guarantee actually changes about the risk split
A guarantee isn't a sixth pricing model. It's a clause bolted onto one of the five that moves the downside from you to the agency, and it's the only mechanism that fixes the retainer's core flaw — that capacity gets paid whether or not it works.
Ours is a retainer with a downside clause. We freeze your trailing-90-day count of qualified leads from organic search on day one, in your CRM, with a date on it. If we haven't beaten that number in 90 days, we keep working free until we do. We do not guarantee a ranking position for a keyword, because nobody controls Google's index and anybody who tells you otherwise is either buying links or planning to argue about wording later.
The thing worth understanding as a buyer: that clause is expensive for the agency. Every guaranteed client who underperforms consumes hours that can't be sold to anyone else, and those hours are the entire inventory. It's why we cap intake at three new clients a month, and it's why most agencies won't write the clause at all. Their reluctance is information, not villainy — see how our SEO engagements are structured.
- The metric. Business outcome or vanity number? Leads and revenue are real. Rankings and traffic are proxies.
- The baseline. Frozen when, measured where? "Your current performance" is not a baseline.
- The remedy. Free work, a refund, or a door you can walk out of? Each is legitimate; vagueness isn't.
- The exclusions. If the guarantee voids because you didn't approve content in five working days, that's fair. If it voids for reasons only the agency can assess, it's decoration.
Which model fits your situation
If you're comparing quotes across two of these at once, normalise them into hours first — the method is in how to compare SEO proposals. Two proposals in different units aren't a comparison, they're a coin toss with extra steps.
| Your situation | Model that fits | Insist on |
|---|---|---|
| One defined job: migration, audit, 40 product pages | Fixed project | Acceptance criteria and 30 days of post-launch support |
| A second opinion, or one specific problem diagnosed | Hourly, capped | A written cap and a document at the end, not just calls |
| Ongoing growth across all four disciplines | Monthly retainer | Itemised hours, named people, short notice period |
| You have in-house SEO and need production capacity | Per-deliverable | The right to reject units, and a sample before volume |
| You want the agency to carry some downside | Retainer with a baseline guarantee | The metric, the frozen number, the date, the remedy |
| A single-conversion property you fully control | Performance-based | Attribution rules agreed in writing before month one |