What "performance-based" actually means in a contract
The phrase covers at least four different deals that behave nothing like each other. Founders usually picture the second one and get sold the first. Work out which is on the table before you compare it to a retainer, because the risk transfer you think you're buying often isn't there at all.
- Pay-per-ranking. You pay a fee for every keyword that reaches page one, or a top-three slot, held for some number of days. The most commonly sold version and the most easily gamed.
- Revenue or lead share. The agency takes a percentage of tracked organic revenue or a fee per qualified lead. Genuinely aligned, and genuinely hard to attribute cleanly.
- Base plus bonus. A reduced retainer covering costs, with an upside payment tied to an agreed outcome. The most workable structure in practice, and the one serious agencies actually offer.
- Money-back guarantee. Not performance pricing at all — a retainer with a refund clause. Read what triggers it, who judges it, and whether the refund covers all months or the last one.
The flaw: whoever picks the keywords picks the outcome
Here's the mechanic that makes pay-per-ranking almost impossible to sell honestly. In nearly every one of these contracts, the agency proposes the keyword list. That single detail hands them control of whether they get paid.
A competent SEO can look at your site and know within an hour which queries you'll rank for anyway. Long-tail variants of your brand name. Queries where you already sit at position 12 and one internal link fixes it. City-plus-service phrases with fifty searches a month and no competitor doing anything. Put forty of those on the list, wait ninety days, invoice for thirty-one page-one rankings, and every word of the contract has been honoured.
You'll have paid a performance fee for rankings that would have arrived anyway, while the four queries your buyers actually type never made the list — because putting them there would have put the fee at risk. That's not a hypothetical failure mode; it's the predictable equilibrium of the incentive. Any model that lets one side define the scoreboard drifts toward the easy scoreboard, and that's true of honest people too.
Position-based versus revenue-based, side by side
The two families of performance deal fail in opposite directions. One is easy to measure and easy to game. The other is hard to game and hard to measure.
- Revenue share sounds cleanest and is the hardest to run. Somebody has to decide whether a returning customer who searched your brand name counts as organic — and whichever way that call goes, one party is unhappy.
- Base-plus-bonus works because the base keeps the lights on and the bonus does the aligning. It's the only performance structure where an agency can afford to spend three months on technical debt that pays off in month five.
| Model | What you pay for | What it quietly incentivises | Where it breaks |
|---|---|---|---|
| Pay-per-ranking | Each keyword hitting an agreed position | Winnable keywords, not valuable ones | Agency writes the list; volume of easy wins beats value of hard ones |
| Revenue / lead share | A % of tracked organic revenue or a per-lead fee | Conversions, which is the right thing | Attribution. Organic gets credited for brand traffic, retargeting and direct returns |
| Base plus bonus | Reduced retainer plus upside on an agreed metric | The metric you jointly chose | Only works if the metric is picked before work starts and can't be redefined later |
| Flat retainer | Capacity — hours, pages, placements | Long engagements; excellent when honest, expensive when the agency coasts | You carry all the risk unless the contract says otherwise |
| Retainer + baseline guarantee | Capacity, with the fee at risk against your own frozen number | Beating a number you both agreed on before day one | Requires clean CRM data and an agency willing to work free if it misses |
Why no honest agency promises a specific Google position
Not because agencies lack confidence. Because nobody outside Google controls the index.
Core updates land without warning and re-sort entire verticals. AI Overviews now sit above the classic results for a large slice of informational queries, so "position one" can mean less traffic than it did two years ago. A competitor can double their content budget the week after you sign. Your own developers can ship a release that noindexes a template. Pricing as if all that were controllable means either the guarantee is fake or you're paying an invisible insurance premium.
Google's spam guidance treats guaranteed-ranking claims as a red flag for a reason: historically the only reliable way to promise a position was to buy links or manipulate the index, and both end with your domain suppressed rather than your competitor's. More on why ranking guarantees are a lie.
There is a version of a guarantee that survives all of this. It just never mentions a Google position.
A baseline guarantee: what it measures and how it's frozen
A baseline guarantee moves the scoreboard from Google's index — which nobody controls — to your business, which you can both see. The commitment is movement against your own past performance, not against a position on a results page. The mechanics matter more than the promise, and three things have to be true or it's marketing copy.
Ours works exactly this way. We freeze your trailing-90-day count of qualified leads from organic search on day one. If we haven't beaten it in ninety days, we keep working free until we do. SEO runs from ₹75,000/mo, with smaller sites from ₹40,000/mo, ex-GST, committed first quarter then month-to-month with 30 days' notice — and you keep every asset if you leave. It's why we cap intake at three clients a month: you cannot carry that risk across forty accounts and mean it. Full terms sit on the SEO page.
Notice what the guarantee does not say. It names no keyword, no position, no traffic number. It says the thing your business already measures should be bigger in ninety days than it was in the ninety before we arrived — a claim you can check without trusting us.
- The metric is yours, not the agency's. Qualified leads or revenue from organic search, pulled from your CRM — not sessions, not impressions, not an average position across queries you never targeted.
- The baseline is frozen before work starts, in writing, from your trailing ninety days. A guarantee agreed in month four against a number chosen in month four isn't one. See how to set an SEO baseline.
- The remedy is specific and non-discretionary. "We keep working free until we beat it", not "we'll review the engagement in good faith" — which means nothing and is enforceable by nobody.
How to build the deal yourself if your agency won't offer one
Most agencies won't volunteer a baseline guarantee, and that's not automatically a bad sign — carrying it needs a cash position and a client-selection discipline most shops don't have. Do these four things at contract stage instead and you convert a capacity purchase into something close to an outcome purchase, without asking anyone to sign a clause they can't fund.
- Write the baseline down anyway. Even with no money attached, a frozen trailing-90-day lead count changes every conversation from month three onward. Both sides now argue about the same number.
- Split the fee. 80% base, 20% quarterly bonus on the agreed metric. Most agencies accept because the base still covers delivery; the ones who refuse are telling you they don't expect to hit it.
- Cap the term at a quarter. Long enough to be fair, short enough to be a real threat. Lock-in and performance pricing are substitutes — an agency shouldn't get both.
- Own the keyword list. You approve additions and removals. Every structure above collapses if this point is conceded, and it costs an honest agency nothing.