Pay-for-Results SEO Pricing: Where It Goes Wrong

Quick answer: Pay-for-results SEO pricing goes wrong when the number you pay on is easier to move than the thing you actually want. A fee per ranking rewards easy keywords and brand terms that add little new revenue, and a share of “organic revenue” turns into an argument about whose report is right. A fair version is usually a hybrid: a fixed fee that covers the work, plus a capped bonus on one narrow metric such as non-brand organic orders above a written baseline. The metric, data source, brand-term list, exclusions and exit terms are agreed in writing before work starts.

If an agency has ever sent you a report full of impressions and no orders, “you only pay when it works” is an attractive sentence. Founders carry the risk on every retainer, and a fee tied to results looks like a way to move that risk onto the agency.

The catch is that a performance deal does not remove the incentive problem. It relocates it. The agency will now do whatever the trigger rewards, and the trigger is almost never the same thing as “more profitable orders from search”. I sell SEO, so read this with that in mind. I am also not saying performance deals are scams. Many are offered in good faith. I am describing specific incentive problems so you can spot them in a proposal and fix them in the contract.

The dollar figures and the percentage in the worked examples are made up for illustration. None is a Prism3 rate or a market benchmark. Facts about Google, GA4, Shopify and the ACCC are as at 31 October 2026.

The three shapes of a performance deal

Providers use different names and some blend the shapes, so ignore the label and ask for the trigger: what exactly has to happen, measured by whom, for money to move. That one question exposes most of the problems below.

Pay per ranking: five ways it bends

1. The cheapest keywords get done first

If every keyword that reaches the top three pays the same bonus, the agency’s best move is the keywords that are easiest to reach, not the ones that sell the most.

Nothing here needs bad faith. An agency paid per keyword simply has a reason to propose the list that pays out soonest, so the keyword list is the contract. A safer list comes from your own data: queries you already get impressions for, filtered to products you want to sell more of. Our guide to page-two keywords and Search Console quick wins shows how to build one.

2. Your brand name already ranks

People who search your brand name mostly know you already, from an ad, an email, a podcast, a friend or last month’s sale. Google’s own announcement of the branded queries filter in Search Console says branded queries typically lead to higher-ranking pages and higher click-through rates, while non-branded queries are the ones that show how new people find you. So any metric that mixes the two pays the agency for marketing that happened elsewhere. We do not run ads or email at Prism3, but both can send people to search for your name, which inflates the number an SEO bonus is measured on.

Search Console can separate the two. Google announced the branded queries filter on 20 November 2025 and updated the post on 11 March 2026 to say it was available to all eligible sites. It has limits. Classification is done by an internal AI-assisted system that Google says can misidentify some queries. It is only available for top-level properties, not for URL-path or subdomain properties, and only for sites with enough query volume. If your property qualifies, use it, and also write your own list of brand terms into the contract (brand name, common misspellings, names of products only you sell).

3. “Position” is not one number

Google’s Search Console help defines position as the topmost place your link held for each query, averaged across queries, recorded only when the link got an impression. It also says what you see in your own search can differ from the average because of variables such as search history and location. A third-party rank tracker is a different measurement again.

So “position 3” needs specifying: which tool, which country, desktop or mobile, on what date, and whether it has to hold (say, four weeks in a row). Otherwise the dispute is two screenshots.

4. A ranking can be real and still not sell

A page-one position for an informational query can bring visitors who never buy. If the bonus is on position or visits alone, you can pay for results that add nothing to orders. Pay on something closer to money: orders, or at least clicks to the product and collection pages that sell. Orders need a source you trust first, which is why GA4 and Shopify Analytics disagreeing matters.

5. Pressure to hurry, and where it can lead

Google’s SEO starter guide says some changes take effect within hours while others can take several months. If a fee depends on a short window, the sensible move for the agency is whatever shows a result inside it. That does not mean a particular agency cuts corners. It means the incentive exists, and Google’s spam policies list what lies at the edges: buying or selling links for ranking, mass-generated pages and doorway pages. Google says violating sites may rank lower or not appear at all, and its hiring guidance says you are responsible for the actions of any company you hire. So put the standard in the contract: the work follows Google’s spam policies, you get a written list of every change with the reasoning, and nothing off-page goes ahead without your approval.

Revenue share: where the argument starts

A percentage of organic revenue sounds like the fairest shape, because the agency only earns when you sell. It also has the most hidden steps.

Whose number is the number?

A revenue share needs one agreed measure of “organic revenue”. Most stores have at least two, Shopify and GA4, and they count differently. Specific ways the figure can move without the SEO changing:

• The attribution model. Shopify’s marketing reports offer several models, including last non-direct click (which leaves direct visits out) and last click (which counts them). Shopify says sales in a marketing report can differ from sales figures in other reports. GA4 offers data-driven, paid and organic last click, and Google paid channels last click, and direct visits get no credit unless the whole path was direct.
• Settings someone can change. Google’s help says that changing the reporting attribution model in GA4 applies to historical as well as future data. The lookback window, which defaults to 90 days for most key events, applies going forward. So a person with the right access can change an already-reported month.
• Channel definitions that Google changes. GA4 counts non-ad search links, including Google’s AI Overviews and AI Mode, as Organic Search, and it says these default definitions cannot be edited. On 13 May 2026 GA4 added an AI Assistant channel for visits from tools such as ChatGPT, Gemini and Claude. A contract signed before that date may not say whether those orders count towards the bonus. Decide, in writing, whether they do.

So name the report, the attribution model and the date range in the contract, screenshot the settings on day one, and agree that neither side changes them without written notice. Our guide to why attribution is broken on Shopify covers the gaps.

You can pay for your own sale (illustrative)

A sale month makes the baseline problem obvious. Say your organic search revenue is about A$30,000 in a normal month. In November you run a sale, email your list and spend on ads, and some of those shoppers search your brand and click an organic result. Organic revenue lands at A$39,000. Last November it was A$34,000. The agreed share is 10%, a made-up rate.

Same store, same month, same percentage, and the payout runs from A$500 to A$3,900, almost eight times as much. None of the three separates what SEO added from what your sale added. A fairer design uses the same month last year as the baseline, counts non-brand revenue only, and adjusts for events you tell the agency about. If the store was already growing before the agency arrived, say so in the baseline, so the agency is paid for growth above the existing trend and not for the trend.

Revenue is not profit

A share of revenue pays the same for a healthy-margin order as for one discounted to break-even, and for orders refunded later. Define revenue as net of discounts, refunds and returns, excluding tax and shipping, and pay after your returns window closes. If you can share margin, a bonus on contribution margin is better still: our worked example shows the arithmetic.

What the agency cannot control

A stock-out on your best collection, a theme change that slows pages, a price rise, a checkout bug, a competitor’s sale, your own ad and email calendar: all move organic revenue and none is the agency’s doing. A revenue-based fee hands those swings to the agency on the downside and credit on the upside. Both are unfair, so write the exclusions down and share your promotional calendar.

Better still, pay on what the work controls and use revenue as a check. For example: the bonus triggers on non-brand organic clicks to an agreed set of collection and product pages (from Search Console), and only pays if orders from those pages did not fall.

Small stores: the noise problem (illustrative)

Suppose your store gets 30 organic orders in a typical month. If orders arrive roughly at random, with no trend, the typical month-to-month swing is about the square root of 30, around 5 or 6 orders, which is roughly 18%. A bonus that triggers at “10% more orders than baseline” would then be hit by chance alone in about one month out of three. A 20% trigger would still be hit by chance roughly one month in six. At 300 orders a month, a 10% trigger is hit by chance about one month in twenty. These are my calculations on a simple random model, and real stores swing more because of seasons and promotions. The smaller the volume, the more a bonus pays for luck, so set the trigger above the normal swing, measure over a quarter, or use a bigger-number metric such as clicks to the pages in scope.

All or nothing picks the work and the stores

A deal with no fixed fee has to be funded by the bonuses on the stores that work. That gives the agency reasons to prefer stores and pages already close to ranking, and quick wins over slow work such as technical fixes, collection page rewrites and site structure. Google says some changes take months, so a 90-day trigger pays for the quick part of the job and drops the slow part. How long Shopify SEO takes covers why timing is hard to promise.

This is reasoning, not a claim about any agency’s behaviour. The test is a question: what will you do in the first 90 days if nothing is rewarded for 90 days?

Forecasts and guarantees in the proposal

Performance offers often arrive with confident forecasts attached. Google’s hiring guidance says: “No one can guarantee a #1 ranking on Google.” A no-win-no-fee deal is not a guarantee, but a forecast is still a claim. The ACCC says a business making claims about future matters, including predictions, must have reasonable grounds when it makes them and is responsible for showing it did. Its list of things a business shouldn’t do includes “make promises they can’t keep, or make predictions without solid evidence”.

So ask what a forecast is based on and get it in writing. Google’s own suggested questions for any SEO are what results to expect, in what timeframe, and how success is measured. This is general information, not legal advice.

What a fair hybrid looks like

Here is the structure I would want to see. It describes the shape of a fair deal, not anyone’s fees.

1. A fixed fee for defined work. Named deliverables and a reporting rhythm, so slow foundational work gets done whatever Google does this quarter. Our guide to retainer versus project pricing shows how that part is usually built.
2. One bonus metric, chosen for what the SEO controls. Non-brand organic clicks or orders to an agreed set of pages, above a written baseline. Rankings on an agreed buying-intent list can be a secondary check.
3. A baseline in writing. The same months last year, adjusted for known events (sales, stock-outs, price changes, migrations) and for any trend that existed before the agency started.
4. A threshold above normal noise. Check how much your metric moves between ordinary months over the last 12, and set the trigger above that.
5. A cap and a payment rule. A ceiling, quarterly measurement, payment after your returns window closes, and a result that holds for two consecutive months.
6. Guardrails on method. The Google spam policy standard, a documented list of changes, and your approval for off-page work.
7. Your accounts, their roles. You own Search Console, GA4 and Shopify. In Search Console, owners have full control and can add or remove users, full users can view all data and take some actions, and restricted users can mostly only view. Give the agency a full or restricted role, not ownership. On exit, remove them under Users and permissions, and remove any verification token they added, or they can re-verify. Our guide to what to prepare before hiring a Shopify SEO agency covers access.
8. A dispute rule and an exit. Which report breaks a tie, how long either side has to query a month’s numbers, the notice period, what happens to an accrued bonus, and whether the agency keeps earning on pages it built after it leaves, and for how long. Month-to-month terms with no long lock-in make all of this easier to agree.

Want a second view of a quote you have? Tell us about your store.

What to define in writing first

Before any performance element is agreed, get answers to these in writing. An agency that cannot answer one has told you something.

• The metric and data source: one number, one report of record, and who holds admin access to it.
• The attribution model and settings: named, screenshotted, locked.
• Net revenue: after discounts, refunds and returns, excluding tax and shipping.
• The baseline: which period, which source, which seasonal adjustment.
• Brand terms: your list, plus the Search Console filter if available.
• Channel rules: whether AI assistant referrals, direct returns and email-driven searches count.
• Exclusions: migrations, stock-outs, store-wide sales, outages.
• Who ships changes, and how fast: recommendations stuck in a queue deliver nothing.
• Allowed methods, cap, review date, exit and tail, in plain language.

For more warning signs on the agency side, see Shopify SEO agency red flags and how to choose a Shopify SEO agency in Australia.

When a performance element is the wrong tool

My own rule of thumb, not an industry standard: skip the bonus and pay a fixed fee for defined work if any of these is true.

• There is no baseline. A new store, a relaunch or a migration has nothing to measure above.
• Organic orders are in the tens per month. The noise swamps the signal.
• Your tracking is untrusted. If Shopify and GA4 disagree and nobody can explain why, fix that first. A bonus on a broken number is a lottery.
• The calendar is busy. A big sale, migration or rebrand inside the window makes the result say little about SEO.

If tracking is clean, there is a real non-brand baseline and the calendar is quiet, a small capped bonus on top of a fixed fee can make sense.

For what it is worth, our retainers are month to month with no long-term lock-in, we do not guarantee rankings or AI-assistant citations, and we send a progress report every two weeks and a monthly stakeholder report tied to revenue and conversion outcomes, not impressions. I will not set out our fees in a blog post.

Key takeaways

• Performance-based pricing is not a scam by definition. Whatever you pay for gets optimised, so the thing paid on has to be what you actually want.
• Pay per ranking rewards easy keywords and brand terms, and “position” depends on tool, location and search history.
• Revenue share often pays for brand searches and sales you drove yourself. Use a same-month-last-year baseline and count non-brand only.
• The attribution model, GA4 settings and channel definitions are part of the price. Name and lock them.
• On a small store, a trigger set inside the normal month-to-month swing pays for luck.
• A fair hybrid is a fixed fee for defined work, one capped bonus metric, quarterly measurement, your own account ownership and a clear exit.

FAQs

What is performance-based SEO pricing?

Any arrangement where some or all of the SEO fee depends on a result: a keyword reaching a position, traffic or orders hitting a target, or a percentage of organic revenue. Many are hybrids. Always ask what exactly triggers payment and who measures it.

Is performance-based SEO a scam?

Not by definition. Plenty of agencies offer it in good faith. The risks are specific: easy keywords, brand searches and your own promotions can move the metric, and the attribution rule can shift the money. Judge each deal by its definitions, not its label.

Is pay-per-ranking SEO worth it for a Shopify store?

Usually only if you control the keyword list and the bonus favours terms that sell. Left to the agency, a per-ranking fee tends to favour easy keywords. Position also varies by tool, location and search history, so the contract must say how it is checked.

How do I stop an agency being paid for my brand traffic?

Write your brand terms into the contract and measure non-brand queries only. Search Console’s branded queries filter is AI-assisted and can misclassify some queries, and it needs a top-level property with enough query volume. Your own written list covers what it misses.

Can an SEO agency guarantee rankings?

Google’s guidance on hiring an SEO says no one can guarantee a #1 ranking. In Australia the ACCC says claims about future matters need reasonable grounds the business can show. Ask what any forecast rests on. General information, not legal advice.

Want a straight look at your store before you agree a fee?

Any pricing model is only as fair as the numbers behind it. Our Shopify SEO service runs month to month and gives no ranking guarantees. For a first look at your own store, tell us about your store or book a free 30-minute diagnostic. We look at your store before we recommend anything, and we tell you honestly if SEO isn’t the right move. No pitch.