Almost every page about SEO pricing quotes the same industry averages, and those averages trace back to a small number of surveys that are hard to verify and harder to apply to your situation. We have taken a different approach here: rather than cite a benchmark, we went and read what agencies publish on their own websites, and reproduced it with the source attached.
This is a smaller sample than a survey and a more reliable one. Every figure below was on the agency’s own site on 19 October 2026. Nothing is estimated, and where an agency publishes guidance rather than a price, the table says so.
Why so few agencies publish anything
Two reasons, one defensible and one not. The defensible one is that SEO scope genuinely varies enormously. Fixing a crawl problem on a sixty-page site and building topical authority in a contested category are different amounts of work, and a single number covering both would mislead everyone.
The less defensible reason is that an unpublished price lets the number move with the buyer. If an agency learns your funding stage before it learns your technical situation, you should at least notice that the discovery call collected those facts in that order.
The practical consequence is that you cannot comparison-shop on price, and attempting to usually produces a worse decision than comparing on scope. Which is why the table above is useful mainly as a sense of range, not as a shortlist.
What actually drives the number
The state of the foundation. A site with crawl and indexation problems needs technical work before content work has any leverage. That is a defined, finite job, and it is often the largest single line in a first quarter.
Competitive density. Winning a query that three well-funded companies are actively defending costs more than winning one nobody has noticed. This is the variable buyers most consistently underestimate.
Content volume and seniority. The gap between a junior writer producing eight pieces a month and a senior strategist approving every brief is most of the price difference between two agencies describing the same service.
Scale of the surface. A forty-thousand-SKU catalogue, a multi-location service business and a twelve-page SaaS site are not comparable amounts of work even when the strategy is identical.
Whether implementation is included. Agencies that hand over recommendations cost less than agencies that ship the changes. The cheaper one is only cheaper if you have the engineering capacity to act on it.
The pricing models, and what each one is really saying
Monthly retainer is the default. It suits ongoing work and it is the easiest to compare. Its weakness is that it has no natural end, which is why so many retainers outlive their usefulness.
Published floor with a stated term, like six or twelve month programmes, is the most honest version of the retainer. It admits that SEO does not produce in eight weeks and asks you to commit to the horizon that makes the work sensible.
Annual pricing filters hard, deliberately. Publishing a six-figure annual floor tells most of the market not to call, which saves everyone a conversation.
Percentage of revenue or funding is the most unusual model in the set. It scales with the company rather than with hours, which aligns incentives neatly and requires you to disclose figures you may not want to share during scoping.
Project pricing suits a defined job with an end: an audit, a migration, a technical remediation. If what you need is finite, paying for it monthly is the wrong shape.
What the bottom of the market actually buys
There is a real floor below which the economics stop working, and it is worth understanding rather than moralising about. A few hundred pounds or dollars a month buys a small number of hours. Those hours can be genuinely well spent on a narrow job — citation cleanup for a local business is a good example, and one agency in the table prices exactly that way, honestly.
What the same budget cannot buy is strategy, implementation and content simultaneously. When an agency offers all three at that price, the arithmetic only works if the work is templated and largely unsupervised, which is how sites end up with forty thin pages and a ranking report that looks busy.
The question to ask is not whether a price is low, but what it is narrow about. A cheap, honest, narrow scope is a perfectly good purchase. A cheap, broad scope is arithmetic that does not close.
What we charge, and why it is not on this page
We do not publish a rate, and it would be dishonest to write a page like this without saying so plainly. Several agencies in the table above do publish, and that is a real advantage they hold over us: it respects your time and it filters before the call.
Our reasoning is that the honest number depends on what the review finds, and we would rather give you a scoped figure in 48 hours than a range wide enough to be meaningless. You can judge whether that is a good reason or a convenient one. We think the published-price agencies have the better of the argument more often than we would like.
What we will commit to in public: the review is free, the reply comes within 48 hours, the figure arrives with the work and the sequence attached, and you will not be sold a retainer for a job that is actually a project.