Three questions that are really one. Nobody asks what it costs without meaning is it worth it, and that cannot be answered without knowing how long it takes.
What SEO costs depends on which of three very different things you are buying. Fixing what is broken is a project, and it is the cheapest and most certain part. Publishing is a continuing cost with a variable return. Earning links is the most expensive per unit and the slowest. Most disappointing engagements are somebody paying for the third while the first is still unfixed.
Four arrangements, and the useful comparison between them is not price. It is what each one is genuinely good at and where each one predictably disappoints.
One experienced person, part time, usually on a monthly retainer or a day rate.
Good for. A specific problem with a known shape: a migration, a technical audit, a recovery. You are buying judgement and you can check whose.
Predictably disappoints at. Volume. One person does not produce a content programme, and a retainer that quietly becomes one is the most common way this arrangement disappoints.
A team, a plan and a monthly report. Usually a minimum term.
Good for. Breadth and continuity. Somebody is responsible when the person who knew your site leaves, and the work continues in the month you are busy.
Predictably disappoints at. Transparency. Almost nobody in this market publishes a price, which means the first thing you learn about a supplier is how they negotiate.
A salary, plus tools, plus the time of everybody the role has to ask for changes.
Good for. Sites where SEO is a product decision rather than a marketing one, and where the work is continuous rather than a project.
Predictably disappoints at. Ramp and breadth. One hire covers one skill set, and the technical, editorial and outreach halves of this are genuinely different jobs.
Software, and the hours of somebody who already works there.
Good for. Small sites, early sites, and any site where nobody has yet checked whether the fundamentals are broken. This is where the free tools on this site are aimed.
Predictably disappoints at. Anything needing authority rather than fixes. No tool subscription earns a referring domain, and that is the constraint on most sites that have already done the mechanical work.
We do not publish a market rate on this page and it is not an oversight. Where figures about other companies exist here, they carry a source and the date they were read, and they live on what this market actually charges, where 4 of the 23 suppliers we surveyed published anything about their own prices. Ours are on the pricing page, in numbers, with no call required.
The honest timeline, in the order things actually happen. The dates are ranges rather than promises, and the last one has no date at all because nobody can give it one truthfully.
What is broken, what is missing, and what is fine. This part is fast, largely automatable and should not be a month. If a proposal has a discovery phase measured in months, ask what is being discovered.
Access, indexability, canonicals, rendering, dates and attribution. Almost all of it is template work, so it is cheap per page and it is the only part of this with anything resembling a guaranteed effect: a page nothing can reach cannot rank at any budget.
New pages, and the usually better bet of improving pages that already have some history. This is where cost becomes ongoing rather than one off, and where a plan either has a reason for each page or does not.
Crawling, indexing and re-evaluation all take time, and search reporting lags. Anything before this is noise dressed as a trend, in both directions. This is the point at which a programme can be judged rather than defended.
Being linked to and cited by other people. It is the least controllable input, it moves the slowest, and on most sites it is the actual constraint once the mechanical work is done. Nobody can put a date on this one honestly.
Four numbers you already have, and one you can find out for free. No industry benchmark required.
Take the value of one customer to you over the time they stay. Multiply bythe share of enquiries that become customers. That gives you what one enquiry is worth. Divide the monthly cost you are considering by that number, and you have the only figure that matters: how many extra enquiries a month this has to produce to break even.
Now compare that against what is realistically available. If breaking even needs forty extra enquiries a month and the entire search demand for what you sell is a few hundred people a month, the arithmetic has answered you and no supplier can argue with it. If it needs two, the decision is easy and the remaining question is only whether this particular supplier can produce them.
The number most people are missing is the last one, and it is the cheapest to get. Real search volume for what you sell comes out of advertising platforms, it is bought once rather than subscribed to, and the method for getting it honestly is a morning of work. Doing that before signing anything is the single highest return hour in this entire process.
One adjustment worth making, because it did not exist five years ago. On a growing share of queries the answer now appears above the results and nobody clicks anything, and that lands hardest on short definitional questions. Forecast those low. Comparative, situational and anything with a purchase behind it are largely untouched, which is where the arithmetic above still holds.
Four situations where the honest answer is no, or not yet.
A genuinely new category has no demand to capture, and search is a demand capture channel rather than a demand creation one. This is a research question with a cheap answer, and it is answered before the budget conversation rather than after it.
The first honest read on a programme is three to six months out. If the business needs revenue sooner than that, paid channels buy time and SEO runs alongside them. Anybody selling you a faster timeline is selling a hope with an invoice attached.
Most of the certain half of this work is template changes. If there is no engineering time and no way to get any, a strategy document arrives and nothing is implemented, which is the most expensive possible outcome because it looks like progress.
Authority decides this and it is not negotiable by writing better. A site with almost no referring domains competing against established publishers on a head term will not place, however good the page is. Build where the competition is beatable and treat authority as a separate workstream.
It changes every subsequent conversation, because it turns a proposal about everything into a proposal about things you already know are wrong.
It depends far more on what you are buying than on the market rate. A one off technical audit and fixes is a project cost. A content programme is a per month cost that continues. Link acquisition is the most expensive per unit of anything in this category. The most useful question is not the price but what is included and what happens when it stops.
The technical fixes can take effect within weeks, because they are removing an obstacle rather than building an asset. Content and authority take months, and the first genuinely readable result is usually three to six months in, because crawling, indexing and reporting all lag. Anybody promising a timeline shorter than that is describing a hope.
It is worth it when people are already searching for what you sell and you can win some of those searches from where you stand. It is not worth it when demand does not exist yet, when the searches are dominated by publishers you cannot outrank, or when the fundamentals of the site are broken and nobody will fix them. The first is a research question and the third is checkable in minutes.
Because the work is variable and because a published price is a negotiating position you have given away. The consequence for a buyer is that the first thing you learn about a supplier is how they handle a pricing conversation, which is a poor proxy for how they handle a site. When we surveyed this market, a small minority of the agencies in it published anything about their own rates.
The cheap version of the technical half is genuinely worth doing and largely free: checking crawler access, indexability, canonicals and rendering costs nothing and settles whether anything else can work. The cheap version of the content and links half is where the money is wasted, because both are labour and both fail badly when done at volume without judgement.
Do the diagnosis yourself first, whatever you decide afterwards. It is fast, it is free, and it changes the conversation with any supplier from a proposal about everything to a proposal about the things you already know are wrong. Buy the parts that need either continuous attention or a skill nobody at your company has.
Three questions. What specifically will be different in ninety days. What happens to the work if we stop paying, which separates assets from rentals. And what would make you tell us this is not worth continuing, which is the one nobody prepares for and the answer is the most informative.
Free account, no card. Paste your URL and get a real, scored report of your AI and search visibility.
Measuring rankings in