How to choose an SEO agency in 2026
A buyer's guide to hiring an SEO agency: what the work is, what it should cost, which proposal lines matter, and the questions that separate two quotes.
Choose an SEO agency on three things: what it will ship each month in writing, who inside the agency will do that work, and what it costs. Everything else in a pitch is decoration. An agency that cannot answer all three in one page is asking you to buy activity rather than outcomes.
The rest of this guide is the long version of that paragraph: what the work is, what it should cost, how to read a proposal, the red flags, and the questions that separate two agencies quoting the same number.
في هذه الصفحة
What does an SEO agency actually do for the money?
An SEO agency does four things, and every retainer is some mix of them: it fixes what stops search engines using your site, it builds pages that answer what your buyers search for, it earns mentions and links from other sites, and it measures which of that produced money. Anything that does not fall into one of those four is overhead being billed to you. Our own SEO services page names them in the same order for the same reason.
The mix matters more than the label. A technical-heavy month is developer time and QA; a content-heavy month is research, writing, editing and publishing; an authority-heavy month is outreach, pitching and waiting. Agencies rarely say which mix you are buying, which is why two quotes at the same price can describe completely different months.
There is a fifth thing now, and it is not optional: being the answer when somebody asks an assistant rather than a search box. That work overlaps with SEO but is not identical to it, and it is worth reading generative engine optimization before you accept a proposal that treats the two as the same line item.
The reason to insist on the mix is that the four kinds of work fail in different ways and on different timescales. Technical work is the only one that can lose you traffic overnight, and the only one where a single mistake by a developer undoes a quarter. Content work almost never fails fast; it fails quietly, by producing pages that are fine and that nobody needed. Authority work fails by producing nothing at all for two months and then producing everything at once, which makes it the hardest line in a retainer to defend halfway through. Measurement fails by being correct and useless — dashboards full of numbers that no decision depends on.
A good agency will tell you which of the four your site is worst at before it tells you what it charges, because that is the diagnosis and everything else is the prescription. If the first call is entirely about packages, you are being sold from a menu rather than assessed.
| Work | What it looks like in a month | What you should see | How fast it moves |
|---|---|---|---|
| Technical | Crawl, diagnose, ship fixes with the developers | A list of shipped changes, not a list of findings | Weeks |
| Content | Research, brief, write, edit, publish, refresh | Published URLs with named authors | Months |
| Authority | Outreach, digital PR, earned mentions | Named placements and who pitched them | Months to quarters |
| Measurement | Analytics, search console, reporting | A paragraph explaining the numbers | Continuous |
What should an SEO agency cost?
Almost nobody publishes a price, which is the first thing to understand about the market. Of the ten best-known global SEO agencies we reviewed, two publish a number on their own website: WebFX prints its tiers openly [2], and Omniscient Digital states a starting figure for full-service engagements in plain text on its homepage [1]. The other eight put a form in front of the question, and the figures those two publish are set out in how much does SEO cost.
That opacity is not a conspiracy, it is a negotiating position. It also means you cannot compare quotes by price alone, because the scopes behind them are invisible until you ask. The practical move is to write your own scope first — what you want shipped each month — and ask three agencies to price it. Then the quotes are comparable, because they are quotes for the same thing.
2 of 10
For what the bands buy and where the money goes, how much does SEO cost takes the question apart properly. Our own plans carry their starting prices on the pricing page, which we did for the same reason we are telling you to demand a scope: a number you can read is a number you can argue with.
One warning about the cheap end. Under every retainer there is a payroll, and the smallest quotes are the ones where the payroll no longer fits inside the fee. What gets removed is never the account management, because that is the part you see; it is the writing and the outreach, because those are the parts you only notice missing a quarter later. A quote that is half the others is usually not a better deal, it is a narrower product sold under the same name.
The expensive end has its own failure. A large retainer with a thin scope is the most common way to waste money in this industry, because the size of the invoice creates an assumption that something substantial is happening. Ask what the extra money buys in units — more pages, more outreach, more senior hours — and if nobody can answer in units, it buys nothing.
How do you read an SEO proposal?
A proposal is a promise about months, so read it as a calendar rather than a brochure. The test is simple: can you turn the document into a list of things that will exist by a date, each with a name next to it? If you cannot, there is nothing in it to hold anyone to.
Three lines carry almost all the information. The deliverables line tells you what is made. The ownership line tells you whether a senior person touches your account after the pitch. The reporting line tells you whether you will find out when something goes wrong, or only at the end of the quarter.
Everything else is context. The case studies tell you what the agency is proud of, not what it will do for you. The tooling list tells you what it pays for, not what it produces. The methodology diagram tells you how the agency thinks about itself. None of that is worthless, but none of it is a commitment, and it is remarkable how many proposals are made entirely of it.
A useful exercise before you sign anything: print the proposal and strike out every sentence that would still be true if it had been sent to a different company. What survives is the actual offer. On a good proposal, a third of the page survives. On a bad one, the survivors are the price and the logo.
Turn every line into a noun with a date
"Improve technical health" is not a deliverable. "Ship the crawl fixes from the audit by 31 October" is. Rewrite the proposal yourself in that form and send it back; the answer tells you a great deal.
Ask who does the work
Name the person writing the pages and the person doing outreach. If the answer is a team or a title, ask again. The gap between the strategist in the pitch and the junior on the account is the most common disappointment in this industry.
Ask what happens in month four
Most proposals describe month one in detail — audit, setup, quick wins — and go quiet after that. Month four is where the retainer is actually spent, so it is the month worth arguing about.
Ask what stops
Every plan drops something. Find out whether it is the writing, the outreach or the technical time, because whichever it is, that is the part of the promise that will not happen.
What are the red flags?
Three claims are reliable disqualifiers, and they are easy to hear in a first call.
The first is a guarantee of rankings. Nobody controls a search result, so the only way to guarantee one is to pick a term nobody searches. Ask which term is being guaranteed and the conversation usually ends there.
The second is a method that cannot be described. Real SEO work is boring and specific: these pages, those fixes, that outreach. A proprietary system nobody will explain is either ordinary work with a trademark on it or something you would not agree to if it were explained.
The third is volume as a strategy. Google's spam policies treat pages made mainly to manipulate rankings as scaled content abuse regardless of how they were produced, and its 2026 updates acted on that at the level of whole domains rather than individual pages [3]. An agency whose plan only adds up if it publishes dozens of unreviewed pages a month is buying that risk with your budget.
How do you compare two agencies fairly?
Put the two proposals side by side and delete the price from both. Almost everyone reads price first and then rationalises, and the rationalising is where the mistake happens. Without the number, you are comparing what you would actually receive.
Then compare on four axes and nothing else: what ships, who ships it, when, and how you will know. Score each proposal out of three on each axis, and write one sentence explaining every score. The sentences are the point; a score you cannot justify in a sentence is a preference, and preferences in this decision are usually about who was warmer on the call.
Only then put the prices back. What you usually find is that the cheaper proposal is cheaper because it contains less, which is fine if the less is the part you did not need. What you occasionally find is that the cheaper proposal contains the same work, in which case the expensive one is selling access to a brand, and you can decide whether that is worth the difference.
How long before it works?
Expect three to six months before rankings move on terms that are not heavily contested, and closer to nine to twelve before you can attribute revenue rather than traffic. That is not a hedge; it is how long it takes for pages to be crawled, assessed, re-assessed and then chosen over the pages that are already there.
Two things change the timeline honestly. A site with technical problems can move quickly at first, because fixing what blocked indexing releases rankings that were already earned. A site in a crowded market moves slowly regardless of budget, because everyone ahead of it is also working.
What should not change the timeline is the size of the retainer. Paying twice as much does not make Google decide twice as fast. It buys more parallel work, which shortens the calendar at the end, not the beginning.
| Quarter | What is happening | What you can judge |
|---|---|---|
| 1 | Audit, fixes shipped, first pages published, tracking corrected | Whether things ship on time |
| 2 | Content compounding, first outreach landing, early impressions | Whether impressions and rankings are rising |
| 3 | Rankings on mid-competition terms, traffic becoming meaningful | Whether traffic is the right traffic |
| 4 | Revenue attribution, refreshes, second wave of pages | Whether the programme pays for itself |
Does it matter where the agency is?
Less than it used to, and more than agencies admit. Time zones and language matter; postcodes mostly do not. If your customers search in a language your agency does not write, that is a real problem and no amount of process fixes it. If your agency is three hours away rather than three floors away, that is a scheduling detail.
Where location does matter is evidence. An agency selling into a market it has never worked in will produce pages that read like they were written from a search result, because they were. Ask for something specific about your market that you did not already know. The answer separates research from repetition fast.
That is also the honest use of a city-level list: it tells you who is actually present. We publish one for our own market — the best SEO agencies in Riyadh — with the methodology and our own bias written into it, because the alternative lists are advertisements that do not say so.
Who should you hire: an agency, a freelancer or an employee?
Each is a different trade between cost, control and range. A freelancer is cheap and deep in one skill; an employee is expensive, wholly yours, and limited to what one person can do; an agency is a team you rent, with the overhead that implies. There is no correct answer, only the one that matches the work in front of you.
The deciding question is usually range. A site that needs technical fixes, bilingual writing and outreach in the same quarter needs three skills, and three skills is more than one person. A site that needs one thing done well needs one person who does that thing.
We took this apart properly in SEO agency vs in-house vs freelancer, including the arithmetic on what an in-house hire actually costs once you add the tools and the time to manage them.
The best question in a first call is not what will you do — it is what will you not do. Every plan drops something, and the agencies worth hiring will tell you what they dropped and why before you ask a second time.
How should an agency report?
A report answers four questions in order: what did you ship, what moved, what did it cost, and what happens next. A dashboard is not a report. A dashboard with a paragraph under it is a report, and the paragraph is the part you are paying for.
Rankings alone are the weakest possible reporting, because they are the easiest number to move without moving revenue. Impressions, clicks, assisted conversions and the specific pages driving them are harder to fake and harder to misread. If an agency cannot connect its work to something that happened in the business, it is reporting on itself.
Ask for the reporting format before you sign, not after. It is the cheapest thing to change in a negotiation and the hardest to change once it has become a habit.
Ask also for the bad-news mechanism. Every account has a month where something breaks, a page drops, or a plan does not work, and what separates agencies is whether you hear about it from them in week two or from your own dashboard in month three. The answer you want is a named person and a channel, not a promise of transparency.
One more thing worth writing into the reporting clause: the numbers should come from your accounts, not from the agency's tools. Search Console and your analytics belong to you and cannot be presented selectively. A report built only on a third-party rank tracker is a report built on a source the agency chose.
What should the contract say?
Three clauses matter. The term, the notice period and who owns what.
On term: an initial period is reasonable, because outreach and content take months to land and an agency that can be dropped in week six cannot plan. A mandatory twelve months with no exit is not the same thing; it is the agency managing its own risk with your budget. Ours are month to month and you can stop at the end of any month, which we can offer because the work is designed to be worth renewing rather than worth locking in.
On ownership: the pages, the accounts, the tracking and the data are yours. Say so in writing. The most expensive version of this mistake is discovering that the analytics property and the search console access belong to an agency you have just left.
On scope change: agree how a scope changes mid-retainer before you need it to. Search results move, priorities move, and a contract with no mechanism for that produces either resentment or invoices.
There is a fourth clause worth adding even though it rarely appears: what happens to work in flight when the relationship ends. Outreach pitched and not yet landed, drafts written and not yet published, fixes specified and not yet shipped. Say plainly that the drafts and the specifications come with you. It costs nothing to agree at the start and is an argument at the end.
What should you ask in the first call?
Five questions, in this order, and the order matters because each one makes the next one harder to dodge.
What would you do first, and why that?
You are testing whether they looked at your site before the call. The answer should be specific enough that it could not apply to another company.
Who does the work, by name?
Then ask what else that person is working on. An account handled by someone carrying eleven others will be handled accordingly.
What does month four look like?
Month one sells itself. Month four is the product.
What have you stopped doing since 2024, and why?
Search changed hard in 2026. An agency whose answer is nothing has not been paying attention.
What would make you tell me to stop paying you?
Everyone claims to be a partner. This is the question that finds out.
If you want a neutral starting point before any of those calls, get an SEO audit done — or run the free one on your own site first. Walking into a pitch already knowing what is broken changes the conversation from a sales meeting into a scoping meeting, which is the meeting you actually wanted.
أسئلة يطرحها الناس
How long should an SEO contract be?
Long enough to do the work and short enough to leave. Six months is a reasonable minimum for a programme that includes writing and outreach, but a rolling month-to-month agreement after an initial period is better for you and costs a serious agency nothing.
Is it worth hiring an agency if I only want technical fixes?
Usually not on a retainer. A defined technical problem is a project: scope it, pay a fixed fee, take the report and the fixes, and stop. Retainers are for work that has to run every week to compound.
What is a realistic timeline before SEO shows results?
Three to six months for movement on terms with little competition, closer to nine to twelve before revenue is attributable. Anyone naming a shorter timeline is describing an outcome they cannot control.
Should I pick the agency that promises the most?
No. Promises are free and the ones that sound best are the ones nobody can keep. Pick the agency whose proposal names what it will ship, who will ship it and when, because that is the only part you can hold anyone to.
What does a good monthly report look like?
What was shipped, what moved, what it cost and what happens next, in that order. Rankings alone are not a report. A dashboard nobody writes a paragraph under is not a report either.
المصادر
- Let's work together — full-service engagements start at $10,000 a month — Omniscient Digital (checked 11 September 2026)
- SEO pricing — WebFX (checked 11 September 2026)
- Spam policies for Google web search — Google Search Central (checked 11 September 2026)
- Creating helpful, reliable, people-first content — Google Search Central (checked 11 September 2026)
See what is actually wrong with your site before anyone quotes you
احجز استشارة مجانية مدتها 30 دقيقة مع إبراهيم وجيريمي. ننظر في موقعك أثناء الحديث، وتخرج بخطة مرتّبة حسب الأولوية.
من كتب هذا
Ibrahem HelalCo-founder
Leads strategy and delivery.
يعمل على: SEO, GEO
راجعه قبل النشر Jeremy Zava، Co-founder.
المزيد عن هذا
- SEO agency vs in-house vs freelancerThree ways to buy SEO, compared on cost, range, speed and risk — with the arithmetic that decides which one fits the work actually in front of you.
- Best SEO agencies in Riyadh (2026)Eleven agencies ranking for SEO in Riyadh, reviewed against the same five criteria, with what each one publishes, what it does not, and where we rank ourselves.
- How much does SEO cost?Published benchmarks put SEO at $1,000 to $5,000 a month, $75 to $200 an hour and $500 to $5,000 a project. What sits behind each band and what moves it.
- SEO Agency AlternativesSix real alternatives to a full-service SEO retainer — in-house hires, freelancers, fractional help, software, boutiques and resellers — and how to pick.