Most advice about SEO proposals tells you what to run away from. That is useful, but it leaves you with a stack of documents and no way to tell which one is actually worth signing. This article is the other half: the positive checklist. Read the proposal in front of you against these nine sections. If a section is missing, that absence is itself a finding — not a blank you fill in later with a phone call.
First, know what you are holding
Almost every other page about SEO proposals is written for the agency, not for you — template posts from Semrush, AgencyAnalytics and SE Ranking teaching agencies how to structure a document that closes. Worth knowing, because the proposal on your desk was probably built from one of them.
That gives you two reference points. AgencyAnalytics tells agencies an effective proposal runs 8 to 15 pages with two or three budget tiers, so a two-page quote is unusually thin and a forty-page deck is padding. And Semrush’s agency-pitch research scored a client-tailored strategy the single most important element, 74 out of 100 across the brands it interviewed — so an agency that sent you a template anyway did it knowing better.
You are not grading an essay. You are looking for the parts of the document that would let you, six months from now, say “you said you would do this, and you did not.” A proposal with no checkable claims cannot be broken, which sounds good until you realise it also cannot be enforced.
1. Findings from your actual site
The first section should prove they looked at your website before they priced it.
That means specifics: pages named, problems described, your real competitors identified by name. “Your service pages have no unique title tags — all eleven use the same one” is a finding. “In today’s competitive digital landscape, businesses need a strong online presence” is not, and if the document opens that way you should expect the rest of it to be a template.
The tell is simple. Search the document for your business name. If it appears only in the header and the price table, nobody looked. This is RF-08, generic template with no site-specific findings, and it matters more than it sounds like it should, because everything downstream depends on it. The keyword list, the mix of deliverables, and the monthly price were all chosen before anyone had evidence about your business. You are being quoted for a generic company of roughly your size.
A stronger test: think of the thing you already know is wrong with your site — the slow product pages, the location page that never got built, the blog that stopped in 2023. Does the proposal mention it? If they missed the problem you already know about, be sceptical about the ones they claim to have found.
2. Deliverables with quantities, cadence, and a definition of done
This is where most proposals quietly fail.
Every line item needs three things: how many, how often, and what “finished” means. Compare these two versions of the same line:
- “Ongoing content optimization and creation”
- “Four articles per month, 1,200+ words, topics agreed with you in advance, written and published to your site, with internal links added to the relevant service pages”
The second one can be audited at the end of the month by counting. The first cannot be audited at all, which is the point of writing it that way. Vague deliverables (RF-02) are also how a retainer shrinks after month three without anybody having to announce it.
Watch for the phrase “up to.” “Up to 15 backlinks per month” commits the agency to nothing — zero satisfies “up to.” Ask for the floor instead of the ceiling, in writing.
Technical work deserves the same treatment. “Technical SEO improvements” should become a named list — the duplicate title tags on the service pages, the missing schema on the location pages, the 34 broken internal links — with a month attached to each.
3. A timeline that says what happens in month one
You want a schedule with the first 30, 60, and 90 days broken out, because that is the window where you can still cheaply decide you were wrong.
Month one in an honest proposal is unglamorous: access and tracking setup, the technical audit, keyword and competitor research, fixing whatever is actively broken. If month one promises ranking improvements, that is a sales document rather than a plan. Google’s own guidance is that site changes take four months to a year to have an effect, so a proposal forecasting results in weeks disagrees with the search engine it is selling access to.
The timeline also tells you when to expect the first thing you can hold in your hands. Ask for that date explicitly. A setup fee with no artefact attached to it (RF-11) is a deterrent against cancelling, not a service — if you are paying an onboarding or strategy fee, the document should name what you receive for it and confirm you keep it whether or not you continue.
4. KPIs tied to leads, not rankings
The reporting section is where you find out what the agency intends to be judged on.
Keyword rankings, impressions, and domain authority are inputs at best — and domain authority is a third-party vendor’s score Google does not use at all. You can rank better on all three and take no more phone calls. Plenty of retainers survive for years on exactly that gap.
What you want named in the document:
- Organic sessions, from your own Google Analytics
- Conversions you actually care about: calls, form submissions, bookings, quote requests, orders
- Revenue or lead value where it can reasonably be tracked
- The work done that month, listed against the deliverables in section 2
Reporting that never reaches calls, forms, or sales is RF-06. It is a soft flag on paper and an expensive one in practice.
One more thing to check: which keywords are being reported on, and when were they chosen? Keywords selected after the fact let anyone show progress. Agree the tracked set at the start and keep it fixed.
The single best clause you can ask for
Ask that the monthly report include the previous month’s plan next to what was delivered. It costs an honest agency nothing and it is the one habit that makes everything else in this list enforceable.
5. Who actually does the work
The proposal should name people. Not “our team of 50 specialists” — a name for whoever runs your account, and an honest answer about whether the writing, the link work, or the technical implementation is subcontracted.
Subcontracting is not a problem. Undisclosed subcontracting is (RF-15), because the person accountable for your account, the quality of the work, and the handling of your data are all different from what you were sold, and you had no chance to assess any of it. The fair version is a sentence: here is who does what, here is where, here is what we subcontract.
Do the arithmetic while you are here: if a proposal claims a large team at a price that could not cover a fraction of it, one of those numbers is decoration.
6. Ownership of everything, stated plainly
This is the most expensive paragraph in the document and it is usually the shortest.
The proposal — or the terms attached to it — should say in plain words that you own, in your own accounts, from day one:
- Your domain, registered in your name at your registrar
- Your hosting and your website
- Your Google Analytics and Search Console properties
- Your Google Business Profile, with you as owner
- Any content written for you, on publication
The agency gets access. Access is revocable; ownership is not. Where a proposal says content “remains the property of the agency until” something, or that they will register the domain or create the Business Profile “on your behalf,” you are looking at RF-09 and RF-14. These are the clauses that make a long contract enforceable in practice: leaving means losing assets you paid for. A dispute about an invoice should never become a website that is offline.
7. Exit terms you would accept on a bad day
Read the termination section as though the work has gone badly. That is the only situation in which you will read it again.
What a fair set of terms looks like:
- An initial term of around three months to cover setup, then month-to-month
- Thirty days’ written notice, by either party, with no penalty
- No automatic renewal — or renewal into month-to-month, with a reminder sent before the date
- A named handover on exit: accounts, content, data, and logins returned within a stated window
A twelve-month minimum (RF-04) transfers all the risk to you. A silent auto-renew (RF-05) is worse, because the notice window usually closes before you would think to cancel. No cancellation clause at all (RF-13) is worst, because silence is not neutral — you would be negotiating your exit from zero while they hold your accounts.
When you see the long term, the silent renewal, and the weak exit together in one document, the combination is the finding, not the individual clauses.
8. Proof you can actually check
Every proposal template tells agencies to include case studies, so almost every proposal has them — which makes the section worthless as a signal unless you test it.
A case study is checkable when it names four things: the client, the starting point, the time it took, and the outcome in a unit you care about. “Increased organic traffic 412%” names none of them — a site going from 60 visits to 307 is a 412% increase. “A three-location dental practice, from 40 to 130 booked appointments a month from organic search, over eleven months” is a claim someone could dispute.
Three quick tests. Does at least one case study come from a business roughly your size, in a category with similar competition? Do the named clients still exist? And will they give you that client’s phone number — a case study you cannot follow up on is marketing copy with a chart in it.
If the section is anonymous throughout — “a leading home services brand” — it is not proof. It may be a genuine NDA, or it may be someone else’s result. You cannot tell, which is the point.
9. A price you can compare
The pricing page should let you reconstruct the number, not just read it: the monthly fee, every one-off fee named separately (setup, onboarding, migration, platform or dashboard charges), what is explicitly not included, and the total over the minimum term. Multiply it out yourself. A $900 quote with a $1,500 setup fee and a twelve-month minimum is a $12,300 decision, and that is the number to compare against the other two proposals.
Two patterns to watch. “From $X” is not a price; ask for the price for your scope. And where three tiers are offered, the bottom one is often built to be unworkable so the middle one reads as sensible — normal sales design rather than a flag, but it means the cheap tier is not the bargain it looks like.
Ranges vary widely by market and scope; what SEO should cost a small business has the published survey figures and their caveats.
What a good proposal is allowed to do
To keep this fair: a proposal is not suspect for declining to guarantee rankings, asking for a three-month initial term, charging a real setup fee with a real deliverable, disclosing that it subcontracts writing, or pricing at the top of the range with relevant case studies behind it. Firmness is not a flag. Vagueness is.
One last thing, which no document will contain: anything the salesperson told you out loud that is not written down. Promises about a start date, a senior person, a keyword, or a price are not deliverables unless they appear in the file. Ask for them to be added before you sign — how they react tells you more than the proposal does.
Before you sign
Print the proposal. Write the section number in the margin where you find each of the nine above. The gaps are your negotiation list — most are fixable before signing and almost none afterwards.
If you would rather have someone else do that pass, order a proposal audit — a fixed-fee, written review of the document you are holding, with the exact clauses quoted and a plain recommendation.
Sources
- AgencyAnalytics, “SEO Proposal: How to Create a Winning SEO Proposal” — the 8-15 page and budget-tier conventions agencies are taught.
- Semrush, “SEO Proposal” — Semrush’s agency-pitch research scoring a client-tailored strategy 74/100, the highest-ranked of ten pitch elements.
- Google Search Central, “Do you need an SEO?” — Google’s own guidance that SEO changes typically take four months to a year to take effect, and its warning-sign list for hiring.