I have read a lot of SEO proposals, mostly because business owners send them to me asking whether they are any good. A pattern shows up quickly.
The weak ones are long. They open with an explanation of what search engines are, include a section on the importance of digital in the modern marketplace, list twenty-five activities in bullet form, and end with a price. You could send that document to any company in any industry and change only the logo.
The strong ones are shorter and more uncomfortable to read, because they contain specific claims about your specific website that could turn out to be wrong.
That is the actual test. A proposal is a prediction, and a prediction you cannot check is not worth paying for. Everything below follows from that.
What a serious proposal contains
Seven things. If more than two are missing, you are looking at a template.
A diagnosis of your site, not a description of SEO. The proposal should tell you something about your website you did not already know. Which pages are competing with each other. Which commercial queries you rank for on page two. What is structurally wrong. If the first three pages are about search engine fundamentals, they did not look at your site before writing.
Named deliverables, not activities. "Ongoing optimization" is an activity. "The commercial litigation service page rewritten, two supporting articles published, internal linking restructured across the practice-area section" is a deliverable. You should finish the document knowing what will exist at the end of month one that does not exist now.
Who does the work. Not the agency's name. The person's. Ask whether the individual on your sales call will be on your account, because frequently they will not be. This pattern is common enough to have a name in vetting circles, the senior-to-junior bait and switch. Junior people do good work under supervision, so this is not automatically disqualifying, but you are entitled to know before you sign rather than after.
What is explicitly out of scope. This is the section that separates honest proposals from optimistic ones, and it is almost always absent. A provider who has genuinely thought about your budget knows what it does not cover and will say so. Its absence usually means nobody did the arithmetic.
How success will be measured, connected to inquiries. Rankings and traffic are intermediate. The measure that matters is qualified conversations. A proposal should say which actions will be tracked as conversions, how attribution will work, and what a realistic figure looks like at three, six, and twelve months.
A timeline with honest expectations. Search work reports back slowly. A credible timeline shows technical fixes landing early, content compounding from around month three, and competitive movement over quarters. Anything faster is a claim about something the provider does not control.
Exit terms and data ownership. Who owns the Analytics property, the Search Console access, the content produced, and the accounts created. What happens to all of it if you leave. This clause is boring right up until the month you need it.
The seven red flags
Some of these are judgment calls. Two of them are not.
1. Guaranteed rankings. This one is absolute. Nobody can guarantee a position, and the people who have been doing this since the 1990s are the most emphatic about it, because they have watched enough algorithm updates to know. A guarantee is either ignorance or a sales tactic, and neither is what you want managing your visibility.
2. Secrecy about method. "Proprietary techniques" that cannot be explained is a phrase covering one of two things: nothing, or something that will get you penalized. Legitimate practitioners will happily explain what they do and why. The methods are not secret. The judgment about what to do first is the actual skill, and that is demonstrable in conversation.
3. Comprehensive SEO at a very low price. Industry consensus puts the floor for meaningful ongoing work at around $500 a month, below which the budget cannot support strategy, content, and technical work simultaneously. If a quote comes in dramatically below the market, the right question is not why are they cheaper, it is what did they remove. Sometimes there is a good answer involving cost structure. Often there is not.
4. Reporting built on vanity metrics. Watch for reports emphasizing keyword counts, raw traffic, or impressions without connecting them to anything commercial. An agency can rank you for five hundred keywords and if none are terms a buyer uses, the rankings are decorative. Ask to see a sample report before signing and check whether a business owner could act on it.
5. Long contracts with no way out. A short initial term makes sense, because meaningful work does take months. A twelve-month lock with no break clause transfers all the risk to you and removes the provider's incentive to keep earning the relationship. Reasonable middle ground is a three to six month initial term with a notice period after.
6. AI SEO expertise with nothing behind it. This has become the most common new claim, and most of it is repackaging. Ask what specifically they do differently for AI answer engines, and listen for whether the answer involves concrete things like structured data, entity clarity, question-led formatting, and citation tracking, or whether it is just the word AI attached to the same service. Ask for a page they have optimized this way and what happened.
7. They quoted without asking about your business. A proposal that arrives after one short call, with no questions about your clients, your margins, your sales process, or which service you actually want more of, is a template. It will be delivered like a template too. A strong provider interviews you before proposing anything, because the plan depends on answers only you have.
The questions worth asking on the call
Five questions, and the way they are answered tells you more than the answers themselves.
"What would you do first, and why that?" Tests whether they have a diagnosis or a checklist. Listen for a reason rooted in your situation.
"What are you not going to do at this budget?" The single most revealing question available. Confidence here is a very good sign.
"Show me something you published in the last month." Not a case study deck. An actual live page. People who ship can produce one immediately.
"What happens if this does not work?" You are testing for a considered answer about diagnosis and adjustment, not a promise that it will.
"What do you need from us?" Every engagement requires client input: approvals, subject expertise, access. A provider who claims to need nothing has either not done this before or is planning to produce content without your expertise in it, which shows.
What honesty looks like in a proposal
The proposals I trust most contain at least one sentence that works against the sale.
Something like: your site has a structural problem that needs solving before content will help, so the first two months will produce very little visible movement. Or: you are in a competitive market and this budget buys progress in a narrow lane rather than across the category. Or, occasionally: based on what you have described, most of your clients come from referral and search may not be your best investment right now.
That last one costs the provider the deal and is the strongest possible signal of competence.
If a proposal contains no caveats at all, that is itself a finding. Real projects have constraints. A document that presents none has been written to close rather than to inform.
Green flags
Worth naming the positives too, because vetting advice tends to be entirely negative.
They asked for access to Search Console and Analytics before quoting, and they actually looked. They named a specific thing that is wrong with your site that you can verify. They set expectations later than you hoped rather than sooner. They distinguished between what they will do and what you will need to do. They were willing to say a competitor is doing something well. They gave you a number for what happens after the initial term, rather than leaving it vague.
None of these are dramatic. That is rather the point. Competence in this field looks unremarkable, and the drama is usually in the sales pitch of people who cannot do the work.
What to negotiate, and what to leave alone
Once you have decided a provider is credible, there is usually room to shape the deal. Knowing which parts are movable saves a lot of awkwardness.
Worth negotiating. The initial term is often flexible, and asking to shorten it from twelve months to three or six is reasonable and frequently granted. The notice period is usually movable. Scope is very movable, and trading breadth for depth is often a better deal than trading price, because a narrower scope done properly beats a wide scope done thinly at the same money. Reporting cadence is almost always negotiable, and reducing it usually buys you more actual work. Payment timing is often flexible for a business with predictable cash flow.
Not worth pushing on. Price itself, usually. A provider who drops fifteen percent because you asked either had margin they were hiding or has just decided to spend fewer hours on you, and the second outcome is more common. If the price is genuinely out of reach, negotiate the scope down instead so the rate per unit of work stays honest. That way you know what you are giving up.
Do not negotiate away. Account ownership, under any circumstances. Your right to a written record of what was done. The exit clause. These exist to protect you in the situation where the relationship has already gone wrong, which is precisely when goodwill will not be available to fall back on.
A note on the discovery fee. Some providers charge for the audit or discovery phase, and buyers often push back on this because free audits are widely offered. I would generally pay it. A paid diagnostic is usually more rigorous than a free one, because a free audit is a sales instrument and its purpose is to produce a reason to hire. If a provider offers the audit free, that is fine too, but read it knowing what it is for.
The larger point is that how someone negotiates tells you how they will behave later. A provider who becomes evasive when you ask to shorten the term is telling you they expect you would want to leave. That is worth learning before you sign rather than after.
Common questions
Should I get multiple proposals?
Two or three. More than that and you spend more time evaluating than the difference is worth. Ask all of them the same five questions above so the comparison is real, since the documents themselves will not be comparable.
Is a cheap proposal always bad?
No. Cost structure varies legitimately, and a provider in a lower-cost market can deliver more hours for the same money without cutting quality. What matters is whether the scope adds up. Count the implied hours against the price and ask directly how it works.
How much should I expect to pay?
Small-business agency retainers generally start around $3,000 to $5,000 a month in 2026, with mid-tier work running $5,000 to $15,000. Meaningful work does happen below that range, but the scope narrows accordingly and a good proposal will say so.
What if I do not understand enough to evaluate the technical parts?
You do not need to. Every test above is about clarity, specificity, and honesty rather than technical knowledge. If you cannot understand what you are buying after asking, that is information about the provider, not about you.
Can I just ask for a trial month?
You can ask, and some will agree. Be aware that one month is genuinely too short to judge search work, so a trial mostly tests communication and delivery rather than results. That is still worth knowing.
A proposal you cannot check is not a plan, it is a brochure with a price on it.
If you have one in front of you and you want an outside read, send it over. I will tell you what I would ask before signing, including when my honest answer is that the other provider is the right call. adphconsulting@gmail.com.