What an SEO proposal should contain, and what a template gives away

Search for this phrase and most of what comes back is templates, which tells you who is looking: providers who need to produce a document, more often than businesses trying to read one. That is worth saying at the start because it explains why so many proposals look alike and say so little. A proposal assembled from a template is recognizable in about thirty seconds, and the tell is not the design. It is that nothing in it could only have been written about your business.

Quick answer

• A real proposal quotes your own data back at you. A template quotes the industry.
• Deliverable counts measure activity, not outcome. Four articles and twelve links is a shopping list.
• Month one should produce almost nothing visible, and a good proposal says so.
• The absence of an exit condition is the most common omission and the most revealing.

What should an SEO proposal include?

A diagnosis built from the site's own Search Console data, the scope stated in hours and by whom, a month-by-month sequence in which month one produces nothing visible, a measurement commitment on a fixed set of queries, and the conditions under which the provider would recommend stopping. Anything written without the data is a template.

The test that takes thirty seconds

Read the proposal and mark every sentence that could not have been written about any other business in the same industry.

On a template the count is near zero: the company name, maybe the city, and a screenshot. On a proposal written after looking at your data, the count is high, because it references your queries, your pages that rank without content behind them, the technical problem found in the crawl.

That is the whole test, and it works without knowing anything about SEO. You are not evaluating the answers, you are evaluating whether anybody looked.

What each section should actually contain

The diagnosis, which comes before the price

A provider who quotes without asking for Search Console access is proposing work without having seen the demand. That is not fraud, it is guessing, and the price attached to a guess is arbitrary in both directions.

What a real diagnosis shows: which queries the site already appears for, which of them have impressions and no clicks, which pages are not indexed, and what that implies about where the first months should go. The questions to ask if this section is missing are in how to choose an SEO agency.

The scope, in hours

SEO is almost entirely skilled time. There is no raw material and no inventory, so a price implies a number of hours at a level of seniority, and a proposal that avoids stating either is avoiding the only arithmetic the buyer can do.

A list of activities is not scope. "Content optimization, technical improvements and link building" describes categories, not commitments.

The timeline, including the boring month

The correct answer for month one usually involves auditing and produces nothing visible, and a proposal that says so is more credible than one promising early wins.

What should be there instead of promises is a sequence: what gets fixed first, what gets written second, and roughly when the data should start moving. The shape of that curve, with a real seven-month progression, is in the Spanish guide on how long SEO takes.

The measurement, with a fixed cohort

The section that separates measurement from decoration. A report built on a set of queries that changes every month cannot show progress, because each month describes something different.

The commitment worth extracting: the same queries, every month, with position and clicks, and a note of what was decided. Everything else in a report is illustration.

The exit condition, which is almost never there

An honest proposal contains the circumstances under which the provider would tell you to stop paying: the demand is not there, the site cannot convert, the vertical is priced beyond the budget.

Its absence is the most informative thing in most proposals. A provider who cannot imagine the engagement failing has not thought about the mechanism, and a provider who has thought about it and left it out has made a decision about what you should know.

Two things that should not be in it

A guaranteed position or timeline. Nobody controls the ranking algorithm, and 2026 alone had four confirmed updates, with the March one changing position for 79.5 % of the top three. A guarantee is not demonstrable, which is precisely why it is the standard pitch of providers who do not intend to deliver.

A request for ownership access to your Google properties. Restricted access reads every report a provider needs. Ownership can add and remove other owners, and it is easy to grant and unpleasant to reverse. How the access levels differ, and where they are set, is in how to set up Google Search Console.

When to stop reading

When there was no diagnosis. Everything downstream of a guess is a guess with a price.

When the deliverables are counted and the outcomes are not. Counting output is how work gets measured when nobody is measuring results.

When the term is three months. That ends exactly where the data would start being informative, which serves the provider and not you.

When the same document, with a different logo, is findable online. It happens more than anyone admits.

The section every template recommends and no proposal should have

Look at the SEO proposal templates that rank for this query and they agree on a section this article has deliberately left out. They call it the ROI projection, and the instruction is usually phrased like this: calculate what ranking for the client's priority keywords would be worth in revenue, and make it tangible.

It is the most persuasive page in a standard proposal, and it is arithmetic on three guesses stacked on top of each other.

Guess one: the search volume. A monthly average from a tool, rounded, for a term that may be seasonal and may bundle several intents.

Guess two: the click-through rate at the target position. This is the one that broke. The tables most proposals still use come from before generative answers occupied the top of the page, and organic click-through has fallen where those answers appear, unevenly and by an amount that depends on the query.

Guess three: the conversion rate. Usually the client's current site-wide rate, applied to traffic that has not arrived yet, from a channel that behaves differently from the ones that produced it. Search traffic arriving on an informational query converts nothing like paid traffic arriving on a product term, and the difference is the whole subject of search intent, covered in the Spanish guide.

Multiply three estimates and the error multiplies with them. The number at the bottom has a currency symbol and the precision of a forecast, and it is a product of assumptions that nobody in the room can check, which is exactly why it works as a sales device.

A proposal that omits it looks weaker and is worth more. What belongs there instead is the measurement plan: which queries, on what fixed cohort, compared against which month, and reported from the client's own Search Console rather than from a dashboard the supplier controls. It is also what makes the engagement reviewable six months in without an argument, which is a separate exercise with its own signals: is my SEO agency working. That is a promise you can be held to, and the difference between the two is the clearest signal of what kind of supplier you are reading. The rest of the signals are in the sections above, and the one that separates them fastest is whether the document names a fixed set of queries at all.

One more thing those templates take for granted. They all recommend a case study with real numbers from a named comparable client. Publishing a client's traffic data needs that client's permission, and it is asked for far less often than it is used. A supplier who names three clients and their figures without hesitating is telling you what will happen to yours.

The figures that do belong in a proposal are the supplier's own, and they make a much shorter conversation once you know what the market actually charges: what SEO costs a small business.

Mistakes that repeat

• Comparing proposals by price. The same figure covers opposite work.
• Accepting one written without your data. It is a template with your logo on it.
• Reading the deliverables and skipping the measurement section. That is where the substance is.
• Not asking what month one produces. The honest answer is unimpressive and reassuring.
• Signing without an exit condition. Nobody agreed what failure would look like.
• Granting ownership access. Restricted reads the same data and reverses cleanly.

Data and transparency

That the SEO proposal templates ranking for this query recommend a revenue projection built from keyword volume, an assumed click-through rate and an assumed conversion rate, and a case study with a named comparable client's figures, is an observation of the first page of results on 12 August 2026 and is recorded as a practice to avoid, not as a source. That organic click-through has fallen where generative answers appear, unevenly and by query, comes from the Seer Interactive study covering June 2024 to September 2025. No figure is offered here for what an SEO engagement returns: it depends on margin per customer and on the market, and any single number would be the same three stacked guesses this section describes.

The 2026 algorithm update data, including four confirmed updates and the March one changing position for 79.5 % of the top three results, comes from SE Ranking. That searches for this term return mostly templates and sample documents is an observation from this project's own keyword harvest of the US market, where the phrase appears alongside variants such as sample and free proposal, collected August 2026. No statistics appear on proposal acceptance rates, typical retainer lengths or how often engagements fail: figures of that kind circulate widely and none I checked traced to a primary source. I sell these services, so this article describes what I think a proposal owes a buyer, and that interest is disclosed here rather than woven into the argument. Verified as of August 2026.

Primary sources, opened on 13 August 2026: the Seer Interactive study; the SE Ranking survey.

What this changes

Most of the advice about buying SEO is aimed at spotting bad providers, and the proposals that cause the most damage are rarely written by them.

They are written by capable people who were never asked for a diagnosis, never asked how it would be measured, and never asked what failing would look like, so they produced the document that gets accepted. A buyer who asks those three questions gets a different document from the same provider. The proposals that go wrong quietly are not the dishonest ones; they are the ones where nobody asked a question that had a wrong answer.

Frequently asked questions