Skip to main content Scroll Top

How to Choose an SEO Company: The Questions That Separate a Real Proposal From a Template

Updated September 2026 · Written and maintained by the Progression Agency strategy team

The questions to ask before signing, how to compare two proposals without collapsing into price, how to check the claims made about past results, how long the work really takes, and the contract clauses that matter more than the fee.

On this page · 14 sections
  1. What it costs, briefly, and where the full answer lives
  2. How to compare two proposals side by side
  3. How to check the claims made about past results
  4. Is SEO worth it?
  5. How long does SEO take to work?
  6. The questions to ask before signing anything
  7. Red flags, in order of how much they should worry you
  8. Contracts: term length, notice and what you keep
  9. What good reporting actually contains
  10. SEO, local SEO and AEO: three jobs sold under one word
  11. Why directories hold the top of almost every agency search
  12. Does hiring locally change the price or the result?
  13. What any quote needs from you to be honest
  14. Running a ninety-day trial instead of a long commitment

The short answerAsk where the site currently ranks for the terms being proposed, measured and dated, and which sites hold those positions today. Everything else in a proposal can be written without opening the website, so that one answer separates work that has been looked at from a template. Then judge the rest on whether it names what will NOT be done, defines which single action counts as a lead, says what happens in month two if month one is flat, and leaves the accounts and content with the business. Nobody can honestly give a date for results.

Search volumes, cost per click and difficulty scores on this page were measured in Ubersuggest for the United States in September 2026 and are shown as measured. No market price benchmarks are quoted, because no reliable public source for them exists.

What it costs, briefly, and where the full answer lives

Enough to judge a quote. The detailed pricing breakdown is a separate subject.

There is no market rate, because the work sold under one word covers repairing an uncrawlable site, writing dozens of pages, fixing a business listing and building measurement that does not exist yet. The variables that decide a number are the starting technical state, how many pages are needed, how contested the terms are, who writes the content, how quickly changes can be approved, and how many markets are involved. A quote that has not asked about those was not calculated. The models, the bands and the method for working out a ceiling from your own figures are set out in full in the SEO pricing guide and in SEO packages and pricing; this page is about the rest of the decision, which is the part that costs people the most when they get it wrong.

The one figure to work out before reading any quote

Gross profit on an average customer multiplied by the share of enquiries that become customers gives the value of a single enquiry. Multiply that by how many extra enquiries a month would genuinely be worth having, and you have the monthly value the work must create. A fee is only judgeable against that number, never against someone else’s budget.

Why the cheapest proposal is often the most expensive

Not because low fees always mean poor work, but because the cheapest proposal is usually the one with the least defined scope. Undefined scope resolves into either very little happening or everything being billed as an extra.

How contested a term is, measured rather than asserted

Cost per click is the clearest public proxy, because it is what advertisers will pay for one visit. Measured in September 2026, local seo pricing carried a cost per click of $24.86 against $7.71 for ecommerce seo pricing. The organic work behind those two is competing with very different levels of spend, which is why one scope cannot carry one price.

What people actually search when they are pricing this work (Ubersuggest, US, September 2026)
Search termMonthly US searchesCost per clickDifficulty (0-100)
seo pricing2,900$16.5949
seo pricing package1,600$11.0346
seo pricing packages720$19.3720
seo contract agreement480$12.7941
how long does seo take to work390$2.7948
local seo pricing320$24.8634
seo contract260$11.6039
website seo pricing260$19.3246
is seo worth it210$16.1530
seo contract template210$1.5339
seo pricing and plans170$48.3926
how long does seo take170$12.8948
monthly seo pricing140$21.7734
seo pricing model140not measured30
how to choose an seo company140not measured33
is seo worth it for small business110$24.5842
ecommerce seo pricing110$7.7122
white label seo pricing90$31.5119
small business seo pricing40$18.3548
enterprise seo pricing40not measured34
questions to ask seo agency30not measured18
affordable seo pricing30not measured23
How the hiring decision actually goes wrongHow the hiring decision actually goes wrong
The failure is at step four. Comparing on price happens when proposals contain nothing else that can be compared, which is a fixable problem before you sign anything.

How to compare two proposals side by side

Most comparisons collapse into price because nothing else in the documents lines up.

A comparison order that does not collapse into price
CompareWhy this orderWhat a weak proposal looks like here
The measured baselineNothing later can be verified without itAbsent, or stated as a target instead
What is excludedExclusions reveal method; inclusions reveal marketingNo exclusions at all
The lead definitionDifferent definitions make cost per lead meaningless“We’ll track it in Analytics”
Where AEO sitsFolded in or priced separately changes both totalsA separate fee with no distinct tasks named
The month-two answerDistinguishes a method from hopeNo answer, or “results take time”
What you keep at the endDecides whether the spend leaves an assetAccounts created under the agency’s login
The totalsLeast comparable item, so read it lastRead first, which is how comparisons go wrong

Why price should be read last

Two totals are only comparable once the scopes behind them are. Reading price first anchors every subsequent judgement to a number that may be covering entirely different work.

What to do when one proposal has no baseline

Ask for it before comparing anything. If it does not arrive, the comparison is between one measured plan and one unmeasured claim, which is not a comparison.

How to check the claims made about past results

Three checks that need no special access and take about fifteen minutes.

Claims and how to test them
ClaimHow to check itWhat a good answer looks like
“We ranked a client number one”Ask for the term, the market and the monthA specific term you can search yourself
“Traffic grew by X%”Ask what the starting number wasA base figure, not only a percentage
“We work with businesses like yours”Ask which ones, and what the work wasNamed work with a describable scope
“We’re partners with a platform”Check the platform’s own partner directoryA listing you can find independently
“Award winning”Ask which award, which year, awarded by whomA body you can look up

Why percentages without a base are meaningless

Growth from four visits to twelve is a two hundred per cent increase. Percentages describe the shape of a change, not its size, so the starting number is the part that carries the information.

Why a specific term beats a screenshot

A ranking screenshot is a moment, taken from one location on one device. A named term and market is something you can check yourself, now, from where you are.

Is SEO worth it?

The conditions under which it is, and the conditions under which it is not.

When it clearly is

When people are already searching for what you sell, when a customer is worth enough to justify a months-long payback, and when you can service more work than you currently get. Under those three conditions the question is only which terms and how fast.

When it clearly is not

When nobody searches for the category yet, when the business cannot survive the months before results, or when the enquiries you already get are not being converted. In the third case more traffic amplifies an existing problem rather than solving one.

The honest middle case

Most businesses are in it: the demand exists but is smaller than expected, and the winnable terms are further down the list than the obvious one. That is still worth doing, but it needs the demand measured first rather than assumed.

Signals that organic search suits a business, and signals it does not
SignalPoints towardWhy
People search your category by nameWorth doingDemand exists and can be measured
High customer lifetime valueWorth doingA long payback is affordable
You can handle more workWorth doingCapacity converts traffic into revenue
Category has no search volumeReconsiderYou would be creating demand, not capturing it
Enquiries already go unconvertedFix firstMore traffic magnifies the leak
Need revenue within eight weeksWrong channelRanking does not move that fast

How long does SEO take to work?

The least satisfying answer, and the most important one to settle before signing.

Nobody can give you a date. Ranking depends on what competitors do, how much authority a site already has, and how often it is recrawled. None of those are controlled by whoever is doing the work, so a specific month is either a guess or a sales tactic. What can be described honestly is the shape.

The shape, in order

Technical fixes show fastest, sometimes within weeks. Newly published pages must be crawled and indexed before they can rank at all, and that is frequently the real bottleneck. Long-tail positions follow. Competitive head terms take many months, and authority changes slowest of all.

What should be visible at ninety days even so

Pages being indexed, impressions rising even without clicks, and positions moving from page five toward page three. Those are leading indicators. Their complete absence at ninety days is a genuine warning sign; a lack of clicks at ninety days is not.

The most common and most expensive mistake

Stopping in month two. The period where nothing appears to happen is not evidence that nothing is happening, and it is exactly when most engagements are cancelled, converting a slow investment into a total loss.

How long different work takes to showHow long different work takes to show
Weeks to first visible movement, not to a finished result. Anyone quoting a date rather than a shape is guessing.
What is reasonable to expect, and what is not evidence of failure
TimeframeReasonable to expectNOT evidence of failure
Weeks 1-4Technical fixes live; pages being crawledNo ranking movement at all
Month 2Impressions for terms you did not appear forStill no clicks
Month 3Positions moving page 5 toward page 3No page-one position yet
Month 6Page one on easier terms; first enquiriesHead term still unreached
Month 12A realistic shot at competitive termsAnything, provided leading indicators moved

Want this done for your site?We build and maintain the search, content and paid programmes described on this page.

Get a free proposal

The questions to ask before signing anything

Six of them. The first one does more work than the other five combined.

Ask where the site currently ranks for the terms being proposed, measured and dated, and which sites hold those positions today. Everything else in a proposal can be written without opening the website; that answer cannot.

  1. Where does this site rank for the proposed terms today, measured and dated?
  2. Which sites hold those positions now, and what are they doing that this site is not?
  3. What will you NOT do? An offer with no exclusions has no method behind it.
  4. Which single action counts as a lead, and how is spam excluded from it?
  5. What happens in month two if month one is flat?
  6. What do I keep if this ends: the accounts, the content, the tracking?
What a proposal should contain before you signWhat a proposal should contain before you sign
The last three are not missing details. Their presence is the warning.

Why a start point matters more than a target

Without a measured baseline nobody can later prove anything moved. Proposals that open with targets and never state a starting position are avoiding exactly that comparison.

Why naming the competitors is the real test

Anyone can write an industry and a city into a document. Naming the three sites currently ranking, and saying what they are doing, requires someone to have looked. It takes ten minutes and most proposals skip it.

What a good answer sounds like

Specific and slightly discouraging: closer to “you are position 34, the top three are all directories, and the realistic target is the four questions underneath that term” than “we’ll get you to page one”.

Red flags, in order of how much they should worry you

Some are dealbreakers. Others are only questions that need answering.

  1. A guaranteed ranking. Nobody controls the systems that decide this, so a guarantee is either a misunderstanding or a manipulation.
  2. No measured start point, which makes anything unprovable later.
  3. No stated exclusions, which means the scope is whatever is convenient.
  4. Vague attribution. Ask which single action counts as a lead and how spam is excluded.
  5. A flat price with no variables behind it. It was chosen to be accepted, not calculated.
  6. Reporting that is only a list of rankings, which is the easiest report to produce and the least connected to revenue.
  7. Ownership of accounts or content resting with the agency rather than the business.
No start point — Red flag. Cannot prove movement.
Guaranteed rankings — Red flag. Nobody controls this.
No exclusions — Red flag. Selling everything.
Vague attribution — Red flag. Cannot judge later.
Locked 12-month term — Depends. Ask what month two looks like.
Flat number, no variables — Red flag. Chosen to sound acceptable.

Contracts: term length, notice and what you keep

The parts of an agreement that matter more than the fee.

What to check in an agreement, and what a reasonable position looks like
ClauseWhy it mattersA reasonable position
Minimum termRanking takes months, so very short terms suit nobodyA stated term with a defined review point inside it
Notice periodDecides how quickly you can stopNotice proportional to the term, stated in days
Scope definitionThe single biggest source of disputeNamed deliverables plus what is explicitly excluded
Ownership of contentDecides what survives the engagementContent and pages belong to the business
Ownership of accountsAnalytics and search accounts hold the historyAccounts created in the business’s own name
Reporting cadenceSets the rhythm of accountabilityA fixed schedule, with what the report contains defined
Price reviewPrevents surprise increasesAny change stated in advance with notice

Why a long term is not automatically a bad sign

Given how long ranking takes, a term that is too short guarantees the engagement ends before the work could have produced anything. What matters is what happens inside the term: whether there is a review point, and whether the work is visible rather than only the invoice.

The clause most often missed

Account ownership. Analytics, search console and business listing accounts hold the entire history of what happened. Created under someone else’s login, they leave with them, and the history is not recoverable.

What good reporting actually contains

If only one thing changes about an existing arrangement, make it this.

Reporting that is useful, next to reporting that only looks useful
UsefulLooks useful but is notWhy
Enquiries by sourceSessionsSessions are the easiest number to move and the least tied to revenue
What changed and what it producedA dashboard exportAn export shows numbers, not decisions
Positions grouped in bandsA list of exact ranksExact ranks vary by searcher and location
What did not workOnly winsA report with no failures is not being written honestly
Next period’s specific plan“Continue optimising”A plan that could be copied to any client is not a plan

The five-minute test

A report should be readable in five minutes by someone who does not do this work. If it requires explanation to be understood, it is being used to occupy space rather than to inform a decision.

Why exact rank lists mislead

Results vary by location, device and search history, so a single reported position is one sample from a distribution. Bands, and the direction of travel between them, carry the information that a precise-looking number does not.

SEO, local SEO and AEO: three jobs sold under one word

Worth separating before buying any of them, because they are priced separately.

The three, side by side
What it targetsWhen it is the priorityHow it is measuredTypical pricing signal
SEOOrganic positions for service termsSelling beyond one metroPositions, then enquiries by sourceScales with page volume and competition
Local SEOMap results and proximity placementMost buyers within driving distanceCalls and direction requestsScales with number of locations
AEOBeing named in AI assistant answersBuyers research by asking rather than searchingWhether you are named, tested with fixed promptsRarely a separate line; shares the SEO foundation

Why AEO cannot honestly be quoted with a search volume

Keyword tools count searches typed into a search engine. They cannot count questions asked of an assistant, and no reliable public source does. Any AI search volume in a proposal is extrapolated from a private sample. What can be measured is whether a business is named when the question is asked, which is a test anyone can run in a few minutes with fixed wording.

Why they are not alternatives to each other

They share a foundation: a site that can be crawled, answers questions directly, and carries consistent business details. The difference is only where effort goes once that foundation exists, which is why AEO priced as a wholly separate product deserves a question about what work it contains that the SEO work does not.

What that means for comparing two quotes

If one quote separates AEO into its own fee and the other folds it in, the totals are not comparable until you ask both what specific tasks sit under that heading.

Technical — Crawl, speed, structure. Fastest to show.
Content — Answers to real questions. Slow and compounding.
Local — Profile and consistency. Decides map results.
Measurement — Leads by source. Usually skipped.
Authority — Independent mentions. Slowest, most durable.
Reporting — What changed and why. Not a rankings list.

Useful context before judging anyone by where they rank for their own category.

They are built for exactly that query

Large directories exist to answer “who are the agencies in X”. They carry substantial authority, update constantly, and maintain a page for every location and service combination. Competing with them head-on for the category term is a poor use of effort for almost anyone.

Which makes that search a weak test

Judging by whether someone outranks a directory for their own category term is close to meaningless. A better test is whether they rank for the specific questions their buyers ask, which is harder to fake and closer to the work you would be buying.

Listings still cost money on a recurring basis

A directory listing is a citation and a referral source, and it is not a substitute for being findable independently. The distinction matters when a quote includes directory spend inside the fee.

Does hiring locally change the price or the result?

The question every location-based search is really asking.

What proximity genuinely helps with

Knowing which competitors are already visible in a market, which directories that sector uses there, and how buyers in the area phrase what they need. That knowledge is real, and it comes from having looked rather than from an address.

What it does not help with

The work itself, which is identical from anywhere. Technical fixes, content and measurement are not improved by proximity, and rates vary more by scope than by geography.

The test that replaces the question

Rather than asking where someone is based, ask them to name the competitors ranking in that market and explain what those competitors are doing. Someone local who cannot answer is no better than someone remote who can.

Rather than asking where someone is based, ask them to name the competitors ranking in that market and explain what those competitors are doing.

What any quote needs from you to be honest

Six inputs. Without them, a number is invented rather than calculated.

  • Which areas are actually served, and whether you travel to customers.
  • Which services carry the margin, because those are the ones worth ranking.
  • Close rate by source, which changes every recommendation that follows it.
  • What a customer is worth in gross profit rather than revenue.
  • What has already been tried, by whom, and what happened.
  • Whether the website can be changed freely, or whether every change needs approval elsewhere.

Why site access matters more than platform

Whether a site runs on one platform or another matters far less than whether changes go live this week or need three approvals. That single fact decides how long everything takes and therefore what it costs.

Why “what has been tried” is worth stating plainly

Previous work leaves traces: pages, links, redirects and tracking that either help or need undoing. A quote made without knowing that history is pricing an unknown quantity of repair.

Not sure which of these applies to you?Tell us the situation and we will say plainly what we would do first, and what we would not.

Talk it through

Running a ninety-day trial instead of a long commitment

A better structure than either a short contract or a leap of faith.

  1. Agree the measured baseline first, in writing, with a date on it.
  2. Pick a small set of realistically winnable terms rather than the head term.
  3. Agree the leading indicators: indexing, impressions, position bands.
  4. Set the review at ninety days and agree in advance what would count as enough to continue.
  5. Keep every account and everything produced in the business’s own name.

Why ninety days and not thirty

Thirty days is not long enough for new pages to be crawled, indexed and settle into a position, so a thirty-day trial reliably produces a negative result regardless of the quality of the work.

What should count as success

Something verifiable without help: pages indexed, impressions rising, positions moving between bands. If success is defined only by a number one side produces, the review is not a review.

AI, AEO and what is changing

Websites and design

Choosing and working with an agency

Social, content and brand

By industry and by situation

Get the measured baseline before you compare any quotes

A free proposal that states where the site ranks today for the terms that matter, which sites hold those positions, and which of them are realistically winnable from that starting point. The measurement comes before any number.

Get a free proposal

Prefer to see the numbers on your own site?We will run the audit described above and walk you through what it finds.

Request an audit

Frequently asked questions

How do I choose an SEO company?
Ask where the site currently ranks for the terms being proposed, measured and dated, and which sites hold those positions today. An agency that cannot answer has not looked at the market. Everything else in a proposal can be written without opening the website, so that single answer separates work that has been looked at from a template.
What questions should I ask an SEO agency?
Where the site ranks today for the proposed terms; which sites hold those positions and what they are doing; what will NOT be done; which single action counts as a lead and how spam is excluded; what happens in month two if month one is flat; and what you keep if it ends.
How do I compare two SEO proposals?
In this order: the measured baseline, the exclusions, the lead definition, where AEO sits, the month-two answer, what you keep at the end, and the totals last. Reading price first anchors every later judgement to a number that may be covering entirely different work.
Why should a proposal say what will NOT be done?
An offer with no exclusions has no method behind it. Naming what is out of scope is what makes the included scope mean something, and it is the fastest way to tell a considered proposal from a generic one.
How can I check the results an agency claims?
Ask for the term, the market and the month behind any ranking claim, and the starting number behind any percentage. Growth from four visits to twelve is a two hundred per cent increase, so a percentage without a base carries no information.
How much does SEO cost?
There is no single market rate, because the work covers repairing an uncrawlable site, writing dozens of pages, fixing a business listing and building measurement that does not exist yet. What decides a number is the starting technical state, page volume, how contested the terms are, who writes the content, approval speed and geographic spread. The models and bands are set out in full in the SEO pricing guide.
How do I know if a fee is reasonable for my business?
Multiply gross profit on an average customer by the share of enquiries you close to get the value of one enquiry, then multiply by how many extra enquiries a month would be worth having. A fee is judgeable against that figure and against nothing else, which is why a market average cannot settle it.
Why is the cheapest proposal often the most expensive?
Because the cheapest proposal is usually the one with the least defined scope, and undefined scope resolves into either very little happening or everything being billed as an extra.
Is local SEO more expensive than regular SEO?
It is priced differently rather than higher, usually scaling with the number of locations rather than the number of pages. Local terms are commercially contested: local seo pricing measured a cost per click of $24.86 in September 2026 against $7.71 for ecommerce seo pricing.
Is SEO worth it?
It is worth it when people already search for what you sell, a customer is worth enough to justify a months-long payback, and you can service more work. It is not worth it when the category has no search demand, the business cannot survive the months before results, or existing enquiries already go unconverted.
Is SEO worth it for a small business?
Often yes, but rarely on the obvious head term. The winnable terms are usually further down the list and more specific, which is why the demand has to be measured before a plan is made rather than assumed from the category name.
How long does SEO take to work?
Nobody can honestly give a date, because it depends on competitors, existing authority and crawl frequency. The shape is: technical fixes in weeks, indexing of new pages in weeks to months, long-tail positions around three months, competitive head terms many months, and authority slowest of all.
What should be visible after ninety days?
Pages being indexed, impressions rising even without clicks, and positions moving from page five toward page three. Their complete absence is a genuine warning sign; a lack of clicks at ninety days is not.
What is the biggest mistake businesses make?
Stopping in month two. The period where nothing appears to happen is not evidence that nothing is happening, and cancelling then converts a slow investment into a total loss.
What questions should I ask an SEO agency?
Where the site ranks today for the proposed terms, measured and dated; which sites hold those positions; what will not be done; which single action counts as a lead; what happens in month two if month one is flat; and what you keep if it ends.
Why does the first question matter so much?
Because everything else in a proposal can be written without opening the website. A measured, dated starting position cannot, so it separates work that has been looked at from a template.
Why should a proposal say what will NOT be done?
An offer with no exclusions has no method behind it. Naming what is out of scope is what makes the included scope mean something.
What are the red flags in an SEO proposal?
A guaranteed ranking, no measured start point, no stated exclusions, vague attribution, a flat price with no variables behind it, reporting that is only a rankings list, and accounts or content owned by the agency rather than the business.
Why can nobody guarantee rankings?
Because nobody controls the systems that decide them. A guarantee is either a misunderstanding of how ranking works or a bet that the claim will not be checked.
Should I sign a twelve-month SEO contract?
A term that is too short guarantees the arrangement ends before the work could have produced anything. What matters is what happens inside it: whether there is a defined review point, and whether the work is visible rather than only the invoice.
What should an SEO contract include?
A stated minimum term with a review point, a notice period, named deliverables plus explicit exclusions, ownership of content and accounts resting with the business, a reporting cadence with the report contents defined, and advance notice of any price change.
Which contract clause is most often missed?
Account ownership. Analytics, search console and business listing accounts hold the entire history of what happened; created under someone else’s login, they leave with them and the history is not recoverable.
What should an SEO report contain?
What work was done, what it produced, what did not work, enquiries by source rather than sessions, positions grouped in bands rather than exact ranks, and a specific plan for the next period. If it cannot be read in five minutes by someone outside the field, it is not a report.
Why is a list of rankings a poor report?
Results vary by location, device and search history, so a single reported position is one sample from a distribution. It is also the easiest report to produce and the least connected to revenue.
What is the difference between SEO, local SEO and AEO?
SEO targets organic positions for service terms, local SEO targets map results and proximity placement, and AEO aims at being named in AI assistant answers. They share a foundation, which is why AEO priced as a wholly separate product deserves a question about what work it contains that the SEO work does not.
Can anyone tell me the search volume for AI questions?
No. Keyword tools count searches typed into a search engine and cannot count questions asked of an assistant. Any AI search volume quoted is extrapolated from a private sample. What can be measured is whether a business is named when the question is asked, using fixed wording.
Does hiring a local agency cost more?
Rates vary far more by scope than by geography. Proximity helps with knowing a market, not with doing the work, so the useful test is asking anyone to name the competitors ranking in that market and explain what they are doing.
Can I run a short trial instead of committing?
Ninety days rather than thirty. Thirty days is not long enough for new pages to be crawled, indexed and settle, so a thirty-day trial reliably produces a negative result regardless of the quality of the work.
What happens to rankings if I stop?
They decay slowly rather than disappearing. Content that ranks tends to keep ranking for a while and technical fixes stay fixed. What stops is the compounding, while competitors carry on.

Want this done for your site?We build and maintain the search, content and paid programmes described on this page.

Get a free proposal

Get a free marketing proposal

Tell us what you are trying to grow and we will come back with a plan, not a pitch deck. Same-day reply on weekdays.

Privacy Preferences
When you visit our website, it may store information through your browser from specific services, usually in form of cookies. Here you can change your privacy preferences. Please note that blocking some types of cookies may impact your experience on our website and the services we offer.
Contact Us