The best SEO agency for your store is the one whose specialism matches your site, your stage, and your market. Merchants under roughly $500K usually get more from fixing fundamentals in house than from any monthly retainer.
Google’s own hiring guidance says organic results typically take four months to a year, and warns that an irresponsible SEO can damage your site. Any agency promising a number one position in ninety days is contradicting the search engine it claims to have mastered.
Most SEO agency shortlists get built backward. A founder searches for a category term, opens five agency websites, compares the headline claims, and picks whichever one sounds most confident. Every site on that list says it is results driven, data led, and transparent. None of that helps you choose, because none of it is falsifiable.
This guide is written from the agency side of that table. I run PN Digital SEO Agency, a London based search agency, so treat the perspective accordingly and weigh it against the alternatives named throughout. What follows is the process I would want a merchant to run on us: the questions that are genuinely hard to answer with marketing copy, the pricing benchmarks that tell you whether a quote is reasonable, and the specific circumstances in which the correct decision is to hire nobody this quarter.
Whether you are doing $30K months or $1M months, the underlying test is the same. An agency should be able to tell you what is limiting your organic performance right now, before it tells you what it sells.
A good SEO agency diagnoses before it prescribes, and can explain why each recommendation matters in terms of your revenue rather than your rankings. That single behaviour separates most of the credible field from most of the rest, and it shows up in the first sales conversation rather than in month four.
The diagnostic looks like this. Before proposing a scope, the agency examines the site’s technical condition, the competitive set, how your customers actually phrase what they want, and where the current organic performance is leaking. It then tells you which of those is the binding constraint. An agency that opens with a package tier before it has looked at your Search Console data is selling a product, not solving your problem.
The clearest disqualifier is a guaranteed ranking. Search results are influenced by competitors, algorithmic systems, and market conditions no agency controls. Google’s own guidance on deciding whether to hire an SEO puts the realistic timeline at four months to a year, and is unusually direct about the fact that a careless provider can damage your site and reputation. An agency promising a top position on a fixed date is either misunderstanding the system or hoping you do.
Ratings and reviews are useful evidence, but read them for content rather than for the star average. Feedback describing communication cadence, how the agency handled a month where results stalled, and whether reporting connected to commercial outcomes tells you far more than a five star aggregate. Case studies matter for the same reason: you are looking for evidence they have solved a problem structurally similar to yours, at roughly your scale, not evidence that somebody somewhere was happy.
An agency buys you multiple disciplines at once, a freelancer buys you one specialist cheaply, and an in house hire buys you institutional knowledge that compounds. These are not three price points for the same thing, and choosing on price alone is how merchants end up paying for the wrong shape of help.
An in house SEO lead integrates deeply with your merchandising, product, and development teams, and learns your catalogue in a way no external partner will. The constraint is coverage. Technical SEO, content strategy, digital PR, and analytics are four different skill sets, and one person rarely has all four at senior level. That gap is where in house programmes usually stall.
A freelance consultant is the right call for bounded work: a technical audit, a migration plan, a keyword and intent architecture you then execute yourself. Ahrefs’ survey data shows agencies charge roughly 138 percent more than freelancers on average, and a good part of that premium buys execution capacity rather than better thinking. If you have execution capacity in house already, you may be buying the wrong half.
An agency earns its premium when a strategy genuinely requires several specialists working in parallel, or when you need throughput you cannot staff for. Many merchants land on a hybrid: an internal owner who holds context and priorities, with an agency supplying technical depth and production volume.
Before committing to any of the three, it is worth reading how other operators evaluated this same decision. eCommerce Fastlane publishes an unranked field guide to ten agencies marketing AI search services, including honest limitations for each and a stage by stage view of when hiring makes sense at all. Reading a comparison written by someone with no commercial stake in your choice is a reasonable counterweight to reading ours.
Most SEO work is priced as a monthly retainer, and the credible market range for a serious programme runs from roughly $1,500 to $10,000 per month depending on scope, competition, and geography. Anything materially below that band is usually buying you a checklist rather than a strategy.
The most useful public benchmark comes from Ahrefs’ survey of 439 SEO providers, which put the average monthly cost at $2,917 and found that 78.2 percent of providers bill on a retainer. The distribution matters more than the average: 23 percent of respondents charged between $500 and $1,000 per month, while the average hourly rate sat at $111 and roughly one in ten providers charged above $150 per hour. A handful of enterprise retainers pull the mean upward, so treat the average as context rather than as a target.
What actually moves the number is scope. Target market count, current technical condition of the site, keyword competitiveness, content volume required, digital PR needs, catalogue complexity, reporting depth, and how fast you want to move all push the figure up or down independently. A single market local business and a multi region ecommerce catalogue will never require the same investment, and any agency quoting both the same way has not looked at either.
Stage guidance is straightforward. Under $250K in annual revenue, most of the available gains are in work you can do yourself, and a retainer is premature. Between $250K and $2M, a focused engagement in the $1,500 to $4,000 range is usually where the maths works, provided the scope is narrow enough to actually execute. Above $2M, the question stops being what SEO costs and becomes what a percentage point of organic revenue is worth defending. Ask every agency to connect its number to specific deliverables and intended outcomes, then compare those, not the headline fee.
Ecommerce SEO lives or dies on collection and product page architecture, not on blog output. For a Shopify store, the highest value work is usually collection page targeting, product page structure, internal linking, faceted navigation control, duplicate content management, and complete product schema, in roughly that order.
The mechanics are platform specific enough that they are worth naming. Shopify generates multiple URL paths to the same product, which creates canonicalisation decisions most generalist agencies handle carelessly. Filters and sort parameters can multiply crawlable URLs until crawl budget is being spent on combinations no customer searches for. Collection pages are the pages that capture non branded demand, and they are the ones most often left with a default template and no descriptive content. If you want the detail, this breakdown of how collection page architecture actually earns non branded organic traffic covers the structural decisions well.
Keyword work has to separate informational from transactional intent, and route each to the right page type. Someone searching for advice needs a guide. Someone searching a product category needs a commercial page that helps them compare and buy. Sending the second person to a blog post is a common and expensive mismatch.
Machine readability is now part of the same job. eCommerce Fastlane’s 2026 field guide reports Adobe research from April 2026 finding that ecommerce product detail pages average only 66 percent machine readability, which means most catalogues are handing incomplete data to the systems that increasingly decide what gets recommended. Complete product schema with accurate identifiers, pricing, availability, and review aggregates is unglamorous work that pays into traditional search and AI search at the same time. Ask any prospective agency who on their team owns schema, and what tooling they use to validate it at catalogue scale.
Specialism matters most where the underlying mechanics differ, and there are three cases where they genuinely do: local search, long cycle B2B, and large scale technical SEO. Outside those three, most claimed specialisms are positioning rather than capability.
Local SEO is a distinct discipline because the ranking inputs are different. Google Business Profile completeness, location and service page structure, citation consistency, and review signals drive geographic visibility in ways that have little to do with domain authority. For a small business with a physical footprint, prioritisation is the whole game: a limited budget spent across too many locations or services produces nothing anywhere.
B2B and SaaS SEO differs because the buying cycle is long and the first search happens well before anyone knows your category exists. As an illustrative benchmark, considered B2B software purchases commonly run six to eighteen months from first informational search to signature, which means the content has to span problem awareness, category education, use cases, integrations, comparisons, and alternatives pages. Success is measured in qualified demos, trials, and pipeline, not sessions.
Enterprise technical SEO is a scale problem. On a catalogue with hundreds of thousands of URLs, a single template level error propagates instantly, and crawl management, indexation control, JavaScript rendering, faceted navigation, canonicalisation, and international configuration all interact. The harder requirement is practicality. A technically perfect recommendation that cannot pass your engineering roadmap or CMS constraints has a real world value of zero, so an enterprise agency is being hired as much for prioritisation and communication as for technical depth.
The most efficient vetting tool is a fixed question set run identically across every agency on your shortlist, because the comparison only works when the input is constant. Budget one 45 minute call per agency and ask all eleven of these.
How will you assess our current organic performance, and what will you look at first? Which opportunities would you prioritise in the first ninety days, and why those? How do you determine search intent for our category? How do you approach technical SEO on our platform specifically? Who writes and reviews the content, and are they in house? How do you build authority, and what would make you refuse a link opportunity? Which metrics will you report, and which of them do you consider vanity? How will you connect organic performance to leads or revenue? Who manages our account day to day, and how senior are they? How often will we talk, and in what format? What happens if the first two quarters do not produce the expected results?
Two answers carry disproportionate weight. The last question tells you whether the agency has a plan for being wrong, which every honest agency needs. The question about refusing a link opportunity tells you whether there is a line they will not cross for a short term metric.
What you are listening for is specificity without jargon. An agency that cannot explain its approach in language you understand either does not understand it well enough itself, or is relying on the complexity to keep you from asking follow up questions. Neither is a good sign at the start of a twelve month relationship.
SEO ROI should be measured against the commercial objective, not the traffic number, using the formula (value generated minus investment) divided by investment, times 100. For an ecommerce business that means organic revenue, transactions, average order value, and acquisition cost. For a lead generation business it means qualified enquiries, booked consultations, opportunities, and pipeline value.
Supporting metrics still matter, but they are diagnostic rather than definitive. Non brand organic sessions, keyword visibility, click through rate, and conversion rate tell you why the commercial number moved. They should never be presented in place of it. Attribution will never be perfectly clean, because customers touch several channels before converting, so good reporting sits organic performance alongside blended commercial data rather than claiming sole credit.
On contract length, treat minimum terms as a constraint the strategy has to justify, not as a commitment device. Early months are weighted toward research, technical remediation, and foundational optimisation, with content and authority work compounding on top of that later. That sequencing is the legitimate reason six to twelve month terms are common. This month by month view of what the first year of SEO work actually contains is a useful reality check against any proposed timeline.
Before signing, get four things in writing: the minimum term, the notice period, the renewal mechanism, and what happens to content and assets if the relationship ends. A term length is reasonable. A term length with no accountability inside it is not, so agree the ninety day review points and the reporting format at the same time you agree the term.
Search visibility now spans traditional results, AI Overviews, and assistant platforms, and an agency reporting only blue link positions is measuring a shrinking share of the surface. This is the fastest moving part of the brief and the part where agency capability varies most.
The commercial stakes are no longer theoretical. eCommerce Fastlane’s field guide cites Shopify data from early 2026 showing orders originating from AI search platforms growing nearly 13 times year over year. Google has moved in the same direction on the guidance side: in June 2026 it published new material on evaluating third party SEO tools and services alongside generative AI optimisation, and was pointed about the fact that third party tools do not have access to Google’s internal ranking data.
Practically, the work that earns AI citations overlaps heavily with good traditional SEO, which is reassuring rather than disappointing. Content that is technically accessible, clearly structured, factually specific, and genuinely answers the question performs on both surfaces. What is new is the measurement layer, and it needs its own tooling. It is worth knowing what that tooling does before an agency tells you it has this covered, and Fastlane’s roundup of the platforms that track prompt level visibility and citation share is a reasonable starting map.
The vetting question is simple. Ask any agency to show you how it currently measures your visibility inside AI answers, and what it would change on your site in the first ninety days to improve it. An agency that answers with schema completeness, entity consistency, answer first content structure, and third party authority is describing real work. An agency that answers with the phrase AI powered and nothing else is describing a landing page.
Choose the SEO agency whose demonstrated specialism matches your platform, your stage, and your market, rather than the one with the strongest general claims. Run an identical question set across three agencies covering how they would assess your current performance, what they would prioritise in the first ninety days, who writes the content, and how they connect organic performance to revenue. Ask for client attributable case data at roughly your revenue scale, not house metrics from the agency’s own site. Read reviews for descriptions of communication and problem solving rather than for star averages, and treat any guaranteed ranking as an automatic disqualification.
Most businesses running a serious SEO programme pay between $1,500 and $10,000 per month, with Ahrefs’ survey of 439 providers putting the average at $2,917 and finding that 78.2 percent bill on a retainer. Where you sit inside that band depends on your market count, your site’s technical condition, keyword competitiveness, content volume, and reporting depth. Under $250K in annual revenue, a retainer is usually premature and the same money is better spent on fundamentals in house. Compare deliverables and intended outcomes across quotes rather than headline fees, because the cheapest proposal is frequently the narrowest one.
Google’s own hiring guidance puts the realistic timeline at four months to a year before organic work shows meaningful results, and that range holds for most ecommerce sites. Certain fixes land faster: a robots.txt file blocking crawlers, a broken canonical configuration, or missing product schema can produce visible movement within weeks of being corrected. Competitive category rankings take substantially longer because they require sustained content and authority work that compounds. Any agency committing to a specific position on a specific date is making a promise about systems it does not control, which is a reason to look elsewhere rather than a reason to be reassured.
Hire an agency when your strategy needs several specialists working in parallel or more execution capacity than you can staff, a freelancer when the work is bounded and specialist, and in house when you need someone holding institutional context full time. Ahrefs data shows agencies charge roughly 138 percent more than freelancers on average, and much of that premium buys throughput rather than better strategy. One in house person rarely covers technical SEO, content, digital PR, and analytics at senior level, which is where in house only programmes usually stall. Many growing brands settle on a hybrid: an internal owner for context and priorities, with an agency supplying depth and volume.
An SEO agency should be measuring your visibility inside AI answers and working on the inputs that drive citation, not just reporting traditional keyword positions. Concretely that means complete and accurate product schema, consistent entity information across the web, answer first content structure that resolves a question in the opening sentence, and third party authority signals that corroborate your claims. Ask the agency to show you its current measurement of your AI visibility and name what it would change in the first ninety days. Note that Google has stated third party SEO tools do not have access to its internal ranking data, so treat any tool derived score as directional rather than authoritative.
