How to Choose an SEO Company in the US (Without Getting Sold)

Most US businesses hire an SEO company after a pitch, not after a comparison. That single sequencing mistake explains the majority of disappointing engagements we hear about. A pitch is designed to make one option look inevitable; a comparison is designed to make trade-offs visible. When you run the pitch first, you evaluate presentation quality. When you run the comparison first, you evaluate the thing you are actually buying — a team's judgement about where your next unit of revenue comes from.

This guide is written from the buyer's side of the table. It assumes you are a marketing leader, founder, or operator who does not want to become an SEO practitioner but does need to tell a good programme from an expensive one. Every section below maps to a decision you will have to make anyway, usually under time pressure, usually with three proposals that all look similar.

Define the shape of the work before you take a single call

Deliverables lists are interchangeable across proposals because they are written to survive procurement, not to describe strategy. What genuinely differs between agencies is where the centre of gravity of the work sits. In the US market there are four recognisable shapes, and most engagements are dominated by one of them.

Technical remediation. The constraint is your stack: rendering, indexation, site speed, faceted navigation, migrations, or a CMS that fights every change. The work is diagnostic and engineering-adjacent, and progress depends on your development capacity as much as the agency's.

Editorial production at volume. The constraint is coverage: you do not have pages that answer the questions your buyers ask. The work is research, briefing, writing, editing, and internal linking, sustained for quarters rather than weeks.

Authority and digital PR. The constraint is trust: your pages are good, but the domains outranking you are cited everywhere and you are not. The work is data-led story production, outreach, and relationship building — the slowest and least predictable of the four.

Conversion and commercial optimisation. The constraint is monetisation: traffic exists but does not convert, or converts into unqualified leads. The work is page-level messaging, offer design, tracking, and lead quality feedback loops.

Write down which of these describes your bottleneck, then ask each agency to argue against the other three. The good ones will qualify their answer with what they would need to see in your analytics first. The weaker ones will confirm whichever shape they happen to staff for. That single question is the cheapest filter available to you.

Match the quote to the difficulty, not to the vendor

A price is neither expensive nor cheap in isolation. It is either matched or mismatched to the competitive difficulty of the outcomes you asked for. Before you compare proposals, spend an hour looking at who currently ranks in the top five for your five most commercially valuable queries. Note what those pages are: publisher round-ups, well-funded competitors, marketplaces, or thin local pages.

The point of the table is not the numbers, which vary by vertical and city. It is the diagnostic: if the pages beating you are backed by teams, an hours-based retainer is a multi-year plan whether or not it was sold that way. Being told that honestly in a sales call is a strong buying signal.

Ask who does the work after month one

Senior attention in the pitch and junior execution by month three is the most common failure pattern in US retainers, and it is almost never disclosed. It also is not always wrong — a well-supervised junior executing a clear plan is fine. What matters is that you know the structure before you sign.

• Named individuals on your account, with roles and approximate weekly hours.
• Which of those people you met during the pitch, and which you did not.
• What happens when your strategist is reassigned or leaves: who absorbs the account, and does the retainer change?
• Whether writing, link outreach, or development is subcontracted, and to whom.
• The escalation path when something is late, and what a make-good looks like.

Get the answers into the statement of work rather than the proposal deck. Decks are marketing documents; the SOW is the only thing that survives a change of account manager.

Insist on a measurement definition before you discuss price

Any agency can grow a traffic chart. Fewer can tell you which of those sessions were worth having. Agree in writing, before pricing, on four things: the primary success metric, the attribution window, the reporting cadence, and which parts you can verify yourself.

The primary metric should be the closest thing to money that your tracking can honestly support. For most B2B companies that is qualified pipeline or opportunities created; for e-commerce it is non-brand revenue; for local services it is booked jobs or calls that lasted longer than a threshold. Rankings and sessions are diagnostic instruments, not outcomes, and a proposal that defines success only in those terms is not accountable to your business.

Insist on the brand versus non-brand split in every report. Brand search grows when your paid media, PR, or product does well, and it is the single easiest way for an underperforming SEO programme to look successful. If an agency resists separating the two, you have learned something important for free.

Read the contract for exit terms before you read the scope

Twelve-month terms are normal in the US market and are not automatically a red flag — meaningful SEO outcomes rarely land in a quarter. What matters is what happens at the end, and what you keep.

• Notice period. Thirty to sixty days is standard. Ninety days after a twelve-month lock is not.
• Content ownership. Everything written for you should be yours on payment, without a licence clause that lets the agency reuse it for a competitor in the same city.
• Link and placement disclosure. You should receive a full log of placements built on your behalf. If the agency will not disclose them, you cannot audit risk after the relationship ends.
• Tooling and access handover. Analytics properties, Search Console, tag manager containers, and any dashboards should live in accounts you own, with the agency invited in — never the other way around.
• Documentation. Roadmaps, briefs, technical specifications, and tracking documentation transfer to you at exit.

Ownership clauses tell you more about an agency's confidence than any case study. Teams that expect to earn renewal do not need to hold your assets hostage.

Interrogate the case studies you were sent

Case studies are marketing artefacts, which makes them a document to interrogate rather than accept. Four numbers make one checkable: the absolute starting and ending non-brand organic sessions, the brand versus non-brand split, conversions or pipeline attributed with a stated model, and the retainer band. A 400% increase from a near-zero baseline is arithmetic, not achievement.

Ask for the engagement that did not renew. Every agency has churn, and the ones worth hiring can describe why a client left and what they changed afterwards. That answer predicts your second year far better than a testimonial wall does.

The five questions to ask on every call

1. 1What would you refuse to do for us in the first six months, and why?
2. 2Show us a client where results came in slower than forecast. What did you change, and when did you tell them?
3. 3Which of our competitors have you studied, and what are they doing that we are not?
4. 4What internal work do we need to own for this to succeed — approvals, developer time, subject-matter interviews?
5. 5What does month one look like, week by week, and what will exist at the end of it?

Score the answers, not the deck. Specificity beats polish every time: an agency that names your competitors, your CMS, and your two biggest template problems in the first call has done more work than one that arrives with a beautiful generic playbook.

Red flags that are worth walking away from

• Guaranteed rankings, guaranteed timelines, or a promise of map pack dominance across an entire metro.
• A refusal to name the people doing the work, or to put them in the SOW.
• Reporting that never separates brand from non-brand.
• Link building described only as 'high-authority placements' with no examples and no log.
• Pressure to sign before you have spoken to a current client of similar size and vertical.
• A proposal that would be identical if your company were in a different industry.

A shortlist process that takes about a week

1. 1Day one: write the bottleneck statement — which of the four work shapes you are buying and why — plus your primary metric.
2. 2Day two: build a longlist of six to ten agencies filtered by state, service focus, and vertical experience rather than by advertising spend.
3. 3Day three: cut to four using published evidence: case study specificity, transparency about pricing and team, and whether their own site demonstrates the discipline they sell.
4. 4Days four and five: run the calls with the five questions above, in the same order, so answers are comparable.
5. 5Day six: take two references each from the final two, including one that ended.
6. 6Day seven: compare SOWs on measurement definitions, named staffing, and exit terms — not on the deliverables table.

This is the sequence our state and service pages are designed to support. Build the shortlist from published facts first, then take the calls, and the pitch becomes what it should be: the last input to a decision you have already framed, rather than the first.