What Should You Verify Before Selecting an SEO Company?

Before relying on a pitch deck, ask the company to discuss its own website and acquisition approach. A live self-review is not a perfect competence test, and a provider may reasonably prioritize client delivery over its own marketing, but the exercise reveals how the team reasons, communicates, and handles inconvenient evidence.

Give the company advance notice of the subject while avoiding a request for a polished presentation. The goal is not to surprise staff or manufacture embarrassment. Ask for the person who would lead strategy or quality review on your account to participate.

Begin with positioning. Ask which audience the company serves, which problems it is trying to be discovered for, and how the site supports the sales process. A credible answer should distinguish brand demand, educational traffic, service demand, referrals, partnerships, and other channels rather than claiming that all growth comes from organic search.

Review information architecture. Ask the team to explain the role of the homepage, service pages, articles, case studies, locations when genuine, and contact paths. Look for clear page ownership, useful internal links, and a rationale for which topics deserve separate pages.

Review technical decisions. Discuss indexation, canonical patterns, redirects, performance, rendering, sitemaps, structured data, and measurement at a level appropriate to the engagement. The provider should be able to explain what is controlled, what is monitored, and what remains a tradeoff.

Review content quality. Select a public article and ask how the topic was chosen, which sources were used, who reviewed it, how AI assistance is governed, how corrections are handled, and what business or reader decision the page supports. Do not equate a ranking with factual or editorial quality.

Review authority activity. Ask how the company earns mentions, relationships, links, and brand discovery. Require a distinction between editorial work, sponsorship, partnerships, directories, public relations, and other methods. The team should explain relevance and risk rather than relying only on third-party authority scores.

Discuss weaknesses openly. A provider may have an outdated page, an underdeveloped content area, or a deliberate choice not to target a query. The useful signal is whether the team recognizes the issue, states the reason, and explains the decision process.

The exercise passes when the company demonstrates coherent reasoning, evidence discipline, technical fluency, and intellectual honesty. It fails when staff become evasive, misrepresent visible facts, use tool scores as final proof, or refuse to explain methods that would affect your brand.

Compare serious candidates using the same agenda. Two or three observations across providers can reveal meaningful differences, but the count is an operating example rather than a statistically valid sample.

If a company's own site is weak but the explanation is plausible, request delivery samples and references from work closer to your scope. Do not accept self-review alone as proof or disqualification.

Ask each provider what it will be accountable for and what remains outside its control. Rankings, impressions, clicks, qualified leads, revenue, assisted conversions, retention, and brand demand answer different questions. A serious engagement defines the business question first and then selects supporting indicators.

Do not demand a revenue guarantee from an SEO company. Revenue depends on product, price, sales process, inventory, seasonality, attribution, implementation, and other channels. The company should nevertheless explain how search work is expected to support the customer journey and how the client will evaluate that contribution.

Start with the conversion definition. Identify the action that represents progress: a purchase, qualified inquiry, booked demonstration, subscription, application, visit, download, or another event. Confirm that tracking exists and that the event has enough quality information to distinguish valuable actions from noise.

Map search intent to pages. For every priority topic, identify the likely reader, decision stage, page type, offer, and next action. A high-volume query may be useful for awareness but irrelevant to near-term pipeline. A low-volume query may be valuable if it describes a specific commercial need.

Review the reporting sample. It should connect implemented work with observable effects while avoiding causal overstatement. Look for page-level and query-level analysis, conversion quality, implementation status, technical findings, content performance, authority activity, blockers, and next decisions.

Use a scenario question: if traffic increased substantially while revenue remained flat, how would the company investigate? A credible answer reviews query mix, landing pages, conversion tracking, page experience, offer alignment, lead quality, sales follow-up, brand effects, attribution, and seasonality. It does not simply declare traffic a success or accept responsibility for every commercial issue.

Define the client's responsibilities. The client may own analytics, customer data, sales feedback, landing-page approval, pricing, inventory, engineering, or expert review. If those inputs are missing, the provider should state which conclusions cannot be made.

Agree on a baseline and observation plan. Record the starting period, known tracking changes, releases, market events, and implementation dates. Avoid comparing incompatible periods or interpreting a single ranking movement as strategy validation.

The measurement review passes when the provider explains how search indicators relate to the business without pretending complete attribution. It fails when reporting stops at vanity metrics, invents revenue causation, or ignores the quality of the visitors and actions produced.

If the business cannot measure conversions reliably, make measurement repair part of discovery. Do not sign a performance-linked agreement until the data, control, attribution rules, and dispute process are sufficiently clear.

A proposal should translate limited discovery into a transparent initial plan. It should not disguise uncertainty or promise a complete strategy before the provider has access to the necessary data.

Red Flag One: a specific result or timeline guarantee without a site and market baseline. Search outcomes depend on variables the company cannot fully control. A provider can commit to deliverables, response standards, review quality, and implementation support, but a page-one promise without evidence should trigger deeper scrutiny.

Ask what the forecast assumes: domain history, existing authority, competition, content quality, implementation speed, technical condition, market demand, and client resources. Require the company to distinguish a planning hypothesis from a contractual guarantee.

Red Flag Two: a deliverable-heavy plan without strategic logic. The source used four blog posts and twelve backlinks as an example. Preserve those values as an editorial illustration, not as a recommended production mix. Every output should have a purpose, acceptance standard, owner, dependency, and connection to the diagnosed constraint.

Ask why the proposed content exists, which intent it serves, how topics avoid overlap, who supplies expertise, how facts are reviewed, and what happens when the client cannot approve production at the planned rate. For authority work, require sourcing, relevance, editorial, disclosure, and risk controls.

Red Flag Three: no discovery or diagnostic phase. An initial proposal may contain hypotheses, but it should state which data is missing and how the plan may change. Discovery can include technical access, analytics validation, customer interviews, content inventory, competitor review, conversion analysis, and stakeholder alignment.

Review staffing. The proposal should identify roles, responsibilities, quality reviewers, subcontracting, and the process for team changes. Do not assume the senior seller performs the work.

Review implementation. Determine whether the provider writes recommendations, deploys changes, coordinates developers, reviews releases, or only reports issues. A recommendation cannot be evaluated if nobody owns implementation.

Review exclusions and change control. Identify out-of-scope systems, revision limits, content approvals, legal review, emergency work, and additional fees. Vague scope often becomes conflict after the contract begins.

The proposal passes when strategic logic, discovery, staffing, deliverables, dependencies, implementation, measurement, and commercial terms are inspectable. It fails when activity volume substitutes for diagnosis or when assumptions are presented as established facts.

When several issues appear together, request a revised proposal before negotiating price. A cheaper version of an unclear plan remains unclear.

Authority work should support truthful expertise, useful information, reputable relationships, and clear entity understanding. It should not be reduced to a third-party domain score, a link quota, or the repeated use of EEAT terminology.

Begin with content authority. Ask how the provider identifies the questions a qualified reader needs answered, obtains direct experience or expert input, verifies claims, selects sources, handles updates, and differentiates a central guide from supporting pages. Review a complete brief and a finished article from the proposed production process.

Assess authorship. Determine whose name appears, what experience supports the content, who reviews regulated or technical statements, and how corrections are made. Do not accept invented first-hand experience, generic biographies, or markup that claims expertise absent from the visible page.

Assess link and mention acquisition. Ask which relationships and editorial opportunities are pursued, how relevance is evaluated, whether compensation or sponsorship is involved, how placements are documented, and what methods are prohibited. Link quality depends on context and editorial legitimacy, not simply volume or a tool score.

Assess site architecture. Strong subject coverage requires clear page roles, descriptive internal links, non-duplicative intent, accessible navigation, and a maintenance plan. Publishing many related pages without boundaries can create competition and thinness rather than authority.

Assess brand and entity clarity across the web. Mentions, profiles, references, and structured information can help users and systems understand the organization, but no specific markup or profile activity guarantees ranking or inclusion in Google AI Overviews or other Google AI features.

The source used enterprise authority as a label. Treat it as a broad concept describing brand recognition and cross-web credibility, not a documented ranking factor or a required service for every business.

Ask the company to distinguish owned, earned, paid, and partner-driven activity. Require disclosure and approval rules that fit your brand, industry, and legal obligations.

The review passes when the provider can demonstrate source discipline, real subject input, editorial quality, relevant relationships, technical consistency, and honest limitations. It fails when authority is treated as a purchasable score or content is produced without accountable expertise.

The source claimed authority-led work compounds over 12-24 months. No supporting URL was provided, so preserve that range as a historical planning statement requiring reconciliation rather than a forecast or causal claim.

If the company's sample content looks strong but production quality is uncertain, meet the editors, request the briefing and review workflow, and define approval criteria in the initial engagement.

A buyer should control part of the sales agenda. Use the same questions across candidates and record the answers so presentation style does not replace due diligence.

Question one: how do you prioritize search opportunities for a business like ours? A strong answer considers customer intent, page type, commercial relevance, existing performance, competition, implementation, and evidence quality. A weak answer begins and ends with volume.

Question two: show a content asset you consider successful and explain the reader need, source process, structural decisions, distribution, measurement, and limitations. Confirm whether the proposed team produced it.

Question three: describe a recent permitted authority or link example. Ask how the opportunity was sourced, why it was relevant, whether payment or sponsorship applied, what editorial control existed, and how the result was documented.

Question four: what would you advise us not to do? The answer should identify a specific mismatch, risk, or opportunity cost rather than repeat a generic warning about prohibited tactics.

Question five: what would failure or an inconclusive result look like, and how would you diagnose it? Ask for early indicators, escalation, client dependencies, decision rules, and the circumstances that would cause the provider to stop or change direction.

Question six: which technical changes do you implement, which do you specify, and who validates deployment? Both advisory and implementation models can work when responsibility is clear.

Question seven: who performs strategy, analysis, writing, editing, links, project management, and quality review? Meet critical delivery staff before signing and confirm how substitutions are handled.

Question eight: how does the team keep guidance current? Look for primary documentation, internal testing, incident review, peer discussion, and dated decision records rather than a claim of always knowing algorithm changes.

Question nine: what happens during onboarding? The answer should cover access, analytics, discovery, stakeholders, risks, baseline, priorities, communication, and the initial delivery plan.

Question ten: what client behavior most improves or harms the engagement? A thoughtful answer reveals how the provider handles approvals, access, implementation, expertise, and disagreement.

Ask follow-up questions when answers rely on broad claims. Request an example, artifact, or process document where confidentiality permits. Do not demand disclosure of another client's private data.

The interview passes when answers are specific, consistent with the proposal, and supported by work samples or documented processes. It fails when sales staff cannot identify the delivery team, methods remain hidden, or accountability disappears behind confident language.

If candidates provide comparable answers, run a paid working session or small diagnostic to observe collaboration before committing to a larger scope.

Price should be evaluated after the buyer understands the objective, scope, team, responsibilities, implementation, and measurement. A high fee can reflect senior attention, broad production, brand positioning, or overhead. A low fee can reflect focus, efficiency, reduced scope, junior staffing, or hidden omissions.

The source described four common models. A monthly retainer can support ongoing strategy, production, analysis, and implementation coordination. It requires prioritization rules, service levels, staffing, review cadence, and a clear method for changing the work as evidence develops.

Project pricing can fit technical audits, migrations, content architecture, measurement repair, or another bounded decision. Define assumptions, inputs, deliverables, acceptance criteria, revisions, implementation support, and the transition after completion.

Performance-linked pricing requires reliable attribution, a shared baseline, sufficient provider control, access to data, and a dispute process. The easiest metric to measure may not represent commercial value, and the provider may not control product, sales, inventory, or client implementation.

Hybrid pricing combines a base fee with a variable component. Examine whether the variable component is material, whether the trigger can be audited, and whether it encourages useful behavior.

Hourly billing can be appropriate for advisory work, investigation, training, or uncertain incidents. It becomes problematic when a buyer expects a defined outcome but scope remains unmanaged.

Review contract length separately from quality. The source used twelve months or more, the first ninety days, and month six as examples. Those tokens are preserved as historical commercial illustrations without a supporting URL. Select a term that matches the work, procurement requirements, implementation cycle, and available evidence.

Inspect termination, renewal, notice, fees, data access, intellectual property, confidentiality, subcontractors, conflicts, non-solicitation, warranties, liability, and transition support with qualified counsel where appropriate. Confirm that the client retains access to its own accounts and data.

Normalize proposals. Compare total expected cost, staffing, deliverables, client effort, tools, production, implementation, revisions, meetings, risk, and likely follow-on work. Do not compare only the monthly headline.

The commercial review passes when the fee supports the delivery model and both sides can understand how scope, performance, and exit work. It fails when recurring revenue is protected by vague obligations or when payment depends on unverifiable attribution.

If a price difference remains unexplained, request the staffing and scope assumptions behind it. Do not infer quality from price in either direction.

Positive indicators are useful when they are connected to observable behavior and work samples. Avoid turning them into another superficial checklist.

Green flag one: the company states what it does not do. Clear limits around industries, implementation, link methods, content types, or analytics can reduce risk. Verify that exclusions appear in the proposal and contract, not only in conversation.

Green flag two: its public presence is coherent. The site should communicate audience, services, evidence, ownership, and next steps. Dominant rankings are not required, but the strategy should be understandable and the content should meet the quality standard promised to clients.

Green flag three: the company challenges assumptions with evidence. Pushback should be respectful and specific. Automatic agreement may indicate sales pressure, while contrarian language without evidence is not strategic value.

Green flag four: the provider connects recommendations to the business model. It asks about customers, conversion, deal value, sales cycle, implementation, measurement, and constraints before prescribing production.

Green flag five: diagnosis comes before a broad commitment. A paid audit, discovery sprint, or proportionate assessment can reduce uncertainty. A free review may also be useful when its scope and data boundary are clear.

Green flag six: communication distinguishes commitments from forecasts. The provider explains which deliverables it controls, which outcomes depend on external variables, and how the plan will be adjusted when evidence changes.

Additional signals include clear documentation, client-owned accounts, correction procedures, transparent subcontracting, consent-based references, secure access practices, and an implementation tracker.

Validate each signal. Request artifacts, meet the team, compare statements with the proposal, and confirm reference permission. A sales claim about transparency is not evidence until the operating process reflects it.

The review passes when positive signals remain consistent across sales, samples, contract, onboarding, and delivery-team conversations. It fails when the pitch promises senior strategy but the proposal assigns a different model or hides methods.

If a provider has strong positives and a few unresolved concerns, negotiate evidence-building checkpoints or a smaller initial project rather than ignoring the concerns or rejecting the provider reflexively.

After the calls, proposal review, work samples, and commercial analysis, compare finalists across four qualitative areas. Do not create false precision with arbitrary weighting when the underlying evidence is incomplete.

Area one is strategic fit. Does the proposed sequence address the actual site condition, customer intent, competitive situation, and business objective? Which assumptions remain untested, and how will discovery change the plan?

Area two is communication integrity. Did the team acknowledge limits, correct misunderstandings, explain uncertainty, and take responsibility for mistakes? Review both written and verbal communication.

Area three is team quality and access. Identify the strategist, analysts, technical reviewers, writers, editors, authority specialists, project manager, and subcontractors. Confirm allocation, senior oversight, and substitution rules.

Area four is structural accountability. Review deliverables, implementation tracking, quality checks, data, reporting, change control, review points, termination, and transition. Accountability should exist during delivery, not only at renewal.

Create a decision memo for each finalist. Include strengths, risks, missing evidence, reference notes, staffing, commercial terms, client responsibilities, conflicts, and the reason the provider may be unsuitable.

If two or three candidates remain close, use a paid working session, diagnostic, or deliverable review with the actual team. Define the task and confidentiality so the session tests collaboration without requesting unpaid speculative work.

Consult internal stakeholders who must implement or approve the work. Engineering may identify technical feasibility issues, legal may identify contract or claims risks, sales may clarify lead quality, and content teams may assess review capacity.

Choose the option you can explain clearly to a skeptical stakeholder. A defensible rationale should rely on fit, evidence, team, operating model, and contract rather than urgency, familiarity, or presentation.

The decision passes when material assumptions are documented, the client can meet its responsibilities, and the initial scope creates useful evidence. It fails when the choice depends on gut feel while staffing, methods, data, or exit remain unclear.

If no candidate passes, do not hire one for the sake of momentum. Improve readiness, revise the brief, or seek a narrower specialist.

The usual advice is to inspect case studies, reviews, references, and the provider's position on risky tactics. These checks are useful, but each has limitations. Case studies are selected by the seller, reviews may reflect only a subset of clients, references are normally positive, and broad statements about ethics may not reveal the methods used in daily delivery.

The deeper problem is incentive and operating-model fit. A monthly provider may reasonably value retention, while the client may expect a defined problem to be solved and the relationship to change. A team reporting rankings may believe it is succeeding while the buyer expects qualified opportunities.

A content operation may optimize production efficiency while the client requires expert review and careful evidence handling.

Do not assume that commercial incentives automatically produce bad work. Instead, make them visible. Ask how priorities are selected, how unused budget or delayed implementation is handled, what triggers a scope change, how the company reports unsuccessful experiments, and whether completion of a project can lead to a smaller engagement rather than automatic renewal.

Another common mistake is treating all SEO companies as interchangeable. Some specialize in technical remediation, migrations, local search, content systems, digital PR, large-site governance, analytics, or advisory work. A provider can be highly capable and still be unsuitable for the problem you have now.

Buyers also overlook internal constraints. A company cannot publish accurate expert content without access to expertise. It cannot validate conversions without analytics and sales data. It cannot implement template changes without engineering support. It cannot resolve executive disagreement about the offer or audience.

The correct question is not which seller appears most credible. It is which operating model, team, evidence, and commercial structure best match the current diagnosis, and whether both sides can fulfill the responsibilities required for the work to produce useful evidence.

SEO companies can become highly sophisticated at technical analysis, content production, and authority work while remaining weak at connecting those activities to the client's commercial decision. Buyers can also create failure by entering without a clear objective, reliable measurement, implementation owners, or an agreed evidence boundary.

A better relationship begins with mutual accountability. The provider explains assumptions, methods, responsibilities, and uncertainty. The client supplies access, customer context, expert review, implementation, and timely decisions. Both sides distinguish activity from deployment and deployment from observed outcome.

The most useful standards are the ones a provider would accept for itself: real work samples, clear staffing, transparent methods, documented corrections, appropriate claims, client-owned data, and a contract that does not hide risk.

Selecting an SEO company should therefore produce more than a signed agreement. It should produce a shared operating model that can detect when the original diagnosis is wrong, adapt without concealing failure, and end responsibly when the work is complete or the fit no longer exists.

Days 1-3

Define the business objective, customer group, conversion event, search opportunity, known constraints, internal owners, and the evidence that would support a useful decision before contacting providers.

Outcome: A one-page buying brief describing the target customer, commercial objective, site condition, implementation capacity, measurement, and required provider capabilities.

Days 4-7

Create a shortlist of four to six companies through referrals, public research, industry contacts, and relevant search discovery. Inspect each provider's public work, ownership, methods, and focus before outreach.

Outcome: A candidate list with an initial record of public evidence, delivery focus, content quality, authority practices, and unresolved questions.

Days 8-14

Conduct structured calls using the ten buyer questions, review commercial alignment, and request a live self-review with the top two or three candidates when they agree to the format.

Outcome: A comparable record of diagnosis quality, team access, accountability, commercial understanding, and fit with the actual business problem.

Days 15-21

Review proposals against the three risk areas, request permitted work samples, verify subcontracting and quality controls, meet the delivery team, and document missing evidence or contract concerns.

Outcome: Two or three finalists with a clear comparison of strategy, staffing, scope, methods, implementation, measurement, and commercial risk.

Days 22-28

Compare finalists across the four decision areas. When two options remain close, commission a paid working session or bounded discovery task before a larger commitment.

Outcome: A documented selection decision based on evidence, team, fit, accountability, and client readiness rather than presentation or price alone.

Days 29-30

Finalize scope, milestones, data access, implementation tracking, reporting, change control, intellectual property, review points, and exit terms before onboarding begins.

Outcome: A signed engagement with clear responsibilities, evidence standards, and a ninety-day review point stated in words rather than treated as a guaranteed result period.