Thirty-eight thousand dollars. That’s roughly what twelve months of an average SEO retainer costs — industry pricing surveys put the typical monthly engagement a little over three thousand dollars — and it’s the number one marketing leader was staring at when she finally asked her agency a question she should have asked in month one: “Can you show me exactly what shipped?”

The answer was a slide with a visibility score on it. Not pages published, not fixes deployed, not rankings on the clusters that feed pipeline — a proprietary index, trending gently upward, explaining nothing. Thirty-eight thousand dollars had bought twelve months of activity she couldn’t audit, in a channel where results legitimately take six to twelve months to prove — which means a bad provider choice doesn’t fail fast. It fails silently, one polite monthly report at a time.

That’s why choosing an SEO services provider is a risk-management exercise before it’s a capabilities comparison. This guide gives you the seven questions that surface real answers, the red flags each one is designed to catch, and an honest framework for the decision underneath the decision: whether you should buy a team’s time at all — or build the capability internally with the right system.

Why Is This Decision So Expensive to Get Wrong?

Because SEO compounds in both directions, and the market’s structure hides failure for months.

The money is material. Industry pricing data converges on wide but consistent ranges: monthly retainers commonly run $2,000–$20,000 depending on scope and market, enterprise engagements reach $10,000–$50,000+ per month, hourly work typically falls between $100 and $300, and project-based work spans a few hundred dollars to $30,000. Contract norms match the channel’s physics — most providers ask for 6–12 month commitments, because early movement typically appears within 3–6 months and meaningful ROI within 6–12, sometimes longer in competitive categories.

Put those together and the risk profile is clear: you’re committing five to six figures to a channel whose feedback loop is measured in quarters. A mediocre provider doesn’t cost you a bad month; they cost you a year of runway plus the compounding you didn’t get.

And the delivery environment has gotten harder to evaluate from the outside. Providers are navigating SERP volatility, AI-driven content saturation, and the arrival of AI answer engines as a second visibility surface — pressures that push some agencies toward genuine process innovation and others toward opaque reporting that hides thin delivery. The seven questions below are designed to tell those two apart before you sign.

One reframe before the questions: you’re not just choosing a vendor. You’re choosing an operating model — a full-service agency that does the work, an in-house build, or a lean internal team running an execution platform like Iriscale. Each question below notes how the answer differs by model, because the right answer to “who should do our SEO” is sometimes “nobody external.”

The 7 Questions That Protect Your Investment

1. What outcomes are you accountable for — and what’s excluded?

Proposals emphasize outputs: pages optimized, links built, posts per month. Outputs aren’t outcomes, and your CFO will eventually ask about pipeline, not page counts. Force the accountability map into writing: which metrics the provider owns (audit completion, fix backlogs, content shipped, priority-cluster rankings), which dependencies are shared (dev cycles, approvals, tracking integrity), and which stay with you. Then ask what’s explicitly excluded — development work, CRO, analytics cleanup, and digital PR are common exclusions, and exclusions are fine; hidden exclusions are how blame games start in month eight.

Red flags: guaranteed rankings on any timeline (SERPs aren’t contractually controllable); hard deliverable commitments paired with vague outcome language; no mention of dependencies — which usually means they’re pre-building the excuse.

2. What’s the plan for the first 30/60/90 days?

Most failed engagements don’t fail on strategy; they fail on slow starts. A credible provider can walk you through days 1–30 (baseline measurement, technical crawl, analytics validation, quick wins, backlog creation), days 31–60 (remediation sprints, internal-linking and architecture work, first content wave), and days 61–90 (scaled production, refreshes of existing pages, reporting tied to leading indicators). If the first ninety days are mostly discovery meetings, you’ve spent a quarter of a twelve-month contract — potentially five figures — before anything shipped.

Red flags: month one contains research only; no prioritization method (impact versus effort, revenue pages first); no plan to validate conversion tracking before work begins — because untracked work is unprovable work.

3. How do technical fixes actually get implemented?

Technical SEO is where competence becomes visible, because it collides with engineering reality. The question isn’t “do you do technical SEO” — everyone says yes. It’s “show me how a finding becomes a shipped fix”: sample audit output, prioritization logic, how tickets get written, and what happens when our dev bandwidth is the bottleneck. Ask for their ship rate — high-impact fixes reaching production per month — because at $100–$300 per hour, a beautiful unimplemented audit is the most expensive PDF you’ll ever own.

Be equally honest evaluating platforms here, ours included: Iriscale is not a technical remediation tool. It plans your content architecture and internal linking structure so new pages ship structurally sound, but crawl fixes, canonical repairs, and performance work go through your developers regardless of which model you choose. Any provider — agency or platform — claiming technical SEO happens without engineering involvement is describing something you should investigate, not celebrate.

Red flags: technical work is “included” but implementation steps can’t be described; hundred-item issue lists with no sequencing; no verification step confirming fixes changed crawl or indexation behavior.

4. How do you produce, optimize, and govern content at scale?

Generative AI changed this question’s stakes: publishing more, faster is now trivial — and so is publishing sameness. Test whether the provider has a system: how they decide what to write (gap analysis, intent mapping, pipeline relevance — not “topics you approve from a list”), how they refresh existing pages (often faster ROI than net-new), what quality gates exist (fact-checking, brand voice, expert input), and how performance feeds back into the plan. For multi-product companies, ask specifically how they prevent cannibalization — two pages competing for one query is a governance failure that pure volume providers create constantly.

This is the layer where the platform model is strongest, and it’s fair to say so plainly: in Iriscale, Topic Strategy and the Keyword Repository decide what deserves to exist, Content Architecture prevents cannibalization by design, the Articles Hub runs brief-to-publish workflow with approval gates, and Brand Voice Guidelines keep fifty pieces sounding like one company. The point of AI in content isn’t that machines write everything — it’s that strategy and governance stop depending on whoever’s memory is best.

Red flags: “X posts per month” with no intent or refresh plan; no described QA process; no cannibalization prevention method.

5. How do you measure performance — and do we get raw access?

Reporting is where engagements die, often independent of results. The checklist: Do you retain access to raw data sources — analytics, Search Console — or receive only PDFs? Are KPIs tied to business outcomes rather than impressions and average rank? Is there an always-on view, or does insight arrive once a month, three weeks stale? And critically: is there an action log — what shipped, what changed — presented alongside performance, so movement can be attributed to work?

Align the ROI timeline explicitly too: early indicators at 3–6 months, stronger gains at 6–12. Reporting should mature along that curve — leading indicators early (index coverage, priority-cluster rankings, content velocity), pipeline signals later. And in 2026, insist the measurement covers both surfaces: Google rankings and AI engine visibility. If a provider isn’t tracking whether ChatGPT, Claude, Gemini, and Perplexity cite you — the way Search Ranking Intelligence does alongside traditional rankings — they’re reporting on a shrinking share of where your buyers actually decide.

Red flags: proprietary “visibility scores” as the headline metric; no action log; you don’t hold owner access to your own analytics accounts.

6. What’s the real cost model — and what protects us contractually?

The ranges are wide by design; the protections are what matter. Get plain-English answers on: minimum term and exit terms (6–12 month commitments are normal and defensible — if paired with milestone checkpoints); what survives a pause or cancellation; who owns the content, the accounts, and every login created during the engagement; and where add-on fees live — extra reporting, extra locations, extra revisions are the classic margin recovery levers.

The ownership question deserves special paranoia. Content published on your domain, analytics properties, and Search Console access should be yours unambiguously and in writing. Providers who resist this are telling you how the relationship ends.

Red flags: long lock-in with no milestone-based exit ramps; provider-owned accounts “for convenience”; pricing that scales through surprise hours rather than transparent tiers.

7. What risks should we plan for — and how do you prevent them?

Mature providers name their own risks unprompted; immature ones have a proprietary method they can’t describe. The risk categories worth hearing addressed: tactic risk (link schemes, spun content, doorway pages — penalty and brand exposure that outlives the contract), operational risk (work done “in the background” with no documentation), dependency risk (your dev cycles and approvals becoming their excuse), and drift risk (chasing algorithm news instead of revenue pages).

Ask for the prevention mechanisms in writing: a stated policy of what they won’t do, approval workflows for on-site changes, link quality standards that exclude private networks and disguised paid placements, and documentation norms. Governance sounds boring until you inherit a penalty from an agency you fired two years ago.

Red flags: “proprietary methods” in place of named tactics; refusal to share sanitized past work; no change-management process for touching your site.

Agency, In-House, or Platform: Which Model Fits?

The seven questions assume you’re evaluating providers. Step back one level first, because the honest answer differs by situation.

A traditional agency fits when internal bandwidth is genuinely near zero and you need end-to-end execution — strategy, content, outreach, coordination — done for you; when your category demands heavy custom stakeholder work (complex brands, regulated approvals); and when the budget comfortably supports market retainers for a 6–12 month runway. A good agency is a real answer. The seven questions exist to help you find a good one.

In-house plus a platform fits when you have at least one capable internal owner and your constraint is capacity and system, not judgment. This is the model Iriscale was built for: the strategy layer (Competitor Analysis, Keyword Repository, Topic Strategy, Content Architecture), the production layer (Articles Hub, Brand Voice Guidelines), the AI-visibility layer (AI Optimization Questions and Answers, Knowledge Base), and measurement across Google and five AI engines (Search Ranking Intelligence) — run by your team, with everything visible and auditable because it’s yours. The coordination overhead you’d pay an agency to manage largely disappears when the system is the coordination.

The hybrid is often the mature end-state: platform-run core execution with a specialist agency retained for what platforms genuinely don’t do — digital PR, technical remediation sprints, high-stakes migrations. That combination also keeps every provider honest, because your internal visibility means “mystery SEO” has nowhere to hide.

The platform model is not a fit if nobody internal can own prioritization and approvals even a few hours a week. Systems amplify an owner; they don’t replace one.

Is Iriscale Right for Your Team?

If you read the seven questions and kept thinking “we could answer these better ourselves if we had the system” — that’s the signal. Iriscale fits B2B SaaS teams with an internal owner who needs agency-grade strategy, content operations, and AI-era measurement without agency opacity or agency invoices: everything shipped is visible, everything measured spans Google and the five AI engines your buyers actually consult, and the entity and voice consistency an agency approximates with style guides is enforced automatically by the Knowledge Base.

It is not a replacement for developers on technical work, and it’s not the right model if you need a fully outsourced department. For everyone between those poles, the practical next step is seeing the system against your own site and market.

Book a demo and pressure-test the platform with your own seven questions →

Frequently Asked Questions

How much should SEO services cost in 2026?

Anchor on the ranges the market actually converges on, then judge fit rather than price. Industry pricing surveys consistently place monthly retainers between $2,000 and $20,000 depending on scope, competitiveness, and provider caliber, with typical engagements landing a little above $3,000 per month. Enterprise programs run $10,000 to $50,000+ monthly, hourly specialists bill $100–$300, and defined projects span a few hundred dollars to $30,000. Two judgment rules matter more than the numbers. First, price against the work plan, not the promise: a $2,500 retainer covering strategy, production, technical coordination, and reporting is arithmetic that should make you skeptical — someone is being paid very little to do a lot. Second, compare models, not just vendors: a platform subscription plus internal owner time often delivers more shipped work per dollar than a low-end retainer, while a high-end agency can be worth every dollar when you truly have no internal capacity. Cheap SEO is rarely cheap; it’s deferred cost.

How long does SEO take to show results?

Set the expectation in two phases and hold providers to matching evidence. Early indicators — improved index coverage, movement on priority keyword clusters, rising impressions, conversions on refreshed pages — typically appear within three to six months. Meaningful business results — organic pipeline, revenue-relevant traffic growth — typically take six to twelve months, and longer in genuinely competitive categories. Two nuances refine the picture. Refreshing existing pages usually moves faster than net-new content, which is why competent providers front-load refresh work for early proof. And the AI-visibility surface runs on a partially different clock: engines retrieving live web content can reflect well-structured pages within weeks, making AI citation movement an unexpectedly fast leading indicator when you’re tracking it. The management takeaway: demand leading-indicator reporting from month one, tie the six-month check-in to trend direction rather than absolute ROI, and treat anyone promising material results in weeks as describing either brand-term rankings or fiction.

What are the biggest red flags when hiring an SEO agency?

Five patterns account for most of the regret. Guaranteed rankings — nobody controls SERPs contractually, so a guarantee signals either naivety or an intent to redefine success later. Proprietary-method opacity — refusing to name tactics is how link schemes and spun content hide until the penalty arrives on your domain, not theirs. Provider-owned access — if your analytics, Search Console, or content live in their accounts “for convenience,” you’re learning how the relationship ends. Vanity reporting — proprietary visibility scores headlining reports, with no action log connecting work performed to movement observed. And discovery that never ends — a first quarter of meetings and audits on a twelve-month contract is 25 percent of your runway spent producing documents. One meta-flag ties them together: evasiveness under the seven questions in this guide. Strong providers enjoy those questions because their operations survive scrutiny. Weak ones reframe, generalize, or flatter. The interview behavior is the sample of the engagement.

Should we do SEO in-house instead of hiring an agency?

Decide based on ownership, not headcount. The in-house model works when at least one internal person can own prioritization, approvals, and the operating cadence — even part-time — because judgment about your business is the input agencies spend months approximating and your team already has. What used to make in-house impractical was the labor: research, architecture planning, production, and now multi-engine visibility tracking historically required either a team or a retainer. That’s the specific gap platforms closed. With Iriscale carrying competitor analysis, keyword and topic strategy, content production workflow, AI answer optimization, and measurement across Google plus five AI engines, a single capable owner runs what previously took a department or a $5,000 monthly retainer. The honest counter-case: if no one internal can give it consistent attention, in-house-plus-platform becomes shelfware-plus-guilt, and a well-chosen agency is the better spend. The deciding question isn’t budget — it’s whether an owner exists.

What should be included in an SEO contract?

Six elements protect you, and their absence is diagnostic. Scope with named exclusions — what’s covered, and explicitly what isn’t (dev work, CRO, digital PR), so month-eight disputes have a reference document. Milestone checkpoints inside the term — 6–12 month commitments are reasonable given SEO’s physics, but pair them with defined review gates and exit ramps if agreed work isn’t shipping. Unambiguous ownership — content, accounts, analytics properties, and all access created during the engagement are yours, in writing, surviving termination. Reporting specifics — cadence, metrics, raw data access, and an action log documenting work performed. Tactic standards — a written statement of practices they won’t use, particularly around link acquisition. And transition terms — what handover looks like at the end, because every engagement ends eventually. A provider who accepts these terms readily is showing you operational confidence. One who negotiates hard against ownership and transparency clauses is showing you something else.

Can an SEO platform really replace an agency?

For some functions completely, for others not at all — and precision here saves money in both directions. What a platform like Iriscale genuinely replaces: the strategy and research layer (competitor analysis, keyword intelligence, topic and architecture planning), content production workflow with governance, AI search optimization, and always-on measurement across Google and AI engines — the recurring core that constitutes most of a typical retainer’s hours. What it doesn’t replace: technical implementation (developers ship those fixes under any model), digital PR and relationship-based link earning, and the strategic counsel of a genuinely senior consultant on high-stakes calls like migrations or internationalization. The pattern that works for many teams is platform-run core execution with specialists engaged surgically for the exceptions — which typically costs less than a full-service retainer while giving you more visibility into the work. The wrong version of this decision is buying a platform with no internal owner, or paying an agency retainer for work a system plus four hours a week would cover.

How do we evaluate an SEO provider’s past results?

Ask for evidence structured as story, numbers, and mechanism — and probe each. Story: a client situation comparable to yours in category competitiveness and starting condition; results from an easy niche don’t transfer to a hard one. Numbers: business-relevant outcomes (organic pipeline, conversions, revenue-page rankings) over vanity aggregates, with timeframes attached — “traffic doubled” means little without knowing from what base and over how long. Mechanism: what they actually did — and this is the discriminating question, because providers who achieved results can describe the causal work in specifics, while providers riding a client’s product-led growth speak in generalities. Sanitized examples are fine and normal; refusal to show any work product is not. Two additional checks: references from engagements that ended, not just current happy clients, and in 2026, ask whether they measure AI engine visibility at all — a provider with no answer for how brands get cited in ChatGPT and Perplexity is optimizing for a shrinking share of your buyers’ journey.

What questions should we ask about AI in an SEO engagement?

Two categories, matching AI’s two roles in modern SEO. On AI as their tool: how do they use AI in production, and where are the human gates? The answer you want is specific — AI-assisted drafting inside briefs with named human accountability for accuracy — not defensive denial (which usually means undisclosed use) or volume boasting (which means sameness at scale). Ask how they prevent the quality collapse that unedited AI content produces, because Google’s systems and AI engines’ citation logic both punish it. On AI as a visibility surface: do they track whether AI engines cite your brand, across which engines, and how does that data change what they do next? Providers still treating “SEO” as Google-only are measuring a shrinking surface — surveys through 2026 show roughly a third of consumers now start searches in AI tools. A modern provider should speak fluently about extractability, entity consistency, and multi-engine measurement. If those concepts draw blank looks in the sales process, you’ve learned what you needed to.

Related Reading

• How to Scale Agency SEO Delivery With an Automation Stack
• Mastering SEO in 2026: A Checklist for Content Marketers
• The Best AI Tools for Digital Marketing Automation
• The Marketing Budget Allocation Framework for 2026
• What Modern Marketing Teams Need Next

© 2026 Iriscale · iriscale.com · AI-Powered Growth Marketing for B2B SaaS