Blog

Two years ago, hiring an SEO agency was a fairly boring decision. You compared rankings promised, backlinks delivered, and monthly retainers, then picked whoever seemed least likely to get you penalized. That checklist is now dangerously incomplete. People still search on Google, but a growing share of them never see a blue link — they read an AI Overview, ask ChatGPT for a shortlist, or let an agent do the comparison shopping. Search Engine Journal’s State of SEO survey work has been tracking the same shift from the practitioner side: org charts, budgets and success metrics are all being rewritten at once.

So when an agency puts “AI SEO” on its homepage in 2026, that phrase can mean three completely different businesses. It might mean they use AI tools internally to produce more work per hour. It might mean they optimize your site to get cited inside AI answers. Or it might mean nothing at all beyond a rebrand. Your first job as a buyer is to figure out which one you’re talking to.

1. Make them define “AI SEO” before you do

Open the call with an open question: what does AI SEO mean in your practice, specifically? Don’t lead. A strong answer separates two workstreams — traditional organic search (rankings, crawlability, content quality, links) and AI visibility (whether ChatGPT, Perplexity, Gemini, Copilot and Google’s AI surfaces mention and cite you) — and explains how they feed each other. A weak answer collapses everything into “we use AI to write content faster,” which is an internal efficiency story, not a growth story. Faster content production is the easiest thing in the world to buy in 2026 and the least defensible thing to pay a premium for.

Watch for agencies that treat AI visibility as an entirely separate channel with its own budget line and its own tactics. As Forbes’ Communications Council argued this year, AI answers largely reflect the strength of your existing search presence and your reputation elsewhere on the web. Publishing more pages doesn’t buy you citations; it usually just adds noise.

2. Ask what they measure — and how they’d prove it

This is where most pitches fall apart, because attribution in AI search is genuinely hard. LLM referrals often arrive disguised as direct traffic or branded search, and Google’s AI Overviews still don’t show up as a distinct referrer. Backlinko’s breakdown of LLM visibility is a good primer on why your analytics can show decline while your brand influence is climbing.

An agency that has actually done this work will say something like: we’ll baseline your share of voice across AI platforms on a fixed prompt set, track branded search volume and direct traffic as proxies, tag AI referrers where they exist, and report citation counts by platform. An agency that hasn’t will promise you “AI rankings,” which aren’t a thing — AI answers are non-deterministic and vary by user, session and phrasing.

Also ask about the boring metric that still pays the bills: pipeline. Impressions, mentions and citations are leading indicators. Demos, trials, and revenue are the scoreboard. If nobody on the call mentions your funnel, they’re selling reporting, not results.

3. Look at their tooling — and their data sources

You’re not buying software, but the tools an agency lives in tell you how they think. Ask which AI visibility platform they use and, crucially, where its prompt data comes from. There’s a meaningful difference between tools that monitor a handful of prompts you invent yourself and tools built on large, search-derived prompt sets — Ahrefs’ Brand Radar is the common example of the latter, and their team has published a practical set of use cases worth reading before the call so you can ask sharper questions. Semrush maintains a reasonable comparison of the AI visibility tool landscape if you want a second opinion on the category.

Two red flags: an agency that can’t name its stack, and an agency whose entire deliverable is a tool dashboard with no human interpretation on top. You can buy a dashboard for a few hundred dollars a month. What you’re paying an agency for is the decision that follows it.

4. Test their content standards, not their content volume

Ask for three pieces of content they produced in the last six months, and ask who wrote them, who reviewed them, and what the byline strategy was. Then ask what they’d refuse to publish.

That last question is the tell. Agencies with real editorial standards have opinions about first-hand experience, original data, expert review and where AI drafting stops. Google’s own guidance on AI-generated content has been consistent: the issue isn’t how content is made, it’s whether it’s helpful, original and demonstrably credible. Since AI systems tend to cite sources that are already trusted and frequently referenced, thin AI-assembled content is doubly wasted — it doesn’t rank and it doesn’t get quoted.

5. Check the shape of the engagement

Practical due diligence, in rough order of how often it saves people money:

• Who does the work? Meet the strategist who’ll actually be on your account, not just the founder who closes.
• What’s the ramp? Anyone promising AI-search results in 30 days is guessing. Three to six months to a defensible baseline is honest.
• Is there a technical layer? AI crawlers need to reach and parse your pages. If nobody asks about your rendering, site structure or bot access, that’s a gap.
• Do they know your category? AI answers are topic-clustered. An agency that already understands your vertical’s language starts closer to the target.
• Contract terms. Month-to-month after an initial 90-day commitment is fair. Twelve-month lock-ins with no exit clause are not.
• Reporting cadence. Monthly, with commentary and a next-90-days plan — not an automated PDF.

6. Reference-check for behavior, not results

Ask two current clients one question: what happened when something didn’t work? Every AI SEO agency has losing quarters. The good ones diagnose, tell you early, and change the plan. The rest go quiet and resend the same dashboard.

How to Identify the Right AI SEO Agency?

The right AI SEO agency in 2026 treats classic SEO and AI visibility as one converged practice, measures both honestly, has editorial standards it will defend, and ties its work to your pipeline rather than to mention counts. The wrong one has swapped a few words on its website and added a tool subscription to your invoice. The difference is usually visible within twenty minutes of the first call — if you ask the right questions.