AI for SEO? Nope. Why I Won’t Let AI Fully Run My or My Client’s SEO

TL;DR: AI is a great assistant for SEO work and a terrible replacement for SEO judgment. I’ve been doing SEO for more than 20 years, and I use AI every day. It still can’t win my most important keywords or get my brand cited in ChatGPT and AI Overviews. And even if it could, I wouldn’t hand it the keys. AI can do pieces of your SEO. It can’t do your SEO.

—-

“Can’t we just have AI do our SEO?”

I hear some version of this question almost every week these days… It’s exhausting.

Sometimes I get this from a founder on a sales call. Sometimes it’s a marketing manager whose CFO just asked why they’re still paying an agency when ChatGPT is $20 a month. It usually comes up as a budget question, and I understand why. Your margins are tight, customer acquisition costs keep climbing, and every tool on the market claims it can automate your growth.

Wanting a faster, cheaper answer is completely reasonable. The mistake is assuming the hard part of SEO is the part AI does.

So, can it?

Here’s my answer, in one place, so I can point to it: NO.

Not “not yet.”

Not “maybe with better prompts.”

No.

For context, I’ve been doing SEO for more than 20 years. I’ve lived through hundreds of algorithm updates. I’ve watched thousands of sites get wiped out for trying to game the system.

I even build and update AI SEO tools for my team at Stryde all the time. If anyone should be able to build and hand SEO over to a machine, it would be someone like me. I have the experience, the data, the tooling, and a strong reason to make it work.

Newsflash: It doesn’t even work for my own business.

And I am ruthless with my own SEO. That’s why we rank #1 (and sometimes #2) for ecommerce digital marketing agency on Google.

AI didn’t get us there. It wouldn’t have gotten us there alone. Experienced people making hard calls did, with AI helping along the way.

What AI Is Actually Good At

Let me be clear: I’m not anti-AI. We use it every single day, and it has made my team meaningfully faster.

Here’s where it earns its keep:

• Data analysis. It can work through Search Console exports, crawl data, and rank tracking in minutes, and spot patterns across thousands of URLs that would take a person days to find.
• Keyword clustering. It can group thousands of queries by topic and intent as a starting point.
• First drafts. Content briefs, meta descriptions, schema markup, and outlines all get to a usable starting point much faster.
• Reporting. It can turn a messy spreadsheet into a readable summary a client can actually follow.

Let’s look at a real example from our team… We have a competitive SEO content analysis agent that we use to quickly analyze our client vs their top three online competitors. It looks at their blog content, looks at the helpfulness, depth, linked resources, etc. and helps identify content on their site that needs to be beefed up or gaps that need to be closed.

This used to take our team 3 hours. The tool does it in 15 minutes so our team can then spend more time digging into the data and making educated decisions on how to move forward… and spend more time on building actual helpful content that will perform vs collecting data all day.

That’s real value. But notice what’s on that list. Every item is a task. And SEO was never a list of tasks.

SEO Isn’t a Task List. It’s a Chain of Decisions.

This is the part most people miss.

Think of AI as a calculator, not a CFO. A calculator can produce any number you ask for, instantly and accurately. What it can’t do is tell you which number matters, what it means for your business, or what to do about it. That’s the CFO’s job, and it’s the job that actually determines whether the company makes money.

SEO works the same way. Every result you get from search, good or bad, comes from a long chain of small decisions:

• Which keywords to target, and which to ignore
• Which pages to build, which to merge, and which to leave alone
• What to fix first when there are 400 issues in an audit
• When to act, and when doing nothing is the right move

Each of those decisions depends on the ones before it. One bad call early on, like targeting the wrong keywords or restructuring the wrong section of the site, means everything after it is built on a bad foundation. AI can carry out every step perfectly and still take you somewhere you never should have gone.

Where AI Falls Down (Especially for Ecommerce)

For the ecommerce brands we work with, the gaps show up in predictable places.

It can’t tell revenue from volume. Your store probably ranks for thousands of keywords. Maybe 20 of them actually pay the bills. AI will happily optimize for whatever has the most search volume, because it doesn’t know your margins, your average order value, which products you’re overstocked on, or which ones bring back repeat customers. The keywords that matter most are often not the biggest ones.

It doesn’t know your business. Seasonality, inventory, upcoming launches, the reason your best seller sells, the objection your customer service team hears every day. None of that is in the data AI is looking at, and all of it should shape your SEO strategy.

It makes architecture decisions that can sink a store. Collection pages, faceted navigation, canonical tags, and redirects during a replatform are where ecommerce SEO is won or lost. These decisions can quietly undo years of rankings, and AI makes them with the same confidence it uses to write a meta description.

It doesn’t know when it’s wrong. This is the one that worries me most. AI output looks just as polished and certain when it’s wrong as when it’s right. There’s no hesitation and no “I’m not sure about this one.” The only protection is someone with enough experience to spot the problem.

It makes everyone look the same. If every one of your competitors runs the same tools on the same data, everyone ends up with the same content, the same structure, and the same recommendations. Sameness doesn’t win in search. It just gets you lost in the crowd.

Real-World Scenario

A $6M home goods brand on Shopify runs an AI-powered audit in early October. The tool flags three dozen “thin” collection pages, like small, specific groupings of throw pillows, table linens, and seasonal decor, and recommends consolidating them into a handful of broad categories. On paper, it’s tidy. The logic sounds right, and the report looks professional.

In reality, those small collection pages rank for the long-tail gift searches that drive a big share of the brand’s Q4 revenue. Consolidating them six weeks before the holidays would wipe out those rankings right when they matter most.

An experienced SEO catches that in 30 seconds. Someone without that experience ships it and finds out in January.

AI Tools Don’t Run Themselves

There’s another part of this that nobody mentions when they pitch “AI SEO”: someone has to keep the tools working.

At Stryde, we run 20 different AI agents across our SEO, content, and reporting work. Every one of them needs regular fixes and updates. In the last month alone, we’ve pushed dozens of updates to those agents. That’s real hours, every week, from people who could be doing client work.

Here’s why that maintenance never ends:

• The models change. When an AI provider updates its model, an agent that worked perfectly last month can start giving different, and sometimes worse, answers with no warning.
• Search changes. Google updates its algorithm constantly, and AI search tools change how they find and cite sources. An agent built on last year’s rules gives last year’s advice.
• The data changes. Tools, APIs, and data sources update their formats, and agents break quietly when they do.
• The output drifts. Small errors add up over time, and nobody notices until a recommendation goes out that shouldn’t have.

And here’s the catch: the person maintaining these tools has to know SEO. You can’t tell an agent’s output has gone bad unless you know what good looks like. A developer can fix broken code, but only an experienced SEO can tell you the agent is now confidently recommending the wrong thing.

So even the “just use AI” approach needs an expert. You’ve just moved that person from doing the SEO to supervising the machine that does it.

Even If the Tool Existed, I Wouldn’t Let It Run

People are often surprised by this part.

Let’s say someone built an AI tool tomorrow that could do everything I just described. It understands your margins, your seasonality, and your architecture, and it makes great decisions most of the time.

I still wouldn’t let it run my SEO.

The keywords that drive my business are too important to hand to something that can’t explain its reasoning or take responsibility when it’s wrong. When a strategist on my team makes a call, I can ask why. I can push back. I can find out what they saw, what they weighed, and what they decided to ignore. If they’re wrong, they own it, we learn from it, and it shapes the next decision.

AI has no accountability. It doesn’t lose sleep over a traffic drop. It doesn’t have to explain a bad quarter to a client. It doesn’t care whether your business does well.

Trust comes from accountability, and I’m not going to trust my most valuable keywords to something that has none. I’d rather be slower and right than fast and guessing.

AI Search Makes Expertise Matter More, Not Less

Here’s the irony. The rise of AI search is the strongest argument against handing your SEO to AI.

To show up in ChatGPT, Gemini, Perplexity, and Google’s AI Overviews, your brand has to be a source those tools trust enough to cite. And what earns that trust? Authority, a consistent brand message across the web, original expertise, real customer proof, and other credible sites confirming what you say about yourself.

AI-generated content can’t be the source of AI answers. It’s a rearranged version of what those answers already contain. If your content says the same thing as everyone else’s, AI has no reason to cite you over anyone else.

This is why I keep saying that strong SEO fundamentals are the foundation of visibility in both traditional and AI search. AEO and GEO build on top of those fundamentals. They don’t replace them. And the fundamentals that matter most, like authority, trust, and real expertise, come from people.

The Lie / The Truth / What to Ask Instead

The Lie: AI can do SEO now, so we don’t need an expert.

The Truth: AI can do SEO tasks. Someone still has to decide which tasks matter, in what order, for what reason, and catch it when the AI is wrong.

What to Ask Instead: “Who on our team, or at our agency, knows enough to tell when the AI is wrong?”

The Model That Actually Works

The brands winning in search right now aren’t choosing between AI and experts. They have real subject matter experts using AI as leverage.

That’s how we work at Stryde. AI handles the heavy lifting on data and first drafts. Experienced strategists make the decisions, check the output, and connect every recommendation back to revenue. The AI makes us faster. The people make us right.

Here’s how I put it: AI makes a great SEO faster. It also makes a bad SEO wrong faster.

If you’re evaluating your own setup, or an agency’s, ask these three questions:

1. Who reviews AI output before anything goes live? Get a name, not “the team.”
2. What would they reject, and why? If they can’t give you examples, nobody is really reviewing it.
3. How does each recommendation connect to revenue? If the answer is traffic or rankings alone, keep asking.

So, can AI do your SEO?

It can do pieces of your SEO. It can’t do your SEO.

If you want to see how an experienced team uses AI to move faster on the work that actually drives revenue, let’s talk.