How AI may increase the value of SEO expertise

By now, you’ve heard the doom and gloom.

SEO is a white-collar job. So does that mean our jobs will be eliminated, too? The answer isn’t as obvious as you might think.

Yes, the world is changing. But if you’ve been doing SEO for a while, you should be used to that by now.

SEOs have always been forced to wear strange combinations of hats: part technical analyst, part content strategist, part UX researcher, part marketer, and part analyst.

I don’t think AI will make SEO expertise obsolete. But it will make shallow SEO obsolete.

The people who thrive will be the ones who understand search behavior, business outcomes, technical systems, content strategy, analytics, and how to turn all of that into better decisions.

The old version of SEO stopped working years ago

I’ve been doing SEO since before there was a word for “SEO.” Every few years, there’s a viral article declaring that “SEO is dead.” One of the first to catch fire was a 2005 article by Jeremy Schoemaker, repeating something he’d heard from Jason Calacanis. 

Then, in 2009, Danny Sullivan wrote an article on this site reacting to a blog post by Robert Scoble declaring that “SEO isn’t important anymore.”

We know the reality. SEO never died. But over the years, it’s changed a lot.

Look at this screenshot of a Google search for [flowers] in 2007 versus the same search in 2026.

This example is near and dear to my heart because I wrote that title tag in 2007. I was fortunate enough to lead SEO at 1-800-Flowers at a time when a No. 1 organic ranking meant significant traffic and revenue.

Twenty years later, their team has maintained the No. 1 organic ranking. However, today it’s so buried on the SERP that I wonder whether it gets any clicks at all.

This phenomenon isn’t limited to searches for “flowers.” Search for any competitive head term these days, and chances are you’ll see the organic result buried.

Is SEO “dead”? That really depends on your definition of “SEO.”

If your definition is “getting to the top of Google organic search” by spending your whole day writing title tags, then yeah, SEO is pretty much dead. It has been for a long time.

If your definition of SEO is understanding that people are looking for your goods and services, understanding their needs, answering their questions, and meeting them wherever they go to find information, then your journey as an SEO expert — or whatever you eventually decide to call yourself — is only beginning.

Dig deeper: Could AI eventually make SEO obsolete?

Why true SEO experts are uniquely positioned to thrive 

There’s one phenomenon I’ve noticed with AI, not just in SEO, but across every industry. You might have noticed it too.

On social media, you’ll see a lot of AI-generated videos. The vast majority are silly “look what I can do with AI” videos. You see them, maybe press “Like,” and then forget about them. But the ones with staying power are made by people who understand filmmaking: pacing, framing, lighting, composition, camera movement, editing, sound design, and how to build toward an emotional payoff.

In other words, even though everyone can generate videos with AI now, the differentiator is no longer how “cool” the visuals are. It’s how skillfully creators use AI as a tool to achieve their vision.

There’s an analogous situation happening with SEO and AI. I’ve noticed a lot of people typing simplistic prompts and, like Neo in “The Matrix,” declaring, “I know SEO.”

What these folks don’t realize is that SEO is a lot more than title tags, and it was never just about reverse-engineering search engines. It was always about reverse-engineering the human brain, drawing on knowledge and experience across keyword lists, user behavior, content strategy, technical systems, analytics, persuasion, UX, and business outcomes.

When others are typing simplistic prompts into their LLMs, SEO experts will be having deep conversations with their LLMs, teaching them, challenging them, and finding ways to get the best out of them. Those who excel in this new world won’t be the ones who have all the answers. They’ll be the ones who have the right questions.

While it’s still early, and I’m convinced we haven’t even scratched the surface of ways to use LLMs in SEO, here are just a few ways I’ve been using AI in my SEO work to make it more efficient and effective than ever.

1. Performing SEO basics with unprecedented efficiency and effectiveness

I’m generally not a fan of AI-generated long-form writing. You end up with generic, inauthentic slop that, in the words of Shakespeare, is “full of sound and fury, signifying nothing.” 

I predict that a year from now, most people will be able to spot the clear signs of AI-generated copy: not just obvious tells like excessive use of em dashes and repetitive phrasing (“That’s not X … it’s Y!”), but a lack of authentic personality and stories.

Metadata is one of the places where I don’t mind AI assistance because its job isn’t to invent original thought. It’s to compress the page’s value, intent, and positioning into the right format for the right surface.

The big mistake I see people making with AI-generated metadata is that their prompts are far too generic: “Write a title tag for this page.”

A seasoned SEO knows the goal isn’t to create a “pretty title tag.” It’s to create the most effective title tag possible for human, search engine, and AI discovery. It takes into account various search intents, brand positioning, competitor gaps, conversion drivers, and practical space limitations.

AI opens up new opportunities that weren’t practical before. Not many people know that ideally, your title tag, Open Graph tag, and Twitter card should be distinct from one another because they’ll be shown to different audiences on Google, Facebook, and X. And it took me a few tries to remind AI that title tag length isn’t based on character count, but on pixel width.

Those “in the know” will start using AI to generate everything: title tags, meta description tags, OG tags, Twitter cards, and the right structured data.

Someone without SEO experience will write generic prompts and wonder why their perfectly polished title tags aren’t doing anything for them a year from now.

Dig deeper: The AI writing tics that hurt engagement: A study

2. Turning SEO recommendations into dev-ready tickets

One “edge” I’ve had throughout my career is the ability to translate vague marketing goals into precise technical requirements developers can actually execute.

But as technology has become more complex, I found myself hitting my own limits. I understood the principles of coding, but had a hard time articulating exactly what I needed developers to do. Googling hardly ever helped because I’d just find high-level articles written by consultants, some of whom clearly didn’t understand it either.

A practical example is modern React or single-page app architecture, where a page may look complete to users while key SEO content is assembled after load from JavaScript rather than appearing as crawlable HTML.

In the past, I might’ve written a vague recommendation like “we need more crawlable content on this page,” forcing my poor developer to figure out what that means.

With AI, I can turn that into a real implementation ticket: grounding the LLM in the site’s tech stack, translating the SEO need into concepts like server-side rendering, hydration, DOM content, and crawlable links, and adding examples, test cases, edge cases, and acceptance criteria.

The point isn’t to become a React engineer. It’s to communicate SEO requirements in a way that developers can execute without forcing them to think too much about it. Trust me, your developer will thank you.

3. Mining GSC, GA4, and Semrush or Ahrefs data for actual user needs

Treating AI optimization as long-tail SEO done right has been one of the game-changers for me when it comes to my own productivity.

The holy grail of SEO has always been to read your users’ minds and create content that meets their needs. Anyone who’s spent a lot of time with SEO data knows that there are enormous amounts of insights locked within this data. The first problem is unlocking them. The second problem is getting them into a format that will get people to pay attention.

In the past, I would literally lock myself in a room with a giant spreadsheet open on my screen. I’d go through search terms one by one, categorizing and clustering them, and, if I was lucky, end up with a handful of insights days later.

I might start with a list of 30,000 keywords and get through maybe a few hundred before getting completely exhausted. And when I’d present my insights, along with my giant pivot table, to stakeholders, they’d nod their heads, and then everyone would forget about them.

LLMs are changing the game. You can simply upload data from GSC, GA4, and Semrush and Ahrefs, along with your own business and market insights, and then simply ask your LLM questions.

Here are just a few recent examples of analyses I’ve done for my clients. These would once have taken days or weeks. Now I can get to a strong first pass in minutes.

• Analyze our GSC keyword data and organize the keywords into topical clusters. Which topics do we clearly have a “right to own” in Google’s eyes?
• Review our top competitors and uncover keywords within this topical neighborhood that they rank for but we don’t. What kind of content do we need to “break in”?
• Surface GSC queries that get lots of impressions but few clicks. What improvements can we make to our titles, snippets, or positioning to drive more clicks?
• Examine organic landing pages that attract a lot of traffic but fail to convert. What is the search intent behind the keywords driving traffic to these pages, and how can we improve conversion?
• Find keywords where we’re in “striking distance” of stronger rankings. What additional content do we need to create or adjust to push us to the top?
• Analyze the queries people type into our on-site search. What are examples of searches they might perform on Google or prompts they might use in LLMs when looking for this information?

There are literally an endless number of questions you can ask. I didn’t present these as sample prompts because they’re thought starters. While you’ll probably get a decent answer, the real value from AI comes only when you:

• Dig deep into specific concepts, pages, and keywords.
• Validate the LLM’s responses.
• Challenge it as necessary.
• Recognize hallucinations or context drift.
• Put your findings into immediate action.

Dig deeper: How to use AI to diagnose and improve search intent alignment

---

4. Prototyping page layouts, content modules, and more

Something else I’ve found LLMs can do really well is generate a solid wireframe of a page or page module that you can pass on to your web designer and developer. But this is another area where the quality of the output depends almost entirely on the quality of your prompt and the context you provide the LLM.

Most people will simply type “design me a web page,” perhaps with a few “wish list” items they’d like to see. AI may produce something that looks “complete” on the surface, perhaps a hero section, a list of benefits, some FAQs, and a call to action (CTA). But when executed, it’ll feel lifeless, generic, and disconnected from the actual business problem.

The better approach is to ground the LLM with as much background information as possible. This doesn’t need to include every SEO report, but rather the ones that provide the highest-quality signals, such as the ones we discussed above: topic clusters, competitor gaps, conversion data, and on-site search data. Add other useful information like sales objections, customer reviews, your brand’s unique value propositions, and a clear explanation of what the page needs to accomplish.

With proper context, AI can help lay out something that transcends a generic landing page. For example, it can propose a strong hero section with suggested wording, recommendations for CTAs, section order, comparison tables, proof blocks, FAQs based on real questions, trust elements, and paths for different stages of intent.

Remember that it works in reverse, too. Upload a screenshot of an existing page, either yours or your competitor’s, tell the LLM what your goals are for the page, and ask it to critique the page.

AI can also open up other SEO opportunities that have previously been roadblocks. 

• Want to do A/B testing? Tell the LLM the hypothesis you want to test, and have it come up with variants for you. 
• Want to prototype a simple interactive tool? Provide your requirements, provide the underlying data, and see what your LLM can do. 

In some cases, it can go beyond a static mockup and produce a working prototype that a developer can evaluate, harden, and turn into production code.

Your edge as an SEO is knowing what information to feed the model, what problems the page actually needs to solve, and which ideas are strategically useful versus just AI-generated decoration.

The one thing that I haven’t seen AI do very well yet is generate professional-quality design and production-quality code. But everything up to that point is at your fingertips now. 

5. Making analytics useful again

As I’m sure it was for many of you, July 1, 2024, was a dark day for me. That’s when Google shut down Universal Analytics and forced us all onto GA4.

Since it was called Urchin, I’d all but mastered UA. Then one day, all of my reports and dashboards were simply gone. And I had no interest in spending another decade on a learning curve just to recreate reports that they’d once given me by default.

But with the arrival of LLMs, you can simply ask the LLM to walk you through building whatever report you want.

The first report I had to re-create was the on-site search report, one that’s inexplicably missing from GA4. I wrote my own prompt to walk me through creating this, but for the purposes of this article, I had ChatGPT write the prompt:

Act as a senior GA4 analytics consultant.

I want to rebuild a useful onsite search report in GA4/Looker Studio. GA4 does not provide the same dedicated Site Search report that Universal Analytics had, but I can use the `view_search_results` event, the `search_term` parameter, and any custom parameters needed.

Create a practical, implementation-ready plan that covers:

1. How to confirm onsite search tracking is working.

2. Recommended event name and parameters, including which should be registered as custom dimensions.

3. How to track searches when the site does not use URL query parameters.

4. The most useful report sections, including:
- total searches
- unique searchers
- top search terms
- zero-result searches
- refined or repeated searches
- searches followed by exits
- searches followed by conversions
- searches by page, device, and user type

5. Step-by-step instructions for building the report in GA4 Explore and Looker Studio.

6. A QA checklist to make sure the data is accurate.
Keep the answer concise, practical, and usable by both a marketer and a developer.

The key to writing these prompts, or prompts that generate prompts, is including the phrase “step by step.” One of the nice things about AI is that it doesn’t judge.

Take as long as you need, ask it to break the setup down into steps as granular as you like, and feel free to ask “dumb” questions. It’ll oblige enthusiastically.

You can imagine what this opens up. One of the classic issues with SEO analytics is that all too often, they’re merely vanity metrics. 

Conversions, clicks, impressions, and rankings may look impressive at first, but eventually the dreaded “so what” question will arise. Who really cares if you see impressions and rankings growing like wildfire if your revenue isn’t increasing?

This is where you want to ask your AI to help you tie data to business performance. 

• Which unbranded keywords are actually driving revenue? 
• Which are leading to soft conversion goals like email signup, account creation, or pricing page visits? 
• Which search queries bring in engaged visitors who come back later through brand search, direct traffic, or email?

Again, the sky’s the limit. You can build a report or dashboard to answer just about any question your stakeholders have, provided you’re collecting the right data, and if you’re not, AI can help you create tickets for your web developer to collect that data.

Dig deeper: SEO analytics: How to interpret SEO data & anomalies

The work is changing. The need for expertise isn’t.

Like I said, this is only scratching the surface of how AI can help transform the work we do as SEOs.

But let’s get to the question everyone is really asking: Is your job safe?

I don’t have a crystal ball. But one thing is pretty clear to me. Not every SEO job will survive unchanged. Big companies will likely cut roles. Teams will likely get smaller. A lot of tactical work that used to require specialists may be done faster, cheaper, or “good enough” by someone using AI.

If your value is limited to tasks that AI can perform on command, there may be challenges ahead.

But if your value is understanding customers, interpreting search behavior, connecting data to business outcomes, translating strategy into execution, and helping companies become more findable, useful, and trusted, then AI isn’t the end of your career. It may be the best leverage you’ve ever had.

And there’s another reason I’m optimistic. The same AI disruption hitting SEO is hitting every other white-collar profession, too. If large companies do lay off significant numbers of talented people, many of those people aren’t just going to disappear from the economy.

Some will start businesses. Some will finally pursue ideas they’ve had in their heads for years. Some will use AI to build prototypes, launch products, test markets, and create companies in ways that would have required far more capital and staff just a few years ago.

That should give us hope.

Many of the great companies we know today started with little more than a few people, an idea, and the willingness to figure things out as they went. Steve Jobs and Steve Wozniak, Bill Gates and Paul Allen, Mark Zuckerberg, Jeff Bezos, Larry Page and Sergey Brin, Michael Dell, and many others did not begin with massive corporations behind them. They began with ideas, persistence, and the tools available to them at the time.

If they were able to accomplish what they did with their tools, imagine what a new generation of entrepreneurs will be able to do with AI.

Maybe you’ll be one of those entrepreneurs. Or maybe your role will be helping one of them turn their ideas into businesses people can actually discover, understand, trust, and choose.

Either way, the products, services, brands, and businesses built with AI will still need to be found. They will still need to explain why they matter. They will still need to earn attention, authority, and trust.

SEO is dead. Long live SEO.