AI Marketing Isn't As Scary As You Think: You've Been Doing It for Over a Decade

A while back, I watched a roundtable conversation.

The main guest was Neil Patel, joined by two of his colleagues at NP Digital: Nikki, who runs SEO, and Chad, who runs content. Recorded in late 2025, the three of them tackled one question: now that AI is here, should marketers actually be panicking?

When Neil opened his mouth, I nearly slapped the table.

His point: panic about what? If you've ever done SEO, bought ads, or studied the algorithm of any platform, you've been doing AI marketing for years already.

Marketing to machines is something you've been doing for over a decade. AI is just the newest machine.

What does "marketing to machines" even mean?

Think about it. After Google came along, who were you optimizing your pages for? The crawlers. Only once a crawler understood you would it serve you up to actual humans. TikTok's recommendation algorithm, Amazon's search ranking — same story. You've been courting machines all along; it's just that the machine used to be called a search engine, and now this one can talk.

And it reads far more closely than any human.

A person landing on a page scans the headlines and skims for the highlights. A machine? It reads from top to bottom and doesn't miss a single cell of data.

Which leads to a very practical question.

The Name Doesn't Matter. Where the Traffic Sits Does.

At one point, the conversation hit a small tangent. First everyone argued: should this thing be called SEO, GEO, or AEO? HubSpot calls it answer engine optimization, and everyone has their own label.

Neil's position was refreshingly blunt: call it whatever the client wants to call it. While you're busy correcting the terminology, someone else is closing the business.

And if you want the hard numbers, NP Digital's data is more interesting. Of the AI referral traffic hitting their clients' sites, 85% to 95% comes from ChatGPT. Gemini? Most businesses don't even know where it is. Why don't they know? Because it simply doesn't send traffic, so nobody thinks about it.

So is Google dying?

Look at one number first: 5 trillion Google searches a year, 13.7 billion a day, still growing. Old-school search hasn't dipped at all.

But there's one category of new search that is migrating wholesale.

Nikki told a small story. She asked at what age a child should switch from a car seat to a booster seat. Her husband went down the Google rabbit hole and only got more confused. Try a different tool, she said. ChatGPT just gave her the answer.

Later she had neck pain for a year and suspected her pillow. She described her whole situation to ChatGPT in one go, and it recommended a $200 pillow, with a direct link to Amazon. She bought it. Now she recommends it to everyone she meets.

Neil went further. After who knows how many rounds of chatting with ChatGPT about his back pain, he ended up replacing mattresses for the entire household.

Notice something? These questions share a pattern: long, specific, loaded with context, and requiring follow-ups.

That kind of search, ChatGPT handles beautifully. Quick, simple lookups — exchange rates, transfers — people still take those back to Google. So this isn't about one replacing the other. Search has split in two.

"Our Customers Aren't On That Platform"? Not Necessarily.

A client once told Neil: our users aren't on TikTok. However hot search gets there, I'm not going.

Neil shot back with a question — and it was a sharp one: do you know which major AI model has signed a data deal with TikTok?

You don't. But the moment one does, your content on TikTok gets scraped and cited on other platforms. Your competitors aren't there; you are. You win by default.

Stack the logic one layer up and it gets even clearer.

Many of ChatGPT's answers are scraped from Google. And Google's front page now has a "what people are saying" section stuffed with content from TikTok, Instagram, and LinkedIn. Instagram itself handles 6.5 billion searches a day.

Each layer feeds the next.

Google also quietly changed the rules: it used to serve 100 results per crawl; now it serves 10. ChatGPT's crawling costs instantly multiplied tenfold.

So Neil offered an actionable tip: take UberSuggest or Semrush, plug in your own domain, and see how many TikTok videos rank for the keywords you care about. Plenty of brands get a shock — on their core terms, social platforms show up nearly as loudly as their own websites.

Let the data talk. Don't let feelings decide.

Social Search Is Two Different Businesses

Search on social actually comes in two flavors.

One is deliberate searching. Picking a hotel? First you scroll through Instagram. The other is more interesting: you never searched at all. Neil said he once bought a probiotic without searching once — he just kept seeing it, day after day, until one day it suddenly hit him: I need this. That one.

Passive search.

This also explains the obsession of the Louis Vuittons of the world: when a user searches "handbag," I must be there. Note that this is a completely different game from making a viral video. Content built for search and content built for reach look nothing alike.

There's a bigger tide underneath. Both Forbes and EMarketer point to the same thing: brand loyalty is declining. People would rather trust a creator they've followed for three years than a brand shouting "our pillow is the best."

And what about Reddit?

A couple of years ago, the big AI models cited Reddit like crazy; then citations visibly slid, and Reddit's stock took a tumble. The reason isn't complicated: it's far too easy for marketers to game.

But Nikki came to its defense. When she was booking a hotel in Cabo, she wanted to see the honest bad reviews — the good and the bad, all of it. And some of their clients have genuine brand evangelists who go on Reddit to correct misinformation — no payment, no pretending to be official.

Honesty always has a market.

Is Blogging Dead? No — It Just Found a New Life

The part of the conversation I agreed with most: blogging isn't dead — it's more important than ever.

The logic is plain. In any field, users have endless questions. If you don't have a single piece of relevant content, on what basis are you the authority in your field? Without authority, Google won't rank you, and ChatGPT won't cite you.

But today's blog is not the same animal as the one from five years ago.

NP Digital ran its own research: the number-one reason content gets cited in AI Overviews is that it contains facts, statistics, and original data. Length has nothing to do with it.

The counterexamples are ready-made too. A bunch of shoe companies published "Best Running Shoes of 2025" on their own blogs, ranked themselves first — and actually tricked the models into citing them. Google cleaned up this trick years ago; the AI models will clean it up sooner or later. Don't mistake the arbitrage for skill.

And here's one more thing that hits your wallet directly.

One of their financial services clients had a new customer walk in and say: I asked ChatGPT several questions, and you showed up in every single answer — so you must be the best. I'm signing. That was a mandate worth millions of dollars.

Visibility is becoming reputation itself. Users assume the platform wouldn't recommend an unreliable brand. Whether that assumption is right is a separate question.

Worth noting too: on conversion rate from AI search, NP Digital has seen the ChatGPT channel exceed 8%. Plenty of channels can only dream of that number.

The Most Expensive Wall Is the One Between Departments

Toward the end, the conversation turned to organizations.

Neil said he'd seen too many big companies — the nine- and ten-figure-revenue kind — where the SEO team and the paid team sit on opposite sides of a wall. The paid folks would rather burn money bidding against themselves on brand terms than ask SEO one question: where do we rank for this term?

Ranking first? Turn off the bidding. A competitor comes to poach? Turn it back on. One sentence's worth of coordination, saving real money.

Why should these two crowds be forced into one room every week? Chad gave two reasons.

First, the moment an AI Overview appears, the expected click-through rate on paid links drops. CTR drops, Quality Score drops; Quality Score drops, cost per click rises. Everyone's ads get more expensive. Meanwhile, the SEO team looks every day at which queries are triggering AI Overviews, on what topics, and what content changes would get you in. That intelligence locked inside the SEO team's heads is going to waste.

Second, Google's heavily pushed PMax and AI Max are, at their core, AI buying ads for you. They read your site's headlines, body copy, FAQs, and image alt text, then decide creative and channels on their own. If your site's content is a mess, the AI takes that mess and runs ads with it.

So that old saying needs an update: content isn't just for visitors anymore. It's for the ad-buying machine too.

The CMO's job comes down to one thing: force everyone to sit in the same room. Everything else is a waste.

Finally, the AI Browsers

Someone asked: Comet, Atlas — these AI browsers, would you use them?

Those who've tried them say they're fine. But there's one fatal flaw: they got jailbroken within a day or two of launch. Would you log into your bank account through one?

Didn't think so.

Switching browsers is ten times harder than switching search engines. Remember what it felt like moving people from Gmail to Outlook? Switching browsers feels exactly like that. Your entire life of passwords, bookmarks, and extensions lives in Chrome.

So don't panic about that one.

Back to the Beginning

As that conversation wrapped up, Neil said something I still remember. The gist: marketing as a craft has been pronounced dead several times, and it has come out fine every single time.

So what has changed?

Your audience has formally become two kinds: humans and machines. Humans have the patience to skim for the highlights. Machines have no patience — they read everything.

So fill in every cell of your data feeds, and nail down every fact in your content.

Thoroughness is the new bar in the age of machines.

Maybe in a few years, when your kid needs homework help, they won't ask Google, and they won't ask ChatGPT — they'll ask some talking device on the desk. When that day comes, hopefully you'll have already come around to this:

Platforms will change. Machines will change.

Courting machines and serving people — that has never changed.