When SEO Feels Like You're Just Training the Machine
If you publish informational content today, it's hard not to feel like unpaid R&D for Google and every LLM on the planet.
You write something useful, it ranks, and then one day your traffic falls off a cliff while an AI overview box above your result paraphrases your work back at the user. They get their quick hit of information. You get vanity impressions and a 0.5% CTR.
So the question isn't just "How do I get my clicks back?"
It's: What do I publish now that the machine eats all the free snacks?
I've been thinking about this a lot, and my answer keeps landing in the same place: there's less incentive to publish plain answers, and a lot more incentive to ship products.
Let's unpack that.
Free Content vs. AI Overviews: You're Playing the Wrong Game
A lot of web content was built on a simple trade:
- You give away clear, well-structured answers to common questions
- Search engines reward you with traffic
- Traffic eventually turns into customers, subscribers, revenue - whatever
AI overviews quietly changed that contract.
For any query where a one-paragraph answer is "good enough," the AI layer has zero reason to send the user to you. In fact, its job is to reduce clicks. Summarize, don't route. The whole value proposition of AI search is not having to visit your page.
That means things like definitions, glossaries, "What is X" explainers, and basic how-tos that can be flattened into bullet points - they all sit in the danger zone now. They're the easiest to scrape, summarize, and strip of all context and personality. If your business model depends on people clicking through to read that kind of content, you're exposed.
You're trying to sell bottled water in a world where the tap suddenly works everywhere.
So what do you do? You stop trying to win on being the tap.
There Are Two Kinds of People on the Internet
Here's something worth building your whole content strategy around: not everyone is the same kind of information seeker.
The first type is the free-info person. They Google, they skim, they take the AI answer and move on. They are not your customer. They never were. They never will be. They want the answer, they got the answer, done. Trying to convert this group is a losing battle - they weren't going to pay you anyway, and now they don't even need to visit your site.
The second type is different. These are the people who, even when a free answer is sitting right in front of them, still go looking for the course, the book, the tool, the paid resource. Not because they couldn't find the free information - they could. They just don't fully trust it. They're convinced the real insight, the shortcut, the thing that will actually work for them, is one layer deeper. Behind a paywall, inside a community, in someone's paid course.
This group is who you're building for.
And here's the beautiful thing about AI search for this audience: if someone asks an AI "what's the best course for learning X?" or "what tools do SEOs actually use?" - and your product gets mentioned by name - that's a warmer lead than almost any organic search click you've ever gotten. They didn't just stumble onto your page. They asked specifically for a recommendation, and your name came up.
That's a different kind of intent entirely.
What Still Needs a Click?
Even in an AI-everywhere world, there are categories of content that still require a click:
- Things you can actually buy
- Tools you can actually use
- Experiences you can actually join
- Communities you can actually participate in
AI can describe those things. It can't deliver them. At some point, the user has to land somewhere to access the real thing.
That's your opening.
If all of your work exists as freely copy-pasteable paragraphs, you're a training set.
If your work exists as products, tools, and experiences, you're a destination.
The question shifts from "How do I protect my definitions from being scraped?" to "How do I wrap my expertise into things AI can mention but not fully replace?"
From Publishing to Shipping
Think about your niche, whatever it is. Instead of asking "What content should I publish?", start asking: "What can I ship that someone would actually need to click to use, join, or buy?"
Take a niche like Hawaiian Pidgin vocabulary. The definitions themselves are table stakes - AI can summarize those all day. But you could build:
- A learner's toolkit with phrases by context, audio, quizzes, and cultural notes
- A "Pidgin in the wild" archive with video clips, commentary, and history
- A members-only community for learners and native speakers
- A simple paid app or reference tool people keep on their phone
Suddenly you're not fighting to "protect" the dictionary. You're using it as a base layer for things that have real gravity - products, tools, membership, community.
AI might tell someone what a word means. It can't enroll them in a course, ship them a book, or plug them into a living culture.
That's your job.
The same logic applies to SEO and marketing content. Anyone can ask an AI to explain how to do keyword research. What AI can't do is give someone access to your template library, your private Slack community, your live audits, your done-for-you service. Those still require a click. Those still require you.
Use AI as Distribution, Not a Thief
Here's the mental flip that helped me most:
Stop seeing AI overviews as "my enemy that stole my content." Start seeing them as free billboards that will never close the sale.
Right now, most content sites are structured like this:
AI: "Here's the definition."
User: "Cool, that's all I needed."
You: "Please click to read the exact same definition, but on my server instead."
There's no reason to click. You're asking people to do extra work for the same reward.
But if the pattern becomes:
AI: "Here's the short answer. If you want the full framework, templates, and a community of practitioners actually using this, [Your Brand] is where people go."
User: "Oh, that I actually want."
Now AI is doing top-of-funnel work for a real thing - not just cannibalizing your blog post. It becomes a referral engine for the second type of person: the one who was already going to pay, who just needed a nudge in your direction.
This is especially true for courses, books, and paid tools. The people who buy those things are actively seeking recommendations. They're asking AI "what should I read about X?" or "what's the best tool for Y?" They want to be pointed somewhere. Make sure that somewhere is you.
Yes, Still Make the Site Good
None of this means you ignore the fundamentals. You still want pages that load fast, clean structure and internal linking, clear titles and descriptions that earn attention, and E-E-A-T signals that tell humans a real person lives here.
But treat those as hygiene, not strategy.
Hygiene keeps you from losing for dumb reasons. Strategy decides whether winning even matters.
If your whole "win condition" is more traffic to ad-monetized definitions, you're playing a game the platform is actively trying to shut down. If your win condition is more people buying, joining, using, or subscribing, then traffic from AI overviews, from search, from social - from anywhere - is useful as long as it flows into something real.
So What Do You Actually Do Next?
If you're feeling this AI squeeze, sit down and answer three questions:
1. What part of what I do actually needs a click?
Not "should," but must - purchases, tools, experiences, community. That's your anchor. Build toward it.
2. How can I turn more of my knowledge into those things instead of free text?
Courses, books, apps, templates, membership, events, interactive tools. Even a well-structured email list with exclusive content is something AI can't replicate. What's the product version of what you know?
3. How do I structure my site so plain answers live in service of the product - not as the end goal?
Definitions and simple explainers become on-ramps. The real value is one layer deeper. The free stuff earns trust and drives discovery; the paid stuff is where the relationship actually happens.
The era of "publish everything and let SEO sort it out" is fading fast.
The era of "ship things people actually need to click for" is just getting started.
AI will keep answering questions. Let it. Your job is to be the place people still have to visit when they want something more than an answer - and to make sure that when AI is handing out recommendations, your name is the one that comes up.