I woke up to a 40% traffic drop last May. Not a slow decline. A cliff. My blog was fine a month ago. Then Google rolled out AI Overviews wider. Suddenly, my “best answer” posts stopped getting clicks. People were staying on Google. Reading the blue box. Never visiting me.
That is when I started searching for answers. I typed “SEO for AI what is it called” more times than I admit.
Turns out, nobody has a perfect name for it yet. Some call it GEO (Generative Engine Optimization). Others stick with “AI Search Optimization.” But the real question is not the label.
The real question is: how do you write for a machine that reads everything but clicks nothing?
I tested seven strategies over six months. Lost money. Made mistakes. Found three things that actually work.
Here is my honest breakdown.
First Thing First – What is “SEO for AI” Actually Called?
You will see four names floating around.
Most bloggers say GEO (Generative Engine Optimization). Coined by researchers at Princeton and Google DeepMind. Sounds fancy. But regular bloggers do not use it yet.
Related Article: What the AI SEO Company Coalition Knows About the Next Algorithm Update?
Then there is AEO (Answer Engine Optimization). Older term. Focused on voice search and featured snippets. Still relevant because AI Overviews pull from the same logic.
Third name is LLM optimization. That means making your content easy for Large Language Models to understand. No fluff. No hidden meaning.
Fourth name is just “AI Search visibility” – boring but practical.
I call it “earning the citation” instead of “earning the click.”
Here is why that matters.
Old SEO was about ranking #1. You win the click. You get the traffic. You show the ad.
New AI-driven search is different. Google shows a paragraph answer at the top. It cites three sources. One of them might be you. But the user never leaves Google.
So your job changes.
You are not optimizing for a blue link anymore. You are optimizing to be the source inside the AI box.
And that changes everything about how you write.
Why My Old Blog Posts Stopped Working
I used to write 2000-word guides.
Long intro. Personal story. Step one. Step two. Step three. Conclusion. CTA.
Classic blogging 101.
AI Overviews killed that format.
Here is what happens now. A user searches “how to use AI for SEO.” Google’s AI reads 50 blogs in 2 seconds. It picks the clearest 200-word answer from the middle of your post. It shows that. No click needed.
If your answer is buried under fluff, the AI ignores you.
If your answer is perfect but your site has low authority, the AI cites someone else.
I learned this the hard way. One of my best posts – a detailed case study on AI SEO optimization tools – dropped from 200 daily clicks to 12. But the AI Overview was quoting my data. Just not my domain.
That stings.
E-E-A-T is Not a Buzzword Anymore – It’s Your Only Shield
Google’s update is called E-E-A-T.
Experience. Expertise. Authoritativeness. Trustworthiness.
In 2023, you could fake it. Hire a writer. Add some stats. Publish.
In 2025, AI Overviews detect fluff instantly.
Here is what changed.
The AI looks for first-person proof. Did you actually use the tool? Did you test the strategy? Or are you summarizing someone else’s work?
Let me give you a real example.
I run a small blog about SEO tools. Last month, I reviewed an AI writing assistant called KoalaWriter. I did not just list features. I showed a screenshot of my failed draft. Then my revised draft. Then the traffic difference.
That post got picked up by Google’s AI Overview for the query “AI SEO optimization tools review.”
Why? Because I showed failure. Real screenshots. Real dates. Real numbers.
The AI trusts messiness. It trusts specific failures more than generic wins.
So here is my advice. Stop writing like a marketer. Start writing like a person who tried and failed and tried again.
The Experience Gap Most Bloggers Miss
Let me show you what I mean.
Bad content (AI detectable):
“AI SEO optimization tools can improve your workflow by 40%.”
(No proof. No context. Fake number.)
Average content:
“I tested Jasper and SurferSEO for two weeks. Jasper helped with outlines. Surfer helped with NLP keywords.”
Better content (what Google Overviews actually cite):
“I tested Jasper on 10 blog posts. It saved me 2 hours per post but added 30 minutes of editing. SurferSEO gave me keywords that felt spammy. Then I tried Frase.io. That one worked for my niche – camping gear reviews. My click-through rate went from 1.2% to 3.8% in 30 days.”
See the difference?
The third one is ugly. It has mixed results. It names specific tools and time frames. It does not promise a silver bullet.
That is experience. And AI Overviews love that because it matches how real humans talk.
How AI Overviews Actually Decide What to Show?
I spent 6 weeks reverse-engineering this. Not a scientist. Just a blogger with too much time.
Here is what I found.
Google’s AI looks for three things in order:
1. Directness. If the query is “how to use AI for SEO,” your first 100 words must answer exactly that. No stories about your dog. No history of search engines. Just the answer.
2. Unique data. If ten blogs say the same thing, the AI picks the one with a unique screenshot, a table, or a contradictory opinion.
3. Source diversity. Google prefers citing one .gov, one medium-sized blog, and one Reddit thread over three big media sites. That gap is your opportunity.
I tested this. I wrote two versions of the same article.
Version A: generic advice. “Use AI for keyword research, content writing, and meta tags.”
Version B: specific failure. “I used ChatGPT for meta descriptions. It wrote 50 descriptions. 12 were great. 38 were repetitive. Here are the 12 that worked.”
Version B got cited in an AI Overview within 11 days. Version A never showed up.
The “Citation Trap” You Need to Avoid
Here is a warning nobody talks about.
Getting cited in an AI Overview does not always help you.
I know. Sounds crazy.
But when Google shows your answer inside the AI box, users do not click your link. They read your 50 words and leave. Your bounce rate goes up. Your ad revenue goes down.
I saw this happen to a friend. He runs a recipe blog. Google started showing his “substitute for buttermilk” answer directly in AI Overviews. His traffic from that query dropped 70%. But his brand mentions went up 300%.
So you have to choose.
Do you want citations (brand visibility) or clicks (traffic)?
For new blogs, citations help. Google sees you as a source. You build authority over time.
For established blogs that rely on ad income, citations hurt. You want the click, not the quote.
I made this mistake. I celebrated being cited for 6 keywords. Then my monthly pageviews dropped by 15,000 within 4 weeks.
Now I optimize differently. I write answers that are incomplete on purpose. Just enough to get cited. Then I add a clear reason to click – like a downloadable checklist or a video demo.
How to Write for Both AI and Real People?
You cannot ignore AI Overviews. But you cannot ignore human readers either.
Here is my four-step process after six months of trial and error.
Step 1: Answer the query in the first 100 words.
No exceptions. If the user asks “what is SEO for AI called,” you say: “It is called GEO (Generative Engine Optimization) or AEO (Answer Engine Optimization). Most bloggers now simply call it AI Search Optimization.” Then you can expand.
Step 2: Add one unique data point from your own work.
Run a test for 7 days. Screenshot the results. Share the good and the bad. Google’s AI detects real data vs. fake data surprisingly well.
Step 3: Write subheadings as questions.
AI Overviews love pulling from H2 and H3 that are full questions. Example: “Do AI SEO optimization tools actually save time?” instead of “Time-saving benefits of AI tools.”
Step 4: Link out to original sources.
This sounds backwards. But Google’s AI trusts blogs that cite other domains. Link to a study, a Twitter thread, or a competing blog post. It shows you are not building a walled garden.
I tested step 4 aggressively. Posts with 3–5 external links got cited 2x more than posts with zero external links.
Tools I Actually Use (No Affiliate Junk)
People keep asking me for “AI SEO optimization tools.”
Here is my honest stack. Nothing sponsored.
- Frase.io – Good for finding questions real people ask. Bad for writing. Their AI writer feels robotic. I only use the question discovery feature.
- SurferSEO – Great for NLP keywords. Overwhelming for beginners. I use it once per post, not in real-time.
- ChatGPT (GPT-4) – I use this for outlines only. Never for final drafts. The tone is too clean. AI detectors flag it immediately. I rewrite every sentence.
- Perplexity AI – This is my secret weapon. I ask it “what would Google’s AI cite for this query” and it shows me real examples. Helps me reverse-engineer the overviews.
- RankMath (free version) – Basic schema markup. That is it. Do not buy the paid version unless you need local SEO.
Tool I stopped using: Any “AI humanizer” tool. They all leave patterns. ZeroGPT catches them every time. Just write with your own voice. It is faster in the long run.
The Honest Pros and Cons of Optimizing for AI Overviews
Let me be straight with you.
Pros (real ones I experienced):
- Your domain authority grows faster. Google sees you as a cited source.
- You get branded searches. People type “yourblogname + topic” after seeing your quote.
- Your content becomes tighter. No fluff. That helps human readers too.
Cons (nobody talks about):
- Click-through rates can drop 40-70% for cited queries.
- You spend more time on fact-checking. AI Overviews amplify your mistakes.
- It is unpredictable. I have perfect posts with zero citations. And messy posts with multiple citations.
Who should optimize for AI Overviews:
- New blogs (under 1 year old)
- Niche experts (doctors, lawyers, tradespeople with real licenses)
- Writers who enjoy testing and tweaking
Who should NOT optimize:
- Ad-heavy content sites (you need clicks, not citations)
- Affiliate sites (citations rarely convert to sales)
- Anyone who wants a “set it and forget it” strategy
I fall into the first category. Small blog. Niche audience. I like testing. So I keep optimizing for AI Overviews even with the traffic risk.
But I have friends in the second category. They stopped caring about AI citations. They focus on email lists and direct traffic instead. That is a valid choice.
Final Reality Check – Is This Worth Your Time?
Here is my honest answer after losing 40% traffic and gaining back 25%. AI Overviews are not going away. Google wants users to stay on Google. That is the business model.
You have two options. Option one: ignore AI Overviews. Build direct traffic via email, social, or YouTube. That works. It is just slower. Option two: learn to earn citations. Accept lower click rates. Use the brand visibility to grow other channels.
I chose option two. Not because it is perfect. Because I am too stubborn to quit blogging. But I changed my definition of success. I no longer track only pageviews.
I track “citations per 1000 words” and “brand search volume.” Last month, my citations went up. My traffic went down. But my newsletter signups went up 18%.
That is the weird new math of SEO for AI. You cannot control the AI. You can only control how clearly you write, how honestly you share your failures, and how patient you stay.
That is boring advice. But it is the truth.
And the truth is what Google’s AI actually wants.