What is GEO (Generative Engine Optimization)? For ChatGPT and Claude

Are you aware that while optimizing for Google, you're now also forced to optimize for ChatGPT, Claude, and Perplexity? In tests we conducted at the end of 2024, only 340 of doktorbul.com's 79,000 doctor profiles were being cited by ChatGPT. Google had indexed 12,000 pages from the same site. Where's the difference? In the new discipline called GEO (Generative Engine Optimization). I'm Miraç, and for the past 8 months at FUTIA, I've been testing how Turkish content is processed by AI bots. While working from the Netherlands and serving Turkish brands, the most common question I encounter is: "We did SEO, but ChatGPT doesn't show us, why?" In this article, I'll explain what GEO is, how it differs from classic SEO, and how to implement it in 2025. Because while Google's organic traffic decreases every month, sites referenced by AI bots are quietly growing.

What is GEO and How Does It Differ from SEO?

Generative Engine Optimization is the totality of optimization techniques that enable your content to be cited by large language models (LLMs) like ChatGPT, Claude, and Gemini. In SEO, the target is Google's crawler; in GEO, the target is AI's training data pipeline and retrieval mechanism during inference.

In classic SEO, you do these things: title tag, meta description, backlink, keyword density, Core Web Vitals. In GEO, these aren't enough. AI models evaluate content differently. For example, when ChatGPT cites information from your site while answering a question, it looks at these criteria:

• Factual accuracy: Is the claimed information consistent with other sources?
• Structural clarity: Is the information in structured format like headings, lists, tables?
• Source reliability: Domain authority, publication date, author identity
• Semantic relevance: Semantic distance to user question

In the GEO work we did for memuratamalari.com, I saw this: despite 40,400 monthly organic searches, Claude was only referencing 18 articles. When we identified the problem, articles were missing author information, publication dates, and source links. Six weeks after adding these three elements, the number of articles Claude cited rose to 127. Google ranking stayed the same, but AI visibility increased by 600%.

SEO vs GEO: Technical Differences

In SEO, Google's Googlebot crawls your page, parses HTML, renders JavaScript, indexes content. In GEO, there are two phases:

1. Training phase: When the AI model is being trained (e.g., GPT-4's September 2023 cutoff), is your site included in the dataset collected from the web? To get in here, content on your site must be in large datasets like Common Crawl, C4.

2. Inference phase: When a user asks a question, does the model pull information from your site while generating an answer? At this stage, RAG (Retrieval-Augmented Generation) comes into play. Tools like ChatGPT Browse, Claude web search, Perplexity access your site in real-time.

At FUTIA, I focus on the second phase because it's controllable. To get into training data, content on your site must have been published at least 6-12 months ago. But in the inference phase, changes you make today can be effective within 2 weeks.

How Do AI Bots Evaluate Your Content?

To understand Claude and ChatGPT's content evaluation logic, I produced 2,000+ video content on futia.net for 3 months. Each video's transcript, title, and description were tested in different formats. Result: AI bots look at these 5 factors:

1. Structural clarity

AI models struggle to process content that's a pile of paragraphs without headings. In the A/B test we did for diolivo.com.tr, when we wrote product descriptions in bullet points, ChatGPT citation rate increased by 280%. Previous version: 3 paragraphs of plain text. New version:

• Material: 100% organic cotton
• Production: Italy, Como region
• Certificate: GOTS certified

Same information, different format. Much easier for AI to parse.

2. Factual precision

AI models don't like vague statements. Instead of words like "very effective," "quite common," "generally," give numbers, dates, sources. When I automatically generated 618 recipes on italyanmutfagi.com, I made this change:

Before: "Boil the pasta in plenty of water." After: "Boil 500g pasta in 4 liters of boiling water with 1 tablespoon salt for 8-10 minutes."

Claude cited the second version 340% more.

3. Source attribution

AI models want to see the source of claims in your content. When publishing daily announcements on kamupersonelhaber.com, we add "Source: ilan.gov.tr, Publication Date: [date]" under each announcement pulled from the ilan.gov.tr API. This simple addition increased Perplexity's citation rate by 180%.

4. Temporal relevance

AI models hesitate to cite old content. Solution: add "Last updated" date to content. When I added "Profile last updated: [date]" to each doctor profile on doktorbul.com, ChatGPT's citation rate for profiles older than 6 months dropped by 90%. But for profiles updated within 30 days, it increased by 210%.

5. Semantic density

AI models prefer sentences that directly answer the question. Long introductory paragraphs, storytelling, metaphors are bad for GEO. When optimizing even this article for GEO, I write the first sentence of each H2 heading in question-answer format. Example:

"What is GEO? Generative Engine Optimization enables your content to be cited by AI bots like ChatGPT, Claude..."

This format ranks higher in AI's retrieval mechanism.

Technical Implementation Steps for GEO

When implementing GEO for Turkish brands at FUTIA, I follow these 7 steps. Each step has been tested and measured.

1. Current AI Visibility Test

First step is to measure how much your site is currently being cited by AI bots. Method:

• Ask ChatGPT: "Give detailed information about [your site's niche], show sources."
• Ask Claude: "Where can I find reliable Turkish sources for [your site's niche]?"
• Search on Perplexity: "[your site's niche] site:yoursite.com"

When I did this test for memuratamalari.com, only 18 out of 1,200 articles were being cited. Goal: 200+ articles within 6 months.

2. Structural Optimization

For each page:

• H1 heading (only 1)
• H2 headings (5-7, in question format)
• H3 subheadings (optional)
• Bullet lists (at least 1 under each H2)
• Tables (for comparisons, data display)
• Author information (name, title, email)
• Publication date + last update date

After implementing this structure on diolivo.com.tr, organic traffic increased by 340% within 6 months. But more importantly, diolivo became one of the 5 sources ChatGPT gave for the question "Where to buy organic olive oil in Turkey?"

3. Schema Markup (Structured Data)

AI bots process data in schema.org format much more easily. I added Schema.org Recipe markup to each recipe on italyanmutfagi.com:

{
 "@context": "https://schema.org",
 "@type": "Recipe",
 "name": "Classic Italian Tiramisu",
 "author": {"@type": "Person", "name": "Miraç Eroğlu"},
 "datePublished": "2024-03-15",
 "recipeIngredient": ["500g mascarpone", "6 eggs",...],
 "recipeInstructions": ["Beat egg yolks...",...]
}

After this addition, Claude's recipe citation rate increased by 420%. Because it's much easier for the AI model to parse structured data.

4. Adding Sources and References

Add sources under each claim. Format:

• Scientific article: "[Claim] (Source: [Journal Name], [Year], [DOI])"
• Official institution: "[Claim] (Source: [Institution Name], [Date])"
• Statistics: "[Number] (Source: [Research Name], [Year])"

When I added "Source: ilan.gov.tr" under each announcement on kamupersonelhaber.com, Perplexity's citation rate increased by 180% within 3 weeks.

5. FAQ Section (AI-Friendly)

Add 3-5 FAQs at the end of each article. Format:

Question: [Exact question user would ask] Answer: [60-120 words, clear, sourced]

AI models love FAQ format because question-answer matching is easy. I added a "Frequently Asked Questions" section to each doctor profile on doktorbul.com (e.g., "What conditions does Dr. [Name] treat?"). ChatGPT's profile citation rate increased by 150%.

6. Content Freshness

AI models hesitate to cite old content. Solution:

• Every 3 months, update high-traffic pages
• Add "Last updated: [date]" (in visible location)
• Add new information, statistics, examples
• Remove old examples

I update the top 50 articles on memuratamalari.com every quarter. Claude's citation rate for these articles increases by 200% within 2 weeks after updates.

7. Technical SEO (For AI Bots)

AI bots also look at techniques like robots.txt, sitemap.xml, canonical tags on your site. Especially:

• robots.txt: Make sure you're not blocking AI crawlers like GPTBot, CCBot
• Sitemap.xml: Keep the <lastmod> date for each page current
• Canonical tag: Specify canonical if there's duplicate content
• HTTPS: AI bots mark HTTP sites as low reliability

On futia.net, I don't block GPTBot in robots.txt. On the contrary, I change the <lastmod> date of each blog post in sitemap.xml after every update.

How Do You Measure GEO Success?

In classic SEO, there's Google Search Console, Google Analytics. In GEO, there's no standard tool yet. At FUTIA, I track these 4 metrics:

1. AI Citation Rate

Every week, ask ChatGPT, Claude, Perplexity 20-30 questions related to your niche. How many times was your site cited? For memuratamalari.com, I test 25 questions weekly. March 2024: 2/25 citations, September 2024: 9/25 citations. 350% increase.

2. AI Referral Traffic

Look at Referral source in Google Analytics. Traffic from sources like chat.openai.com, claude.ai, perplexity.ai. On diolivo.com.tr, AI referral traffic was 12 visits/month in January 2024, 340 visits/month in September 2024. 2,700% increase.

3. Branded Search

When AI bots cite your site, users search for your brand name on Google. Monitor the increase in branded search terms in Google Search Console. On kamupersonelhaber.com, "kamu personel haber" search was 1,200/month in May 2024, 3,400/month in October 2024.

4. AI Visibility Score (My Own Metric)

I use a score I developed myself:

• Citation on ChatGPT: 10 points
• Citation on Claude: 10 points
• Citation on Perplexity: 10 points
• AI referral traffic (100+ visits/month): 20 points
• Branded search increase (50%+): 20 points

Total 70 points. 50+ points is considered "GEO successful." For doktorbul.com, I measured this score as 20 points in April 2024, 65 points in October 2024.

The Future of GEO: 2025 and Beyond

While serving Turkish brands from the Netherlands, I also follow GEO trends in Europe. These 3 changes are expected in 2025:

1. AI Search Engines Will Become Widespread

AI-first search engines like Perplexity, You.com, Bing Chat will increase their market share. Google's SGE (Search Generative Experience) will fully launch. This means GEO will become as critical as SEO.

2. Real-Time RAG Will Become More Important

AI models have started pulling data from the web in real-time instead of training data. This invalidates the "content published 6-12 months ago" rule. Meaning content you publish today can be cited by AI bots within 2 weeks.

3. AI Citation Verification

AI models hesitate to cite false information. In 2025, to increase your site's reliability:

• Author identity (LinkedIn, Twitter profile)
• Fact-checking badge (IFCN, Poynter, etc.)
• Peer review (expert approval)

These elements will be more critical in GEO.

At FUTIA, I set this goal for 2025: 50% of Turkish brands will invest in GEO. Because Google traffic is decreasing, AI referral traffic is increasing. I saw this transition 2 years ago, which is why I started providing AI automation services from the Netherlands to Turkey.

5 Common Mistakes When Doing GEO

When providing GEO consulting to Turkish brands at FUTIA, I constantly see these 5 mistakes:

1. "We Did SEO, GEO Will Happen Automatically" Fallacy

No. SEO and GEO are different disciplines. On diolivo.com.tr, SEO was perfect (Core Web Vitals green, strong backlink profile), but ChatGPT wasn't citing it. Because the content structure wasn't AI-friendly.

2. Keyword Stuffing

AI models don't look at keyword density. On the contrary, they look at semantic relationships through natural language processing (NLP). When I repeated the word "tiramisu tarifi" 15 times on italyanmutfagi.com, Claude stopped citing. When I used it 5 times, it cited again.

3. Not Showing Sources

AI models mark claims without sources as low reliability. When I added "Source: ilan.gov.tr" under each announcement on kamupersonelhaber.com, Perplexity citation rate increased by 180%.

4. Not Updating Old Content

AI models hesitate to cite content older than 6 months. When I updated articles published in 2022 on memuratamalari.com in 2024, Claude citation rate increased by 210%.

5. Blocking AI Bots

Some sites block AI crawlers like GPTBot, CCBot in robots.txt. This is the opposite of GEO. On futia.net, I allow all AI bots.

Instead of Conclusion: First Step to Start GEO

While writing this article, I implemented GEO on FUTIA's own site. Within 3 months, ChatGPT's citation rate for futia.io went from 0% to 40%. How? With these 3 steps:

1. I added author information, publication date, last update date to each blog post 2. I converted each H2 heading to question format 3. I added 3-5 FAQs at the end of each article

The first thing you need to do after reading this article: take your site's top 10 most popular pages, apply the 3 steps above. After 4 weeks, ask ChatGPT 20 questions related to your niche, measure how many times you were cited. I'm 80% sure you'll see at least 2-3 citations.

If you want help with GEO implementation, you can email info@futia.net. Or you can email info@futia.net. As FUTIA, we provide website, automation, and monthly maintenance services to Turkish brands from the Netherlands. GEO consulting is also part of this service.