SEO in the generative AI era

SEO is not dead. It has changed. In 2026, Google processes more than 8.5 billion searches daily. But it now shares the stage with generative engines: ChatGPT Search, Perplexity, Gemini with Google Search, Bing Copilot. People no longer only search on Google. They ask AIs that search for them.

This creates two playing fields:

• Classic SEO: ranking your content in Google results (organic, featured snippets, People Also Ask).
• GEO (Generative Engine Optimization): getting AI models to cite your content when they answer user questions.

They are not mutually exclusive. Content that performs well in classic SEO also tends to be cited by generative AIs. But there are specific techniques for each. AI can help you on both fronts.

Keyword research with AI

Traditional keyword research involves tools like Ahrefs, Semrush or Google Keyword Planner. They remain useful, but AI adds an analysis layer that previously required experience and hours of manual work.

Generating keywords with AI

Prompt to generate a keyword map:

"I am a [type of company] and I sell [product/service] to [ICP]. Generate a keyword map organized by search intent (informational, navigational, commercial, transactional). For each keyword: suggest estimated volume (high/medium/low), difficulty (high/medium/low) and an optimized article title. Give me at least 30 keywords grouped into thematic clusters."

AI does not have exact volume data (you need Ahrefs or Semrush for that), but its ability to generate semantic variations, long-tails and related questions is superior to manual searching.

Competitor keyword analysis

Combine AI with real data. Export your competitors' keywords from Ahrefs or Semrush (CSV) and pass them to Claude or ChatGPT:

"Here are the top 200 organic keywords of my competitor [paste CSV]. Compare with my current keywords [paste CSV]. Identify: 1) Keywords where they rank and I don't (opportunities). 2) Keywords where we both rank but they're better positioned (direct competition). 3) Long-tail keywords that neither is covering well. Prioritize by traffic potential and alignment with my ICP."

This analysis, which manually takes half a day, AI completes in 2 minutes. Quality depends on the quality of the data you provide.

Long-tail keywords and questions

Long-tail keywords (phrases of 4 or more words) usually have less competition and higher purchase intent. AI excels at generating them:

"For the topic [X], generate 20 questions someone would ask on Google before buying [product/service]. Include question types: comparative ('X vs Y'), cost ('how much does X cost'), process ('how to implement X'), validation ('does X work for Y?'), alternatives ('alternatives to X')."

Search intent analysis

Search intent is what the user actually wants when they type a keyword. Google understands it increasingly well. If your content does not satisfy the intent, it won't rank. It doesn't matter if it has the right keywords.

The 4 intents

• Informational: "what is lead scoring." The user wants to learn. Your content: educational article, guide, tutorial.
• Navigational: "HubSpot login." The user is looking for a specific page. Not very relevant for your content strategy (unless you are HubSpot).
• Commercial: "best CRM for SMBs 2026." The user is researching before buying. Your content: comparison, review, use case.
• Transactional: "buy HubSpot license." The user wants to buy. Your content: landing page, pricing page, direct CTA.

Automated classification with AI

If you have a list of 200 keywords, manually classifying them by intent is tedious. AI does it in seconds:

"Classify these keywords by search intent (informational, navigational, commercial, transactional). For each one, suggest the optimal content type and an SEO title. Format: table with columns Keyword, Intent, Content type, Suggested title, Priority (high/medium/low)."

Then, prioritize commercial and transactional keywords for your product pages, and informational ones for your blog. It's a distribution that seems obvious but many companies ignore: they publish informational articles on pages that should sell, and sales pages where they should educate.

Search-optimized content

Generating content with AI that ranks well on Google requires more than writing an article. You need to optimize structure, depth and quality signals.

SEO content structure

An SEO-optimized article has these elements:

• Title tag: main keyword + modifier (guide, 2026, step by step). Maximum 60 characters.
• Meta description: summary that invites the click. Maximum 155 characters. Includes the keyword naturally.
• H1: visible title, can be different from the title tag but must include the keyword.
• H2 and H3: subtitles covering the main subtopics. Each H2 answers a question or subtopic from the cluster.
• First paragraph: directly answers the user's question. Google values the answer appearing early.
• Internal links: link to other relevant articles on your site. Minimum 3 internal links per article.
• External links: cite authority sources. This improves content credibility.

SEO content prompt

"Write a 2,500-word blog article about [main keyword]. Search intent: [informational/commercial]. Structure: H1 with keyword, 5-7 sections with H2, at least 2 H3 per section. Include a comparison table, a step list and an FAQ with 5 questions (question-answer format). The first paragraph must directly answer the main question. Tone: [according to your tone guide]. No filler. Every paragraph must add new information."

E-E-A-T and AI content

Google evaluates content with 4 criteria (E-E-A-T): experience, expertise, authoritativeness and trustworthiness. AI-generated content can fail on experience (it has no real experiences) and authoritativeness (it is not a recognized expert).

The solution is not to hide that you use AI. It is to complement what AI generates with your real experience:

• Add data from your own experience: "in our case, we tested X and obtained Y result."
• Include screenshots, charts or proprietary evidence.
• Sign with a real author who has a verifiable profile (LinkedIn, bio on the website).
• Cite primary sources, not just other pages talking about the topic.

GEO: generative engine optimization

GEO (Generative Engine Optimization) is the discipline of making AI models cite your content when they answer questions. When someone asks Perplexity "what is the best CRM for B2B startups," you want your article to appear as a cited source.

How AIs decide which content to cite

Generative models with search access (Perplexity, ChatGPT Search, Gemini) select sources based on:

• Domain authority: sites with a history of quality content and relevant backlinks.
• Direct relevance: content that answers the question concisely and accurately.
• Structured data: content with schema markup, tables, lists and formats that AI can parse easily.
• Freshness: recently updated content (publication and modification dates matter).
• Specificity: content that gets to the point rather than generic pages covering everything superficially.

Practical GEO techniques

1. Direct answers: include paragraphs that directly answer common questions. Format: question as H2 or H3, answer in the first paragraph below.
2. Citable data: include statistics, percentages and concrete data with source. AIs tend to cite content with specific data.
3. Comparison tables: AIs parse them easily and use them to build their responses.
4. Ordered lists: "the 5 best X for Y" is a format that AIs frequently cite.
5. llms.txt file: a file at the root of your site (similar to robots.txt) that tells AI crawlers which content is relevant and how to interpret it.
6. Frequent updates: update your best articles every 3 months with new data and the modification date.

Schema markup and structured data

Schema markup is JSON-LD code that helps Google and AIs understand your page content. It's not visible to the user, but it improves how engines interpret and display your content.

Essential schemas for marketing

• Article: for blog articles. Includes title, author, publication date, modification date.
• FAQPage: for FAQ sections. Google can display them as expandable rich snippets.
• HowTo: for step-by-step guides. Appears with a special format in search results.
• Product: for product pages. Includes price, availability, ratings.
• Organization: for the main page. Name, logo, social networks, contact data.
• BreadcrumbList: for navigation. Helps Google understand your site structure.

Generating schema with AI

AI can generate schema markup from your content: "Generate FAQPage JSON-LD schema for these 5 questions and answers [paste]. Valid format for the Google Rich Results validator."

Always validate the generated schema at search.google.com/test/rich-results before publishing. AI generates valid schema most of the time, but can make formatting errors that prevent Google from recognizing it.

AI-assisted technical SEO

Technical SEO ensures that Google can crawl, index and understand your site correctly. AI can help with technical audits:

Technical SEO audit with AI

Export a crawl of your site (with Screaming Frog, Sitebulb or similar) and pass the CSV to the AI:

"Here is a crawl of my site with 500 URLs [paste CSV]. Identify: pages with 4xx/5xx errors, pages without meta description, pages with duplicate title tags, pages with load time over 3 seconds, redirect chains, orphan pages (without internal links). Prioritize issues by SEO impact (critical/high/medium/low)."

This analysis turns hours of manual review into minutes. The key is giving it real data (the crawl), not asking it to "audit my website" without data.

Optimizing existing URLs

For pages that already rank but could improve: "This page ranks at position 8-12 for the keyword [X]. Here is the current content [paste]. Analyze: what is it missing to compete with pages in position 1-3? What subtopics do they cover that I don't? What data or formats (tables, lists, FAQs) should I add? Suggest concrete improvements without rewriting the entire article."

SEO tools with AI

• Ahrefs / Semrush: keyword, backlink and competitor data. Both integrate AI for suggestions and analysis. Essential.
• Surfer SEO: real-time content optimization. Analyzes pages that rank for your keyword and tells you what to improve (length, semantic keywords, structure).
• Clearscope: similar to Surfer, focused on topical coverage. Tells you what topics and entities your article should cover.
• Frase.io: generates SEO-optimized outlines from SERP analysis. Combines keyword research with content generation.
• Google Search Console: real data on how Google sees your site. Free and essential.
• Perplexity: to verify how generative AIs display topics in your sector. Search your keywords and see if your site is cited.

Errores comunes

1. Generating SEO articles without keyword data. AI doesn't know real search volume. Use SEO tools for data and AI for analysis and generation.
2. Endless content without value. 5,000-word articles full of filler don't rank better. Google measures user satisfaction, not length.
3. Ignoring GEO. If you only optimize for Google and not for generative engines, you lose a growing share of traffic. AI searches increase every quarter.
4. Not updating existing content. Updating an article that already ranks (with new data and modified date) is more effective than writing a new one from scratch.
5. Incorrect schema markup. Badly formatted schema is worse than having no schema. Always validate with Google's tool.

Ejercicio practico

Puntos clave