AI Marketing Is the New SEO: One Strategy for Google and the AI Engines

For twenty years, search marketing meant Google. You ranked. You got the click. The end.

In 2026 that picture has split in two. Half the high-intent queries still go to Google — but more and more of them get answered before the user ever sees a search result. ChatGPT answers it. Perplexity answers it. Gemini answers it. Google's own AI Overview answers it. The user gets the answer. You get a citation — if you're lucky. Or nothing, if you're not.

This is GEO — Generative Engine Optimization — and it isn't a fad or a separate discipline. It's the new shape of the same problem SEO was always solving: how do you make sure the engine that mediates between users and your content picks your content?

The mistake most teams are making right now is treating GEO as a new project. Hire a new specialist. Buy a new tool. Bolt it onto the side of the existing SEO team. That's the wrong move. AI SEO and GEO are the same problem from two angles, and the smart play is to run them as a single unified strategy, powered by AI agents that read your real first-party data.

This piece is the practical playbook for doing that.

---

TL;DR

• AI SEO is using AI to optimise for traditional search engines (Google, Bing).
• GEO (Generative Engine Optimization) is optimising for citation by AI search engines (ChatGPT, Perplexity, Gemini, Google AI Overviews).
• The signals that drive both are 80% overlapping: authoritative content, structured data, clear definitions, E-E-A-T markers, internal linking, page experience.
• Running them as a unified strategy is dramatically more efficient than running them separately — one analysis, one content brief, one publish, two channels.
• The unified strategy needs three things: live data from Search Console + GA4 + AI traffic attribution, a model (Claude is the default in 2026) that can reason across both surfaces, and a workflow that turns insight into shippable action.
• Cogny Cloud at $530/month ships the full harness for this — connecting your real data via MCP, running unified SEO + GEO analyses on a schedule across every channel, and producing falsifiable Growth Tickets with the Truth Ledger to close the loop. Cogny Solo at $9/month is the entry tier (bring-your-own-Claude, starter MCP set, one channel at a time) — useful to start, but not the full harness.

---

Definitions, Cleanly

Before going further, let's pin down terms. Most of the confusion around AI marketing for search comes from people meaning different things with the same words.

SEO (Search Engine Optimization) is the practice of structuring a website so that traditional search engines — primarily Google and Bing — rank its pages high for relevant queries. The output is a ranking position. The success metric is organic clicks and conversions.

AI SEO is using AI to do SEO work. AI generates keyword analysis, identifies content gaps, audits technical issues, drafts content, builds internal linking strategies, predicts which pages will rank. The search engine being optimised for is still Google.

GEO (Generative Engine Optimization) is the practice of structuring content so that AI search engines — ChatGPT, Perplexity, Gemini, Claude itself, Google's AI Overviews, Bing Copilot — cite that content as the source of their answer. The output is a citation. The success metric is AI-referred traffic, citation share-of-voice, and influence on the AI's response.

AI search is the user behaviour: humans asking AI engines for answers instead of (or in addition to) typing queries into Google. AI search is what makes GEO necessary.

AI marketing for search is the unified discipline: using AI agents to optimise content for both Google ranking AND AI citation simultaneously, off the same data pipeline.

---

Why SEO and GEO Are Actually One Problem

The internet narrative in 2025 was that GEO was a brand new discipline requiring brand new tactics. That narrative was wrong, but it was useful for selling courses.

Here's what the actual signal overlap looks like once you dig into how both Google and the AI engines select content.

The signals that drive citation by ChatGPT or Perplexity are largely the same ones that drive ranking on Google. The differences are around how those signals are weighted and what surfaces the engine puts you on — not around what makes content good.

Practically, this means: content optimised for SEO is usually 80%+ of the way to being optimised for GEO. The remaining 20% is structural — making the content easier for an AI engine to parse, cite, and attribute.

The teams running them as separate disciplines are doing twice the work for marginal additional impact.

---

The Unified AI Marketing Strategy for Search

Here's the loop we recommend, and the one Cogny is built to run automatically.

Step 1: Connect Live Data

You need three sources, minimum:

1. Google Search Console — query/page performance, impressions, CTR, ranking position. Source of truth for SEO.
2. GA4 (or your analytics tool) — what users do after they arrive, conversion data, and increasingly, AI-engine referrers (ChatGPT, Perplexity, etc. now show up as identifiable sources).
3. Your CMS / content database — what you've published, when, and what topics it covers.

Optional but powerful:

• BigQuery export of GA4 — for deeper analysis, cohort tracking, AI-referrer attribution
• Search Console export to BigQuery — for query-level historical analysis beyond the 16-month limit
• First-party data from your business systems — customer records, revenue data, attribution

This is the data layer. Without live access to it, AI agents are guessing. With it, they're reasoning over your actual numbers.

Step 2: Run Unified Analyses

A serious AI marketing platform should run weekly or daily analyses that cover both SEO and GEO in one pass:

• Pages losing impressions — same diagnostic for SEO and GEO. The page that's losing Google impressions is probably also losing AI citations.
• Queries where you should be cited but aren't — for both Google (you rank #4–10 = SEO opportunity) and AI engines (you're a near-miss citation = GEO opportunity).
• Content gaps — questions being asked of AI engines that nobody on your site has answered well, AND queries with significant impressions and weak top results.
• Technical issues — schema errors, broken internal links, crawl failures hurt both surfaces.
• Authority signals — outdated author bios, missing E-E-A-T markers, weak source attribution. Hurts both.

Cogny's geo-search-optimization and geo-conversion-report templates encode this kind of analysis. They run on a schedule, read live data, and produce a unified opportunity queue.

Step 3: Generate Tickets, Not Suggestions

The output of each analysis should be specific, actionable, falsifiable tickets — not generic recommendations.

Bad: "Improve E-E-A-T signals on the blog."

Good: "Add author bio with 5+ external citation links to /blog/best-ai-seo-tools-2026. Currently no author byline; page ranks #14 for the target query despite 4,200 monthly impressions. Predicted ranking lift +6 positions over 30 days. Predicted AI-citation lift on Perplexity for best ai seo tools query."

The second one can be executed and measured. The first cannot. We wrote about why this matters at the schema level.

Step 4: Execute and Measure

A human approves. The change ships. The platform measures the result on a schedule appropriate to the action:

• Ranking changes — 30/60/90 days
• AI citation changes — 30 days (AI engines update their knowledge faster than Google)
• Technical fixes — 7–14 days for crawl to reflect
• Content publishes — 60–90 days for full ranking + citation maturity

The measurement gets written back to the system. The next analytical cycle reads it and recalibrates.

Step 5: Compound

Six months in, the system knows which kinds of changes work in your specific site. The recommendations get sharper. The approval rate climbs. The AI citation share-of-voice grows. The Google rankings improve. The two compound on each other because the underlying content quality is rising on both axes.

This is what AI marketing for search looks like when it actually works.

---

The Specific Techniques That Drive Both

A few specific patterns that punch above their weight for unified SEO + GEO.

1. Lead Every Page With a Crisp Definition

If your page is about "AI marketing harness," the first 80 words should define what an AI marketing harness is, cleanly, in plain language. Google's featured-snippet algorithm and Perplexity's citation extractor are doing the same job — looking for a concise authoritative answer near the top of the page.

We do this on every Cogny piece. The TL;DR section near the top is not just user-friendly. It's citation-friendly.

2. Use Comparison Tables Generously

AI engines pull comparison tables out of pages and surface them in answers. So does Google for "X vs Y" SERPs. The marginal cost of writing a table is small. The marginal upside on both surfaces is large.

3. FAQ Sections With Schema

Every pillar page should have an FAQ section, with FAQPage schema markup. Google uses it for rich SERPs. AI engines use it for direct Q&A citation. The same questions, marked up the same way, work on both.

4. Specific, Quantified Claims

Vague claims don't get cited. "Cogny is fast" is uncitable. "Cogny runs 39 analytical templates across 25+ data sources and produces falsifiable tickets within 24 hours of connection" is citable. Specificity is the universal currency of citation.

5. Real Author Bios With External Authority Signals

Anonymous content gets less weight in both surfaces. An author bio with a real photo, real credentials, and links to public work (LinkedIn, conference talks, GitHub) is an E-E-A-T signal Google reads and an authority signal AI engines weight.

6. Topic Clusters With Tight Internal Linking

Don't write one isolated page on a topic. Write a pillar page plus 4–7 supporting pages, all interlinked. Google reads this as authority on the topic. AI engines read this as "this site has covered the topic comprehensively" and prefer it as a citation source.

7. Fresh Updates With Visible Dates

Both surfaces decay stale content. Update key pages at least quarterly, with a visible "Updated: [date]" marker. The recency signal is one of the strongest in GEO.

8. Structured Data Beyond the Basics

Article, FAQ, BreadcrumbList, Organization schema are table stakes. Going further — ItemList, HowTo, ProfessionalService, Product schema where appropriate — gives both Google and AI engines more structured surface to lift.

---

The Role of First-Party Data

Here's the part of the story that's underrated. Most "AI SEO tools" out there don't read your real numbers. They scrape SERPs, run keyword tools, and produce generic content recommendations.

That's not AI marketing. That's a content marketing tool with AI features bolted on.

Real AI marketing for search uses your first-party data:

• Your actual Search Console queries (not what some third-party tool guesses you might rank for)
• Your actual GA4 conversions (not what the industry average says)
• Your actual AI-referrer traffic (which most generic tools don't even track)
• Your actual customer data (so recommendations are grounded in what converts for you)

The reason this matters: a content recommendation is only as good as the data it's grounded in. "This topic gets 50,000 monthly searches" is useless if the topic doesn't convert for your business. "This topic gets 8,000 monthly searches, you currently rank #18, your competitors get cited by Perplexity for it, and you have three customer case studies that prove authority on this topic" — that's a recommendation you can actually ship.

This is why MCP matters so much for AI marketing. The MCP primer covers it in depth. MCPs let Claude (or any compliant AI client) read your real numbers from Search Console, GA4, BigQuery, and the rest. The AI marketing strategy is grounded in your data, not the industry's averages.

---

How AI Engines Are Different From Google (And Why It Matters)

The two surfaces are similar enough to optimise together. But the differences are worth knowing.

Google ranks pages. AI engines cite passages. Google sends the user to your page. AI engines lift a passage from your page into their answer and cite the source. This means which paragraph on your page matters for GEO in a way it doesn't for SEO.

AI engines update faster. A new article can be cited by Perplexity within hours of publishing. Google ranking takes weeks. This makes GEO a faster feedback loop — useful when you're testing whether a topic resonates.

AI engines weight authority differently. Backlinks are still huge for Google. AI engines weight them less and weight in-content authority signals (citations within the text, author credentials, structured data) more.

AI engines surface niche queries. A long-tail question like "what is a marketing MCP and why does it matter for AI-powered ad spend optimization" might have ~50 Google searches a month — and be a very valuable AI citation if a high-intent buyer asks it of Perplexity.

Click-through behaviour is different. Google sends the click; you get the visitor. AI engines often answer in-place; you get a brand impression and maybe a click-through. The conversion model for GEO is closer to PR than to direct response, and the measurement has to adapt.

The unified strategy accounts for these differences but doesn't fragment the team or the tooling. Same data, same analysis, same content publish, two surfaces.

---

Measuring Success Across SEO and GEO

A few metrics worth tracking in a unified dashboard:

SEO metrics (you probably already track these)

• Organic impressions and clicks (Search Console)
• Average position by query and page
• Ranking trajectory over 30/90 days
• Conversion rate from organic traffic
• Revenue from organic

GEO metrics (newer; track them now)

• AI-referrer traffic (ChatGPT, Perplexity, Gemini, etc. — visible in GA4 referrer reports)
• AI citation share-of-voice for key queries (sampling-based monitoring)
• Branded query lift in AI engines (an indirect signal — when AI tools surface your brand, users follow up with branded searches)
• Conversion rate from AI-referred traffic (often higher than organic average)

Shared health metrics

• E-E-A-T signal completeness on key pages (author bios, dates, schema)
• Page experience scores
• Internal link density on pillar topics
• Content freshness scores

Cogny's geo-conversion-report template tracks the AI-referrer side specifically, with dual BigQuery + GA4 MCP support so you can answer the question "what did AI search drive in revenue last month?" — a question most teams currently can't answer at all.

---

The Strategic Case for Doing This Now

There are two timing arguments for moving on unified AI marketing in 2026.

1. The AI surface is still being mapped. ChatGPT, Perplexity, and Gemini have not converged on a stable citation model. The brands that establish authority in this window will benefit disproportionately for the next several years. By 2027, the strong citation patterns will be largely set.

2. The cost of running AI agents has collapsed. The same analysis that cost $50 in inference 18 months ago costs $5 today. Tomorrow it'll cost $1. The implication is not "run AI for cheaper tasks" — it's "run AI 50x more often." Hourly monitoring of search visibility across both surfaces is now economically obvious.

The teams who wait will face the same setup work in 2027 with a smaller window of advantage. The teams moving now build durable position.

---

How to Start

If you want to run this strategy without building the infrastructure yourself:

Cogny Cloud at $530/month is the full configuration: scheduled SEO + GEO analyses running weekly across every channel, the complete 25+ MCP catalogue, the Growth Tickets queue with falsifiable hypotheses, and the Truth Ledger that closes the loop. This is the version that compounds.

Cogny Solo at $9/month is the entry tier — bring-your-own-Claude, starter MCP set (Search Console included), one channel at a time, 7-day free trial. A useful place to test the model on your real Search Console and GA4 data before stepping up to the Cloud configuration.

If you want to keep your current SEO tooling and add the AI agent layer on top, Cogny's MCP server is available as a standalone — bring your own Claude or Claude Code, point it at our MCP, and ask questions of your data directly.

Either path beats running SEO and GEO as separate disciplines with separate tools.

---

FAQ

What is GEO (Generative Engine Optimization)? GEO is the practice of structuring content so that AI search engines like ChatGPT, Perplexity, Gemini, and Google AI Overviews cite that content as the source of their answer. Unlike traditional SEO which optimises for ranking position, GEO optimises for citation. The signals overlap significantly with SEO — authority, structured data, clear definitions — but the success metric is being the source of the AI's answer, not being a ranked search result.

Is GEO replacing SEO? No, both matter. Google still drives substantial click-through traffic and will for years. But the share of high-intent queries that get answered in-place by AI engines is growing fast. The right strategy is to optimise for both off the same content pipeline, not to pick one or replace the other.

How is AI SEO different from regular SEO? AI SEO uses AI to do SEO work — keyword analysis, content gap identification, technical audits, content briefs. The target is still Google. AI marketing for search goes further: same AI engine optimising for both Google and the AI search engines, off the same data, on the same schedule.

What data sources do I need? At minimum: Google Search Console (for query/ranking data), GA4 (for behaviour and AI-referrer attribution), and your CMS or content database. Adding BigQuery exports of both gives you deeper historical analysis. First-party customer data adds another layer of grounding.

Does Claude work for both SEO and GEO? Yes. Claude is the default model for serious AI marketing platforms in 2026. Its strength at structured reasoning over tabular data fits SEO analytics, and its calibrated approach to citation and source-attribution makes it well-suited to GEO. Cogny's harness runs Claude across both surfaces.

How long until I see results? SEO changes show in 4–8 weeks as Google reindexes. GEO changes can show in 1–4 weeks — AI engines update their knowledge much faster than Google. The compound effect kicks in around month 3, once enough cycles have run for the system to calibrate on your specific site.

Is the unified SEO + GEO strategy more work than running them separately? Less work, not more. The signals overlap 80%+. One analysis covers both. One content brief covers both. One publish, two surfaces. Running them as separate teams with separate tools is the expensive way to do it.

What about local SEO and GEO? Local search is interesting because AI engines weight local intent heavily — "best plumber in Stockholm" is a high-citation-volume query for AI engines, and proximity signals matter. Cogny's local-seo template covers the unified local strategy.

---

About Tom Ström

Tom is CEO and co-founder of Cogny. He spent eleven years building Campanja, the AdTech platform behind growth campaigns for Netflix, Zalando, and Telia, before starting Cogny to build the AI marketing harness for the post-search-engine era. He writes about AI marketing, SEO, GEO, and what changes when AI agents have real-time access to your first-party data.

Run the unified SEO + GEO strategy against your own data

The full configuration — scheduled unified analyses across every channel, all 25+ MCPs, Growth Tickets, the Truth Ledger — ships with Cogny Cloud at $530/month.

Cogny Solo at $9/month is the entry tier: bring-your-own-Claude with a starter MCP set including Search Console, 7-day free trial. The right place to start if you want to test the model on your real search data before committing to the full Cloud configuration.