AEO for SaaS Founders: Get Cited by ChatGPT, Perplexity, and Gemini

Organic Traffic to SaaS Sites Is Down 40-60% Since AI Overviews Launched

If you run a SaaS site, you have already felt this. Search Console graphs that look like ski slopes. Articles that used to bring in 2,000 visitors a month now bring in 300. Position 1 rankings on informational keywords getting fewer clicks than position 8 used to get a year ago.

Generic "what is X" content gets close to zero clicks at position 1 now. AI Overviews answer the query inside Google, ChatGPT answers it before users even visit Google, and Perplexity wraps it all up with citations that point back to a few authoritative sources, not to you. The game changed in 2024. Most SaaS founders are still writing for the old game.

Here is what works now, and how to get cited by ChatGPT, Perplexity, and Gemini instead of summarized away.

What's Actually Dying

Before talking about what works, let's be specific about what is dead. Not "evolving." Not "becoming less effective." Dead. If you are still publishing content in these categories, you are paying writers to produce inventory that will never get read.

Generic Informational Queries

Searches like "what is product market fit," "what is churn rate," or "what is a value proposition" used to drive massive traffic. AI Overviews now answer these in two sentences directly on the SERP. Click-through rates on the top organic result for these queries dropped from around 30% to under 5% in many SaaS niches. The clicks are gone. They are not coming back.

Top-Funnel Keyword Content

The classic SaaS playbook of writing 50 top-of-funnel articles to capture broad keywords and convert later through email is broken. Top-funnel queries are exactly what AI handles best. Users no longer click through to read your "ultimate guide to email marketing." They get a summary, ask follow-up questions, and never visit a website.

Thin Programmatic SEO Pages

Programmatic pages built around templates ("[city] + [keyword]" or "[competitor] alternatives" with 200 words of templated copy) are getting deindexed. Google's helpful content updates penalize these explicitly. AI tools also ignore them as low-quality sources. If you have a folder of 5,000 templated pages, check your Search Console. Most of them are zombies.

"What Is" Articles

Any article whose primary intent is defining a term is dead inventory. ChatGPT defines it. Wikipedia defines it. Investopedia defines it for finance terms. Your blog has zero chance of being the source AI cites for a definition. Stop writing them.

Generic "Best Practices" Lists

"10 tips for better email subject lines." "7 ways to improve your homepage." These were already weak in 2022. In 2026 they are invisible. AI generates this exact format on demand. There is no reason for anyone to read your version unless it contains data or specifics no one else has.

Here is the test: if a 7B parameter open-source model running on a laptop can write a credible version of your article in 30 seconds, you should not be publishing that article. AI is the floor for content quality now. You have to clear it.

What Survives AI Overviews

Six content types still drive traffic and citations. Ranked by how durable they are against further AI improvement.

1. Interactive Tools

This is the most defensible category. AI cannot replicate an interactive experience. When someone uses our SaaS Positioning Grader, they get a personalized score for their actual homepage in 10 seconds. There is no way for ChatGPT to give them that. The tool itself becomes the destination, the result page becomes a shareable artifact, and the tool generates proprietary data we can write about (which feeds the next category).

If you can ship one piece of content this quarter, ship a tool. A calculator, a grader, a checker, an analyzer. Even a simple one beats 20 blog posts.

2. Original Research With Proprietary Data

"We analyzed 100 SaaS homepages and found 80% fail the 5-second clarity test." That sentence is citable. AI cannot generate it without you. Proprietary research positions your brand as the source other people quote, including AI tools. When ChatGPT is asked about SaaS homepage patterns, it pulls from sources that have actual data. Your "10 tips" article does not qualify. Your "we ran a study on 100 sites" article does.

3. Specific Teardowns With Real Screenshots

Our clarity maps teardown specific SaaS homepages with screenshots, named patterns, and concrete copy rewrites. AI cannot produce a teardown of Notion's homepage without us doing the work first. These are the kind of pages Perplexity loves to cite because they contain specific, verifiable claims tied to a real artifact.

4. Implementation Guides With Exact Steps

"How to add Article schema to a Laravel Blade template" beats "what is structured data." Step-by-step guides for narrow technical problems still get traffic because users need to see the exact code, the exact configuration, the exact button to click. AI helps with these but often gets details wrong, so users still verify against a written guide.

5. Product Comparisons With Nuanced Takes

"Linear vs Jira" content used to be SEO bait. Now it works again, but only if you have actually used both products and have an opinion. Generic comparison tables get summarized by AI. Comparisons that include "Linear feels lighter for engineering teams under 50, Jira wins once you have multiple product lines because of its custom field flexibility" survive. Nuance from real experience is uncopyable.

6. Strong Opinion Content

AI gives neutral, balanced summaries. Humans want takes. "Email marketing is overrated for B2B SaaS at under 1,000 customers" is the kind of headline that gets shared, debated, and linked to. ChatGPT will never write that sentence with conviction. Neutral content is commodity. Opinionated content is brand.

How to Write Content ChatGPT Will Cite

Getting cited by AI is not random. There are specific patterns AI tools use to pick sources. Here is what actually moves the needle.

Be the Authoritative Source on a Narrow Topic

AI tools cite sources they trust on the specific topic. They do not pull from generalists. The path to citation is owning a niche so thoroughly that any query in that niche surfaces your content.

Pick one narrow topic. SaaS homepage positioning. Cold email deliverability. B2B pricing pages. Whatever it is, publish 20 deeply specific pieces on that topic before publishing one piece on anything else. Topical density signals authority. AI tools weigh that heavily.

Lead With Extractable Specific Facts

AI tools extract sentences. They are looking for sentences that contain a complete, citable fact. Compare these two openings:

"Many SaaS homepages have positioning issues that affect their conversion rates."

"After grading 50 SaaS homepages, 78% scored under 60/100, with the average headline scoring 42/100 on clarity."

The second sentence is the kind ChatGPT pulls and cites. The first is filler. Every paragraph of your article should have at least one extractable fact: a number, a percentage, a named pattern, a specific URL, a dated event.

Use Clear H2 Questions and H3 Specifics

Structure your content the way users actually ask AI questions. H2s should be questions: "How long does AEO take to work?" "What schema markup matters most?" H3s should be specific subtopics under those questions. AI tools chunk content by heading and look for sections that directly answer the query. If your H2 is "Best practices," AI cannot tell what query it answers. If your H2 is "How do I get cited by Perplexity?" the answer is obvious.

Add Structured Data

Article schema, FAQ schema, and HowTo schema in JSON-LD. This is non-negotiable for AEO in 2026. Gemini in particular pulls heavily from structured data because it is integrated with Google's existing infrastructure. ChatGPT and Perplexity use structured data as a quality signal. If two competing pages have similar content but only one has FAQ schema, the schema page wins citations.

For SaaS founders running Laravel, drop a JSON-LD block in your article template. It takes 30 minutes and pays off for years.

Include FAQ Sections With Explicit Answers

FAQs are the highest-density format for AI extraction. Each Q&A is a self-contained fact AI can pull. Add 5-7 FAQs at the end of every long article. Make the questions match real queries users would type into ChatGPT. Make the answers under 100 words and self-contained (no "as we discussed above").

Build Topical Clusters

One pillar article plus 8-12 supporting articles, all internally linked. AI tools follow internal links to assess depth. A site with 12 articles on AEO, all linked together, looks more authoritative than a site with one AEO article and 50 unrelated posts. Cluster strategy is back, but now it is about AI authority signals, not just Google PageRank.

Update Regularly

Perplexity heavily weights recency. ChatGPT prefers content updated in the last 6-12 months. Gemini favors fresh sources. Pick your top 10 articles. Schedule quarterly updates. Add new data, refresh stats, note any pattern changes. The "last updated" date matters more than the original publish date.

How to Optimize for Perplexity vs ChatGPT vs Gemini

The three big AI tools have different ingestion patterns. If you want citations from each, optimize for each.

Perplexity

Perplexity is the easiest to win. It runs live web searches, weights recent content heavily, and shows citations directly in its responses. To get cited:

• Publish original research with clear data points (Perplexity loves numbers it can quote)
• Update content frequently (Perplexity preferences sources updated within the last 6 months)
• Add prominent author bylines and dates (visible authorship boosts trust signals)
• Write extractable single-sentence claims (Perplexity pulls quotes verbatim)
• Get cited elsewhere first (Perplexity uses link signals from other sites it already trusts)

ChatGPT

ChatGPT is harder because base model knowledge is trained on a snapshot of the web, not live search. Browse mode pulls live, but most queries hit base knowledge. To get into the base training set:

• Publish on a domain with sustained authority (ChatGPT's training data weights established sources)
• Get referenced by Wikipedia, major publications, or industry roundups (training data secondary sources matter)
• Use clear, declarative sentences (ChatGPT's training prefers prose over marketing fluff)
• Build content that gets shared and quoted (frequency in the training corpus correlates with citation likelihood)
• For browse mode specifically: structured data and FAQ format help

Gemini

Gemini is integrated with Google. If your content does well in Google Search and AI Overviews, it tends to do well in Gemini. To optimize:

• Structured data is the biggest single lever (Gemini relies on it heavily)
• Featured snippet formatting (definition lists, ordered steps, tables) gets pulled directly
• E-E-A-T signals matter: clear author bios, credentials, and expertise indicators
• Site architecture from Google's perspective: clean URLs, sitemap, fast loading
• If you already rank well organically, Gemini citations follow

The 7-Step AEO Checklist

Run every article you publish through this checklist before hitting publish. If any item is missing, fix it first.

1. Title under 60 characters with primary keyword. AI tools use the title as a quality signal. Over 60 characters and Google truncates. Include your target query in the first 5 words.
2. Meta description 50-160 characters with a click reason. Tell the user what they will get specifically. "Learn AEO" is dead. "The 7-step AEO checklist used to get 4x more AI citations in 90 days" works.
3. Article + FAQ schema markup in JSON-LD. Both. Article schema for the post itself, FAQ schema for any Q&A section. Test with Google's Rich Results Test before publishing.
4. Lead paragraph answers the query in 2-3 sentences. Inverted pyramid. The first paragraph should be standalone-citable. Save the long preamble for never.
5. H2s structured as questions readers ask AI. Pull from "People Also Ask" and from your own ChatGPT logs. The H2 wording should match how users actually phrase queries.
6. At least one original data point or proprietary framework. A statistic from your tools, a pattern from your customer research, a named methodology you created. If everything in the article is available elsewhere, it will not get cited.
7. Internal links to your tools and related articles. Every article should link to at least one tool (where users can act) and 3-5 related articles (cluster signal). Use descriptive anchor text, not "click here."

What We Did at Growth Pigeon

This is the playbook we are running on Growth Pigeon. Real moves, real results so far. I am writing this in 2026, about 7 months into the strategy.

Built the Positioning Grader (Interactive Tool, AI Cannot Replicate)

The SaaS Positioning Grader takes a URL and returns a 6-dimension score in 10 seconds. AI tools cannot grade your specific homepage. They can talk about positioning principles, but the grader gives users a personalized artifact tied to their actual page. The result page becomes shareable, indexable, and serves as a destination AI sometimes points users toward.

Published Clarity Maps (Proprietary Teardowns)

Our clarity maps are detailed positioning teardowns of specific SaaS companies. Each one names patterns, includes screenshots, and proposes specific copy rewrites. These are the kind of pages Perplexity cites when someone asks "how is Linear's homepage positioned." We have over 80 published. Perplexity has started surfacing them in citation lists for SaaS positioning queries.

Created the BELT Framework (Named, Citable Framework)

The BELT framework measures SaaS product durability across Behavior attachment, Economic moat, Loop strength, and Time-to-value. Naming the framework matters. ChatGPT can cite "the BELT framework from Growth Pigeon" in a way it cannot cite "a framework about SaaS retention." Named frameworks become referenceable nouns. Unnamed concepts disappear into the noise.

Sold the Clarity Map Product

The paid clarity map product ($37) gives founders a custom positioning teardown of their own site. Tools and content drive traffic. The paid product converts that traffic. AEO without a clear conversion path is just expensive vanity.

Results So Far

Brand searches for "growth pigeon" are up roughly 4x quarter over quarter. AI citation tracking (using a mix of manual ChatGPT testing and Perplexity reverse search) shows we are now cited in roughly 12% of relevant queries, up from 0% six months ago. Tool usage has grown from a couple dozen graders per week to a few hundred. None of this comes from generic "what is positioning" content. All of it comes from tools, teardowns, and named frameworks.

FAQ

What is AEO?

AEO stands for Answer Engine Optimization. It is the practice of structuring content so AI tools (ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews) cite your work as a source when they answer user queries. Where SEO targets ranking in search results, AEO targets being the source AI uses to generate its answer.

How is AEO different from SEO?

SEO optimizes for click-through from search results. AEO optimizes for citation in AI responses. SEO rewards broad keyword targeting and long-form content. AEO rewards specific, extractable claims with original data. SEO traffic is a user clicking your link. AEO "traffic" is a user reading about your brand inside an AI response and then either visiting directly or remembering you for later. The two strategies overlap but are not identical. AEO weights structured data, original research, and topical authority more heavily than traditional SEO does.

Does schema markup help with AI citations?

Yes, especially for Gemini and Google AI Overviews. Schema markup gives AI tools a structured map of your content's meaning. Article schema, FAQ schema, and HowTo schema are the most useful for SaaS content. Adding schema is one of the highest-impact AEO moves because it costs almost nothing in time and signals quality across multiple AI systems. Run your pages through Google's Rich Results Test to verify the schema is valid.

How long until AEO works?

For Perplexity, citations can start within 2-4 weeks of publishing strong original research. Perplexity uses live search and weights recency. For Gemini, expect 1-3 months as Google indexes and ranks your content. For ChatGPT base model knowledge, expect 6-18 months because the training data is updated periodically rather than continuously. ChatGPT browse mode picks up content faster, similar to Perplexity. Plan AEO as a 6-12 month strategy, not a 2-week experiment.

Should I stop doing SEO?

No. Stop doing the parts of SEO that are dead (generic informational content, thin programmatic pages, top-funnel keyword stuffing). Keep the parts that still work (specific bottom-funnel content, internal linking, technical SEO basics, fast site speed, mobile optimization). AEO is mostly an extension of strong SEO, not a replacement. The technical foundations are the same. The content strategy is what changes.

What is the single biggest AEO mistake?

Writing generic content because it is fast to produce. Most SaaS founders publish "10 ways to improve X" articles because they can write them in an hour. Those articles are invisible to AI. One specific teardown with original data outperforms 20 generic listicles for AEO. Quality and specificity beat volume now.

How do I track AI citations?

There is no perfect tool yet. Manual testing works: ask ChatGPT and Perplexity the queries you want to be cited for, and see if your domain appears in the citations or response. Tools like Profound, Otterly, and SearchAtlas are starting to track AI mentions, but they are still early. Set a quarterly review cadence: pick 20 target queries, run them through each AI tool, log the citations you get. Track the trend over time.

What to Do This Week

If you only do three things from this article, do these:

1. Pick one narrow topic you want to own. Audit your existing content. If you have less than 10 deep articles on it, plan the next 10.
2. Build one interactive tool, even a simple one. A calculator, a checker, a grader. Tools survive AI better than any other content type.
3. Add Article and FAQ schema to your top 10 articles. This is a 2-hour task that pays off for years.

The SaaS founders who survive the AI Overview era are the ones who treat AEO as a real strategy, not a footnote. Generic content is over. Specific, opinionated, data-rich, tool-supported content is the only thing left. Build for that.

If you want to see what AEO-optimized SaaS positioning looks like in practice, run your homepage through the positioning grader or order a clarity map. Both are designed to be citable, specific, and useful, exactly what AI tools surface and exactly what your prospects need.