Why ChatGPT Cites Travel Sites More Than Education (And What That Means For Your AI Search Strategy)
Citation rates in ChatGPT aren’t uniform—they swing wildly by topic. Travel answers link out far more often than education, and the sites being cited change by vertical. Here’s what’s really happening, why it matters for SMEs and agencies, and how to build an AEO/GEO plan you can actually execute.
AI Search is forcing every business to relearn a basic truth: visibility is not evenly distributed. The same question asked in two different topics can produce completely different outcomes—different sources, different links, and sometimes no links at all.
A recent report covered by Search Engine Journal highlighted something many teams are only starting to measure: ChatGPT’s tendency to include citations varies sharply by topic. In Similarweb’s analysis (U.S. desktop), travel answers contained citations far more often than education answers. That gap isn’t just interesting—it changes what “winning” looks like in AI-assisted discovery.
This editorial is a practical playbook for SMEs, ecommerce operators, local businesses, and agencies who need to respond with strategy and execution. I’ll explain what’s changing, why travel is different, what can go wrong if you chase the wrong KPI, and how to build a durable AEO/GEO plan that fits your vertical.
Concise summary
- ChatGPT citations aren’t consistent across topics. Some verticals naturally trigger more “check the web” behavior than others.
- More citations = more surface area to be discovered. If your topic sees fewer cited answers, your “available Impressions” inside AI answers can be structurally limited.
- Different topics cite different kinds of sites (UGC/reviews vs. publishers vs. ecommerce). A single universal AEO playbook is a trap.
- Execution speed matters. AI-driven SERPs and chat answers evolve quickly; delayed approvals and slow implementation are now direct visibility costs.
- AYSA’s role: monitor AI visibility, prepare prioritized site changes, request approval, then execute accepted updates—so strategy doesn’t die in backlog.
Key takeaways (what to remember)
- Citation rate is a vertical-dependent KPI. Treat it like “market conditions,” not a universal benchmark.
- “Be cited” is not the only win condition. When citations are rare, you focus more on becoming the underlying source model answers draw from (and on conversion when traffic does arrive).
- UGC and reviews are a major citation category in many topics, especially travel (per Similarweb data cited by SEJ). That implies reputation and distribution matter as much as on-site content.
- Build for retrieval: clarity, freshness, and verifiability increase the chance an AI system will look outward and include references.
Table of contents
- What changed: ChatGPT citation behavior is not uniform
- What “citations” mean in AI answers (and what they’re not)
- Why travel gets more citations (and why that’s logical)
- Why education may get fewer citations (and what you can still win)
- Why no single AEO/GEO playbook works across every vertical
- What can go wrong when you optimize for AI citations
- New measurement mindset: KPIs that matter beyond “I got cited”
- A concrete SME scenario: the independent hotel vs. the tutoring center
- Action plan: how to improve your chances of being cited (without guessing)
- Where AYSA fits: monitored, approved, executed AI search optimization
- What to do next
- Sources and further reading
What changed: ChatGPT citation behavior is not uniform
The headline insight from the SEJ coverage is simple: Similarweb observed that in May (U.S. desktop), ChatGPT included citations far more frequently for travel questions than for education questions. Across all topics, Similarweb reported an average citation inclusion rate that was materially lower than travel and higher than education.
Two implications matter immediately:
- The “citation opportunity pool” differs by topic. If only a small fraction of answers in your vertical include links, then even perfect optimization has a ceiling: there are fewer moments where a citation could appear at all.
- AI visibility becomes a market-structure problem. This isn’t just “do better SEO.” It’s “understand how the system behaves in your category, then allocate resources accordingly.”
SEJ also notes Similarweb saw overall citation inclusion increase over the prior year (with volatility). The point for operators is not the exact curve—it’s that this surface area is changing quickly, and it may shift without warning because model behaviors, product UX, and safety policies shift.
That’s why, at AYSA.ai, I treat AI search visibility as something you monitor continuously, not something you “fix once.” If you’re not watching the system, you’re arguing about yesterday’s search reality.
What “citations” mean in AI answers (and what they’re not)
In classic Google search, visibility is mostly about ranking and snippets. In AI chat answers, visibility is often about:
- Whether your site is referenced (linked/cited).
- Whether your information is used (paraphrased or synthesized) even without a link.
- Whether the answer suggests a brand by name (also possibly without a link).
A citation is the most obvious and measurable artifact, because you can see the link. But citations are not a perfect proxy for influence. A model can “learn” from patterns in public content or rely on general knowledge, and still answer without linking. Or it can link to a third party that summarized you.
So when we talk about citations as a KPI, we need to keep two realities in mind:
- Citations are a product choice as much as a relevance outcome. The UI decides when to show them, how many, and where.
- Citations are an ecosystem outcome. In some verticals, the most “citable” sources are aggregators, review platforms, or publishers—not brand sites.
Still, citations matter because they do three valuable things for SMEs:
- They create a direct click path out of the AI answer.
- They create trust transfer (“this answer is supported by sources”).
- They create a repeatable visibility unit you can monitor and improve.
Why travel gets more citations (and why that’s logical)
Even without seeing Similarweb’s full gated methodology, the behavior described in SEJ aligns with something we can reason about operationally: travel is inherently dynamic and comparison-heavy.
Think about what people ask in travel:
- “Is this neighborhood safe right now?”
- “What’s the best time to visit?”
- “Which hotel is better for families?”
- “What are the latest entry requirements?”
- “What’s open late on Sundays?”
Those questions share three properties that tend to encourage linking out:
1) Freshness pressure
Travel information changes: hours, seasons, construction, closures, pricing, events, airline schedules, advisories. When a system detects high volatility, it has a rational incentive to check the web and attribute sources.
2) Heavy comparison intent
Travel planning is a string of comparisons: destinations, routes, hotels, tours, restaurants. Comparisons naturally benefit from multiple sources, and citations are the product’s way of signaling “I didn’t make this up.”
3) Experience-based evidence (reviews/UGC)
SEJ’s write-up notes that for travel, a high share of citations were categorized as reviews and UGC in Similarweb’s data. Again, that tracks with user behavior: travelers trust recent experiences and crowd consensus.
Business implication: in travel, you’re not just optimizing your site—you’re optimizing your presence across the review and UGC ecosystem, because that’s what AI systems appear to cite most often (per the Similarweb categorization referenced by SEJ).
Why education may get fewer citations (and what you can still win)
Education queries can be very different:
- More conceptual (“how to learn algebra,” “what is photosynthesis”).
- More evergreen.
- More standardized.
If a question is evergreen and can be answered confidently from general knowledge, an AI system may “feel less need” to link out every time. That could help explain (at least directionally) why education saw lower citation inclusion in Similarweb’s snapshot described by SEJ.
But lower citation frequency doesn’t mean “no opportunity.” It means the win condition changes:
- Own the branded layer: become the organization people ask for by name (“best SAT tutor in Austin”).
- Be the canonical explainer: deep, structured pages that answer a full cluster of questions can still be used—even if not cited.
- Win on conversion: if you get fewer AI citations, each click matters more. Improve landing pages, proof, pricing clarity, and contact paths.
This is where many SMEs get trapped: they chase “citations” like backlinks in 2012. But if your topic structurally produces fewer citations, your best ROI may come from being the best destination for the traffic you do get, and from improving the signals that turn AI mentions into customers.
Why no single AEO/GEO playbook works across every vertical
SEJ’s summary of Similarweb’s breakdown reinforces an uncomfortable truth for agencies selling “AI Optimization packages”: what works in one topic may underperform in another, because AI systems cite different site types depending on the category.
In Similarweb’s categorization (as reported by SEJ):
- Across all answers, reviews/UGC and news/publishers were major citation buckets.
- Beauty leaned heavily toward retail/ecommerce citations.
- Finance leaned toward finance-specific sites and publishers.
- Travel leaned heavily toward reviews/UGC.
This forces a strategic decision: are you trying to be cited as a brand site, or are you trying to be present where the AI systems tend to cite third parties?
Distribution is now part of “SEO” again
For years, many teams tried to make SEO purely an on-site game: publish content, optimize technicals, wait. AI citation patterns pull us back toward a broader definition:
- Reputation (reviews, mentions)
- Publisher relationships
- Data partnerships and feeds (where applicable)
- Product page quality (for ecommerce-heavy citation categories)
That doesn’t mean “spray PR everywhere.” It means build your authority in the places the AI system already trusts for your topic.
Why this matters for SMEs and agencies
If you’re an SME, you can’t afford to waste quarters on the wrong playbook. If you’re an agency, you can’t sell a single “AI SEO retainer” without first answering: what does the model tend to cite in this vertical?
At AYSA.ai we treat that question as a monitoring problem first, and a content problem second. If you don’t have measurement, you’re just rebranding opinions as strategy.
What can go wrong when you optimize for AI citations
AI search optimization has real traps. Here are the big ones I see.
1) Turning citations into a vanity metric
A citation is not a sale. If you optimize for “being linked” without aligning to customer intent, you’ll generate visibility that doesn’t convert. For SMEs, that’s fatal because budgets are finite.
2) Copycat content that blends into the web
When teams hear “travel is cited a lot,” they often respond with generic city guides and “top 10” lists. That content is easy for AI to summarize and easy for big publishers to outcompete.
Better approach: build assets that are specific, operational, and verifiable (policies, live constraints, original photos where appropriate, clear pricing, unique local expertise, tools/templates).
3) Fighting the citation ecosystem instead of joining it
If your topic is UGC-driven (like travel, per the Similarweb categorization cited by SEJ), and you ignore reviews, you’re opting out of the very sources AI systems appear to cite most.
4) Strategy that dies in approvals and backlogs
AI search changes are often small but numerous: page structure tweaks, FAQ sections, clarifications, internal link improvements, schema updates, content refreshes. If you can’t execute quickly, your “AI strategy” becomes a slide deck.
This is exactly the category of work AYSA is built for: monitoring, preparing changes, requesting approval, and then implementing what’s approved. (More on that below.)
New measurement mindset: KPIs that matter beyond “I got cited”
SEJ’s piece is valuable because it pushes the industry toward measurement. But measurement has to be business-grade, not hype-grade.
Here’s the KPI hierarchy I recommend for 2026 planning:
Tier 1: Revenue-adjacent outcomes
- Qualified leads / bookings / purchases from AI-referred traffic (where referral is detectable)
- Conversion rate on AI-entry landing pages
- Brand search lift (people search your brand after seeing you in AI answers)
Tier 2: Visibility and influence
- Mentions in AI answers (linked or unlinked, depending on platform reporting)
- Citations/links in AI answers
- Share of voice vs. competitors for your money queries
Tier 3: Leading indicators you can control
- Content freshness cadence (last updated, change logs)
- Coverage of comparison and decision queries
- Authority signals: expert authorship, clear policies, transparent pricing
- Indexability and technical health (so your content is retrievable)
Notice what’s missing: “number of AI tools we’re on” and “how many prompts we tested.” Activity is not progress.
If you need a starting point for monitoring, AYSA has dedicated pages on AI search visibility and monitoring that reflect this KPI mindset: measure first, then execute.
A concrete SME scenario: the independent hotel vs. the tutoring center
Let’s make this real with two businesses of similar size and marketing budget.
Scenario A: Independent hotel in a seasonal destination
What people ask in AI: best areas to stay, parking rules, beach access, pet policies, seasonal events, weather, how to get from the airport, “hotel vs. Airbnb” comparisons, family-friendly recommendations.
Why citations are plausible: answers depend on changing details (events, hours), comparisons, and experience-based evidence (reviews). That aligns with Similarweb’s reported travel behavior: more citations and more UGC/reviews being referenced (as summarized by SEJ).
What the hotel should do:
- Create “decision pages” that match AI comparison prompts (parking, fees, check-in, amenities, neighborhood guide).
- Build a small set of evergreen-but-updated local pages (events calendar highlights, “what’s open late,” transportation options) and refresh them frequently.
- Strengthen review ecosystem presence (respond, keep details consistent, encourage recent reviews ethically).
- Make policies explicit and easy to quote (pet policy, cancellation terms, accessibility).
Scenario B: Tutoring center focused on math and test prep
What people ask in AI: study strategies, what to expect on exams, conceptual explanations, how many hours to study, the difference between tutoring options.
Why citations may be less frequent: many questions are evergreen and can be answered without live web validation—potentially reducing link-outs (consistent with Similarweb’s lower education citation rate noted by SEJ).
What the tutoring center should do:
- Win on local and branded discovery: become the named recommendation for “best SAT tutor near me” type prompts.
- Publish high-trust pages: tutor bios, methodology, success measurement approach (without inventing stats), pricing transparency, scheduling process.
- Build conversion-first landing pages because fewer citation opportunities can mean fewer clicks—each click must count.
- Target decision prompts: “online vs. in-person tutoring,” “group vs. 1:1,” “how to choose a tutor.” These are more comparison-like and may encourage citations.
Same budget, different physics. The hotel can often capture citation-driven discovery; the tutoring center may need to treat citations as a bonus, not the primary channel.
Action plan: how to improve your chances of being cited (without guessing)
Here’s a step-by-step plan that works whether your topic is high-citation (travel/retail) or low-citation (some education queries), because it starts with measurement and ends with execution.
Step 1: Map your AI query universe (not just keywords)
Classic SEO keyword lists don’t fully capture AI prompts. People ask longer, more specific, more contextual questions in chat.
Build an “AI question map” with buckets like:
- Comparisons: X vs. Y, best for families, cheapest, fastest
- Constraints: open now, cancellation policy, accessibility, parking
- Planning: itineraries, checklists, timelines
- Definitions: what is, how does it work
- Local: near me, neighborhoods, safety, commute
Then decide which buckets are closest to revenue.
Step 2: Identify what AI systems cite in your vertical
SEJ’s write-up is a reminder that citation sources differ by topic. Before you spend on content, answer:
- Are brand sites cited often in my niche, or are aggregators/publishers dominating?
- Is the dominant cited format UGC, reviews, product pages, or editorial explainers?
If you don’t know, treat that as a monitoring requirement, not a debate.
This is exactly why AYSA invests in visibility monitoring: you can’t manage what you can’t observe. Start here: AYSA Monitoring.
Step 3: Build “citable units” on your site
A “citable unit” is a section of a page that’s easy to lift into an answer and attribute. The more your content looks like a clean source, the easier it is for any system to reference.
Examples (SME-friendly):
- Clear pricing ranges and what’s included
- Policy sections in plain English (shipping, returns, cancellation)
- Step-by-step processes (“How our onboarding works”)
- Comparison tables (kept honest and accurate)
- Local specifics (parking instructions, entrance details, seasonal notes)
Don’t hide critical facts inside PDFs or images. Put them in structured HTML.
Step 4: Refresh what changes (and show that you did)
If your vertical is dynamic (travel, retail, sports), stale pages are a silent killer. Refresh doesn’t mean rewriting; it means updating the parts that drift.
Practical refresh habits:
- Add “Last updated” dates when legitimate
- Update hours, pricing, seasonal constraints
- Replace broken links and retired recommendations
- Add new FAQ entries from customer support tickets
Step 5: Strengthen the off-site sources that are actually cited
If UGC and reviews are prominent in your category (as Similarweb categorized travel citations in the SEJ summary), treat reputation operations as part of AI search.
That includes:
- Consistent business info everywhere (hours, policies)
- High-quality photos and descriptive responses
- Ethical review generation (ask, don’t incentivize dishonestly)
This is not a call to spam listings. It’s about making sure the most-cited ecosystems contain accurate, current information about you.
Step 6: Measure outcomes, not just activity
Track:
- Which queries lead to mentions/citations over time
- Which pages are used as landing pages from AI referrals (where visible)
- Conversion performance for those entry points
If you’re only tracking “rankings,” you’ll miss the shift. AI visibility isn’t a single SERP position.
Where AYSA fits: monitored, approved, executed AI search optimization
Most teams don’t fail because they lack ideas. They fail because execution is slow, fragmented, or unsafe—especially on small sites where the owner is also the operator.
AYSA is designed to make AI search optimization operational:
- Monitors your AI search visibility over time (so you can see shifts, not just snapshots). Learn more: AI search visibility.
- Prepares changes that improve clarity, coverage, and retrievability (content updates, internal linking, on-page improvements).
- Asks for approval before implementing—so you stay in control of brand, compliance, and risk.
- Executes accepted website changes—so the plan ships, not stalls.
If you’re evaluating tooling, start with the toolset overview here: AI SEO Tools. If you need to understand fit and cost quickly, see pricing. For ongoing editorials like this, visit the AYSA blog.
My point of view is direct: in 2026, the winners are not the teams with the cleverest AI prompts. They’re the teams that can measure the AI surface area in their vertical and execute improvements at the speed the market changes.
What to do next
- Classify your vertical: is your category dynamic/comparison-heavy (likely higher citation behavior) or evergreen/conceptual (likely lower citation behavior)?
- List 25 real customer questions from sales calls, support tickets, and reviews—not just keyword tools.
- Build 5–10 “citable units” (policies, comparisons, checklists, FAQs) on the pages closest to revenue.
- Refresh one critical page per week for 8 weeks. Track changes and outcomes.
- Audit your review/UGC ecosystem (accuracy, recency, consistency) if you’re in travel, local, retail, or any review-heavy niche.
- Set up monitoring so you can see whether visibility is expanding or shrinking as AI products change. Start at AYSA Monitoring.
Sources and further reading
- Search Engine Journal: ChatGPT Links Out Most On Travel Questions (coverage of Similarweb’s report and topic-level citation behavior)
- AYSA: AI search visibility
- AYSA: AI SEO tools
- AYSA: Monitoring
- AYSA: Blog
- AYSA: Pricing
Note on methodology: The SEJ coverage summarizes a gated Similarweb report and emphasizes that the figures are based on U.S. desktop usage and are estimates/extrapolations. Treat the numbers as directional indicators of behavior, not as absolute truths for every market, device, or time period. The strategic lesson—topic-dependent citation behavior—remains actionable.
Continue the AI search topic inside AYSA.
Use these pages to connect the article with AI SEO tools, AI visibility monitoring, AI Overviews and approved website execution.
Turn this topic into a website action plan.
Use these AYSA hubs to move from reading to technical fixes, AI visibility monitoring, research, glossary context and approval-first SEO execution.
