What 2.4 Million Citations Reveal About What Actually Works for AEO / AI Search
Prepared for: Brand Managers & CMOs, Australian Insurance Industry
Data Period: November 2025 – July 2026 (7 observed months)
Records Analyzed: 2,437,107 | Domains: 28,725
Platforms: ChatGPT standard search, Google
The Summary
We analysed 2.4 million insurance citations across ChatGPT and Google for our Answer Engine Optimization(AEO) research. Here’s what surprised us:
- Discount pages survive, guides die. Content with discount/save language appears 2x more often among domains that survive all 7 months vs domains that appear once and vanish.
- “Does X cover Y?” is an untapped goldmine. 10,120 citations go to coverage question pages but only 1.8% of content uses this format.
- Almost nobody publishes comparison content. Only 1.6% of cited content is “X vs Y” or comparison format, yet comparison sites get 19% of their citations from it.
- 57% of domains are one hit wonders. Cited once, never seen again. The content graveyard is vast.
- ChatGPT is learning who to trust. 67% of persistent domains improve their position over time Canstar went from 3.3 to 2.3, Finder from 3.7 to 2.5.
- Google visibility is not AI visibility. Some domains are dramatically over-indexed in ChatGPT relative to Google: Kiplinger is 28.3x, WSJ is 20.1x, Wikipedia is 5.6x.
- Transactional pages are weaker in ChatGPT. Quote/buy/apply language appears in 6.7% of ChatGPT citations vs 12.4% of Google citations.
- Homepages are fallback visibility, not winning visibility. Brand homepages are rising, but they rank around position 4.5–6.1 while specific answer pages often rank near position 2.
- Climate-risk content is underweight in AI search. Flood/storm/hail/bushfire/cyclone language appears in 3.2% of ChatGPT citations vs 5.1% of Google citations.
- Month to month AEO reporting needs normalization. ChatGPT share swings from 1.5% in Feb 2026 to 30.8% in Jan 2026 because the observed query set changes materially.
- Your URL slug is the one thing you can change today. 2 word slugs generate 3.1x more citations per unique slug than 1 word slugs.
Part 1: The Dataset
| Metric | Value |
|---|---|
| Total citation records | 2,437,107 |
| Unique domains cited | 28,725 |
| ChatGPT citations | 346,172 (14.20%) |
| Google citations | 2,090,935 (85.80%) |
| Observed months | Nov 2025, Dec 2025, Jan 2026, Feb 2026, Mar 2026, May 2026, Jul 2026 |
Each record contains: month, year, platform (chatgpt.standard or google.), domain, URL, position, title, description, and source. The platform_search_type field is the dividing line it lets us compare the same insurance topic across two fundamentally different search architectures.
Platform overview:
| Platform | Records | Share | Avg Position |
|---|---|---|---|
| google. | 2,090,935 | 85.80% | 5.77 |
| chatgpt.standard | 346,172 | 14.20% | 3.93 |
Of 28,725 unique domains: 4,924 appear only in ChatGPT, 18,279 appear only in Google, and 5,522 appear in both. The two platforms are not mirrors they are fundamentally different information ecosystems.
Part 2: Insights
Insight 1: The Discount Formula
Figure: Insight 1: The Discount Formula chart
Discount/save pages generate 10.1% of all ChatGPT citations. Domains that use discount language survive 2x more often than those that don’t.
This is the single most actionable finding in the dataset. The word “discount” or “save” in a page title or URL correlates with survival.
Discount Content Volume
| Metric | Value |
|---|---|
| Citations with discount/save language | 35,079 |
| Share of all ChatGPT citations | 10.1% |
| Top domain | Canstar (5,113 citations) |
| Second | Budget Direct (3,331) |
| Third | Finder (2,676) |
The Survival Connection
We compared content patterns between domains that appeared in all 7 months (survivors) and domains that appeared in only 1 month (one hit wonders):
| Content Pattern | Survivors | One Hits | Survival Multiplier |
|---|---|---|---|
| Discount/save language | 10.6% | 5.4% | 2.0x |
| Comparison language | 8.0% | 4.3% | 1.9x |
| FAQ format | 1.4% | 0.8% | 1.8x |
| How-to format | 4.3% | 2.9% | 1.5x |
| Guide format | 2.0% | 7.3% | 0.3x (dies) |
Discount language doesn’t just get cited it gets cited repeatedly. Canstar’s home-insurance-discounts page alone has 2,150 citations. The slug car-insurance-discounts-what-could-you-get has 1,471.
Top Discount Slugs
| Citations | Slug |
|---|---|
| 2,150 | home-insurance-discounts |
| 1,471 | car-insurance-discounts-what-could-you-get |
| 1,138 | multi-policy-discount |
| 1,021 | how-to-save-money-on-car-insurance |
| 855 | discounts |
| 800 | how-to-save-money-on-home-insurance |
| 752 | what-is-the-online-discount-for-car-insurance |
| 556 | insurance-discounts |
| 475 | save-on-home-contents-insurance |
| 450 | how-to-get-cheaper-car-insurance |
Action: Create a discount/save page for every product category. The slug formula is {category}-insurance-discounts. Canstar has proven this works there’s no reason your brand can’t do the same. See which of your pages AI search actually cites with Somantra Brand Engagement.
Insight 2: The Question Formula
Figure: Insight 2: The Question Formula chart
“Does X cover Y?” pages get 10,120 citations but only 1.8% of content uses this format. It’s the most undersupplied content type in AI search.
ChatGPT is a question-answering engine. When users ask “does car insurance cover flood damage?”, ChatGPT looks for a page that answers that exact question. Almost nobody is publishing those pages.
Coverage Question Pages
| Metric | Value |
|---|---|
| ”Does X cover Y” citations | 10,120 |
| Share of all ChatGPT citations | 2.9% |
| Share of content using this format | 1.8% |
| Supply gap | 1.6x demand exceeds supply |
Top Coverage Questions ChatGPT Answers
| Citations | Question |
|---|---|
| 544 | What does roadside assistance cover? |
| 372 | What pet insurance does and doesn’t cover |
| 299 | Does travel insurance cover covid-19 or other pandemics? |
| 290 | Does car insurance cover fire damage? |
| 278 | Does your pet insurance policy cover pre existing conditions? |
| 237 | What does dog insurance cover? |
| 199 | What does car insurance cover? |
| 183 | Does my home insurance include flood cover? |
| 159 | Does your car insurance cover theft of personal items? |
| 157 | Does landlord insurance cover short term rentals? |
The legalclarity.org Lesson
legalclarity.org is a legal information site that gets cited in ChatGPT with 76% question-format titles and an average position of 2.8 better than most major insurers. Their top pages answer specific questions: “What is landlord insurance in Australia and what does it cover?” (57 citations), “Why is my motorcycle insurance so high?” (52 citations).
This isn’t content farming it’s answering the exact questions people ask AI. The question format matches how users phrase queries to ChatGPT, which is why it ranks so well.
Action: Publish pages answering specific coverage questions. The slug formula is does-{category}-insurance-cover-{specific-risk}. Examples: does-car-insurance-cover-flood-damage, does-pet-insurance-cover-pre-existing-conditions, does-home-insurance-cover-termite-damage. Get a free audit of the coverage questions ChatGPT asks about your brand.
Insight 3: The Comparison Gap
Figure: Insight 3: The Comparison Gap chart
Only 1.6% of cited content is comparison or “X vs Y” format. Comparison sites get 19% of their citations from it. Insurers get 2.4%. The gap is massive.
When a consumer asks ChatGPT “which pet insurance is best?”, the AI looks for comparison content. Almost nobody is providing it except Canstar, Finder, and Choice, who dominate this format.
Comparison Content by Source Type
| Source Type | Comparison Citations | Share of Their Total |
|---|---|---|
| Comparison Sites | 11,708 | 19.2% |
| Insurers | 1,300 | 2.4% |
| Government | 303 | 0.5% |
| Forums/UGC | 299 | 2.2% |
Top Comparison Pages
| Citations | Page | Domain |
|---|---|---|
| 371 | Motorcycle Insurance Cost Comparison | totalmotorcycle.com |
| 215 | Landlord Insurance Comparison Australia | canstar.com.au |
| 176 | Landlord Insurance Comparison Australia | canstar.com.au |
| 173 | Home and Contents Insurance Comparison 2026 | canstar.com.au |
| 158 | Life insurance claims comparison tool | moneysmart.gov.au |
| 126 | Life Insurance Comparison Australia | canstar.com.au |
| 119 | Life Insurance Comparison Australia | canstar.com.au |
| 118 | Home Insurance Comparison Australia | finder.com.au |
| 116 | Life insurance comparison: Compare 20+ policies | finder.com.au |
Canstar owns the comparison format. They have dedicated comparison pages for every category: landlord, home, life, car, pet. Most insurers have zero comparison pages.
Action: Publish {category}-insurance-comparison pages. There’s almost no competition. Even a simple table comparing 3–5 products would fill a gap that ChatGPT is currently struggling to answer.
Insight 4: The FAQ Effect
Figure: Insight 4: The FAQ Effect chart
FAQ pages get 2.4% of all citations. Budget Direct dominates with 1,366 FAQ citations more than any other content type they publish.
FAQ pages are structured, answer specific questions, and match how ChatGPT retrieves information. They’re also one of the few content formats that insurers can own directly (unlike comparison content, which comparison sites dominate).
FAQ Content Performance
| Metric | Value |
|---|---|
| Total FAQ citations | 8,208 |
| Share of ChatGPT citations | 2.4% |
| Top domain | Budget Direct (1,366) |
| Second | 1cover.com.au (497) |
| Third | fastcover.com.au (487) |
Budget Direct’s FAQ Strategy
Budget Direct publishes FAQ pages for every product category. Their FAQ content gets cited more than their product pages, more than their guides, more than anything else. It’s their most reliable citation source.
Other insurers with strong FAQ presence: 1cover (497), fastcover (487), seniors.com.au (291), raa.com.au (291), medibank (272), nationalmotorcycleinsurance (255), petsecure (238).
Action: Build an FAQ section for every product. The slug formula is {category}-insurance-faq or just faqs. Budget Direct has proven this works at scale. Track how your FAQ pages perform in AI answers.
Insight 5: The Content Graveyard
Figure: Insight 5: The Content Graveyard chart
57% of domains are one hit wonders cited once, never seen again. The content that dies is predictable. The content that survives is predictable.
We tracked every domain across all 7 months. The results are stark.
Domain Survival Rates
| Survival Period | Domains | % of Total |
|---|---|---|
| All 7 months | 279 | 2.7% |
| 5–6 months | 438 | 4.2% |
| 3–4 months | 1,481 | 14.2% |
| 2 months | 2,278 | 21.8% |
| 1 month only | 5,970 | 57.2% |
More than half of all domains appear in exactly one month and vanish. The citation base churns constantly different queries each month surface different domains.
What Dies vs What Lives
The content patterns that predict survival:
| Pattern | Among Survivors (7 months) | Among One-Hits (1 month) | Verdict |
|---|---|---|---|
| Discount/save | 10.6% | 5.4% | Survives |
| Comparison | 8.0% | 4.3% | Survives |
| How-to | 4.3% | 2.9% | Survives |
| What Is | 4.0% | 2.0% | Survives |
| FAQ | 1.4% | 0.8% | Survives |
| Best Of | 5.2% | 4.3% | Neutral |
| Claims | 6.3% | 5.6% | Neutral |
| Cost/Price | 4.4% | 5.4% | Neutral |
| Guide | 2.0% | 7.3% | Dies |
Guides are 3.5x more common among dead content than surviving content. “How to” content survives, but “Complete guide to…” content dies. The difference: “how to save money on car insurance” (1,021 citations, survives) vs “Complete guide to motorcycle insurance in Australia” (87 citations, one hit).
The Top 10 Domains That Survive All 7 Months
| Domain | Citations | Type |
|---|---|---|
| canstar.com.au | 23,464 | Comparison |
| finder.com.au | 17,291 | Comparison |
| reddit.com | 13,416 | Forums |
| choice.com.au | 10,537 | Consumer |
| moneysmart.gov.au | 9,736 | Government |
| budgetdirect.com.au | 7,613 | Insurer |
| nationalcover.com.au | 7,540 | Affiliate |
| allianz.com.au | 7,434 | Insurer |
| aami.com.au | 7,357 | Insurer |
| nrma.com.au | 6,024 | Insurer |
Action: Stop publishing guides. Start publishing discount pages, comparison pages, and FAQ pages. The content format determines whether you survive or become a one hit wonder. Find out which of your pages survive in AI search — book a free brand audit.
Insight 6: The Homepage Effect
Figure: Insight 6: The Homepage Effect chart
ChatGPT Starts Citing More Brand Pages Directly
In July 2026, ChatGPT has reduced its dependency on 3rd party publishers for brands and citing brand homepages directly. The scale of the shift: 12.7x increase in brand homepages getting cited. In our research covering Australian General Insurance brands which we have been tracking for 7 months, in July homepages now make up 12.8% of all citations. In March, this was 1.4%
The interesting part: For as long as ChatGPT has cited sources, it needed a specific page to get content and then use it in the response summary, This used to help brands by guiding customers to conversion pages organically in AI Search results. Not anymore. ChatGPT has crawled the brands web pages and now REMEMBERS the brands product and service offerings. In many cases it is guiding the customer to the brand’s homepage but not specific product/service description pages. ChatGPT is building an internal associative map which maps the brand to product/service offerings. The citation simply points to the home page. The secondary order efrect: This now puts onus on the homepage to communicate product and service offerings. Brands that offer multiple product/services should pay attention to the customer experience to ensure the customer can easily navigate to the product/service page from the home page.
Homepage Citations by Brand
| Citations | Brand | What ChatGPT Shows |
|---|---|---|
| 487 | NRMA | nrma.com.au |
| 466 | Budget Direct | budgetdirect.com.au |
| 440 | AAMI | aami.com.au |
| 426 | Allianz | allianz.com.au |
| 389 | Youi | youi.com.au |
Action: Make sure your homepage is optimised for ChatGPT. Include structured data, clear product descriptions, and brand-authority signals. When ChatGPT can’t find a specific answer, your homepage is what it falls back to.
Insight 7: SEO Visibility Is Not AI Visibility
Figure: Insight 7: SEO Visibility Is Not AI Visibility chart
Winning Google does not automatically mean winning ChatGPT. ChatGPT has its own visibility pattern, and it rewards different publishers.
The biggest strategic mistake is treating AI search as a smaller version of Google. The overlap is real, but the ranking logic is different. Some domains with modest Google citation counts are disproportionately visible in ChatGPT.
AI Overindexed Domains
| Domain | ChatGPT Citations | Google Citations | AI/Google Ratio |
|---|---|---|---|
| www.kiplinger.com | 1,104 | 39 | 28.3x |
| www.wsj.com | 1,166 | 58 | 20.1x |
| en.wikipedia.org | 1,425 | 256 | 5.6x |
| hellosafe.com.au | 1,569 | 828 | 1.9x |
| www.investopedia.com | 1,207 | 992 | 1.2x |
AI Underindexed Domains
| Domain | ChatGPT Citations | Google Citations | AI/Google Ratio |
|---|---|---|---|
| www.youtube.com | 9 | 24,400 | 0.04% |
| www.facebook.com | 32 | 13,110 | 0.24% |
| www.mynrma.com.au | 214 | 11,075 | 1.9% |
| www.comparethemarket.com.au | 3,831 | 74,553 | 5.1% |
| www.qbe.com | 2,540 | 40,472 | 6.3% |
The pattern: ChatGPT imports trust from broad reference, finance, and consumer information publishers. It does not simply mirror Google , social platforms, video platforms, or quote heavy commercial pages.
Action: Track AI visibility separately from SEO visibility. Add an “AI/Google parity ratio” to reporting, but don’t treat parity as the only goal. A brand can be strong in Google and still weak in ChatGPT if the content does not answer research questions cleanly. Measure your AI share of voice with Somantra Brand Consideration.
Insight 8: The Transactional Penalty
Figure: Insight 8: The Transactional Penalty chart
ChatGPT is materially less likely than Google to cite quote, buy, claim, and policy-document pages.
The intent gap is measurable. Google surfaces more operational pages. ChatGPT surfaces more explanatory pages.
| Page Signal | ChatGPT Share | Google Share | Implication |
|---|---|---|---|
| Quote/buy/apply language | 6.7% | 12.4% | Purchase pages underperform in ChatGPT |
| Claims language | 8.1% | 17.1% | Operational pages are stronger in Google |
| PDS/policy wording | 1.7% | 2.4% | Document pages are not enough for AI |
| Reviews/ratings | 3.0% | 2.6% | Review context matters slightly more in ChatGPT |
This reinforces the broader finding: ChatGPT is a research engine, not a purchase engine. It does not ignore brands, but it prefers pages that explain, compare, define, and answer.
Action: Keep quote and claim pages for conversion and servicing, but don’t expect them to carry AI visibility. Build a research layer above them: explainers, coverage questions, discount pages, comparison pages, reviews, and FAQs.
Insight 9: The Specificity Premium
Figure: Insight 9: The Specificity Premium chart
ChatGPT is citing homepages more often, but specific answer pages rank better.
The homepage effect is real, especially in July 2026. But the position data shows a second truth: homepages keep a brand present when ChatGPT lacks a better URL. They rarely beat highly specific answer pages.
Homepage vs Specific Answer Pages
| Page | Citations | Avg Position |
|---|---|---|
| Budget Direct homepage | 466 | 4.80 |
| AAMI homepage | 440 | 4.51 |
| Allianz homepage | 426 | 4.54 |
| Youi homepage | 389 | 6.10 |
| Moneysmart: How to save money on car insurance | 568 | 1.96 |
| Moneysmart: Pet insurance | 1,306 | 2.23 |
| Canstar: Home insurance discounts | 2,041 | 2.08 |
| Qantas Insurance: Online discount for car insurance | 752 | 1.93 |
The pattern: A homepage citation is brand awareness. A specific answer page citation is authority. The best AI search strategy uses the homepage as a trust anchor, but wins positions with focused pages.
Action: Optimise the homepage for product clarity and entity trust, but do not stop there. Every major product needs specific pages for savings, coverage, claims, exclusions, comparisons, and FAQs.
Insight 10: The Public-Authority Explainer Advantage
Figure: Insight 10: The Public-Authority Explainer Advantage chart
Plain, trusted, non salesy explainers are some of the strongest pages in ChatGPT.
Moneysmart is the clearest example. It does not win because of aggressive SEO formatting. It wins because its pages are direct, authoritative, and written around consumer questions.
Moneysmart’s High-Position Explainer Pages
| Page | Citations | Avg Position |
|---|---|---|
| Pet insurance | 1,306 | 2.23 |
| Choosing car insurance | 894 | 2.15 |
| Contents insurance guide | 864 | 2.05 |
| Total and permanent disability insurance | 853 | 1.93 |
| How to save money on car insurance | 568 | 1.96 |
| Trauma insurance | 562 | 1.84 |
| How to make a home insurance claim | 468 | 2.53 |
| Storm, flood and fire insurance | 411 | 2.77 |
This is the “boring authority” pattern. ChatGPT rewards pages that answer the user’s question without forcing a commercial journey.
Action: Build insurer-owned explainers that copy the structure, not the authority source. Use neutral titles, direct definitions, answer first intros, clear exclusions, and links to product or quote pages after the explanation.
Insight 11: The Climate Risk Content Gap
Figure: Insight 11: The Climate-Risk Content Gap chart
Climate and property risk topics are more visible in Google than ChatGPT, which suggests an AI search content gap.
Weather risk terms are materially underrepresented in ChatGPT relative to Google.
| Topic Signal | ChatGPT Share | Google Share | Gap |
|---|---|---|---|
| Flood/storm/hail/bushfire/cyclone | 3.17% | 5.06% | 1.6x higher in Google |
| Solar/battery | 0.21% | 0.68% | 3.2x higher in Google |
| Pre existing conditions | 1.39% | 2.09% | 1.5x higher in Google |
| Short stay/Airbnb | 0.44% | 0.46% | Similar |
| EV/electric vehicle | 0.60% | 0.77% | Slightly higher in Google |
The opportunity is not generic “climate content.” The opportunity is specific risk-answer pages that match how consumers ask AI tools about cover.
High-Opportunity Page Formulas
| Formula | Example |
|---|---|
| does-home-insurance-cover-{risk} | does-home-insurance-cover-storm-damage |
| does-contents-insurance-cover-{asset} | does-contents-insurance-cover-solar-batteries |
| {risk}-insurance-australia | flood-insurance-australia |
| {category}-insurance-exclusions | home-insurance-exclusions |
| {category}-insurance-for-{scenario} | landlord-insurance-for-short-term-rentals |
Action: Prioritise climate and property risk pages where consumers need a yes/no coverage answer: flood, storm, hail, bushfire, cyclone, solar panels, home batteries, short term rentals, and high value contents.
Part 3: The Source Type Hierarchy
Who gets cited matters as much as what they publish. The hierarchy is clear and consistent.
| Source Type | Citations | Share | Who They Are |
|---|---|---|---|
| Other (niche/unknown) | 182,870 | 52.8% | Small sites, blogs, niche publishers |
| Comparison Sites | 63,814 | 18.4% | Canstar, Finder, Choice, Mozo, iSelect |
| Insurers (Brand) | 54,088 | 15.6% | AAMI, Allianz, Budget Direct, NRMA, etc. |
| Government | 24,448 | 7.1% | Moneysmart, Smartraveller, ASIC, AFCA |
| Forums / UGC | 13,416 | 3.9% | |
| Media | 7,528 | 2.2% | Forbes, ABC, news outlets |
What Each Source Type Gets Cited For
| Content Type | Insurers | Comparison Sites |
|---|---|---|
| Discount/save | 10.8% | 17.4% |
| Comparison | 2.4% | 19.2% |
| Best Of | 0.2% | 7.1% |
| Question format | 9.7% | 13.8% |
| Claims | 5.7% | 0.5% |
| Guide/how-to | 3.1% | 4.9% |
The pattern: Comparison sites dominate every content type except claims. Insurers own the claims space (Moneysmart aside), but they’re losing on discounts, comparisons, and best-of lists.
Action: You can’t out-Canstar Canstar. But you can own content types they don’t: FAQ pages, coverage questions, audience-specific pages. Play where comparison sites aren’t strong.
Part 4: The Category Landscape
All insurance categories move together. When one rises, they all rise. There is no cannibalization.
| Category | ChatGPT Citations | Share | Google Citations | Share |
|---|---|---|---|---|
| Travel | 48,088 | 13.9% | 213,918 | 10.2% |
| Pet | 43,708 | 12.6% | 238,461 | 11.4% |
| Car/Auto | 41,759 | 12.1% | 617,552 | 29.5% |
| Motorcycle | 31,472 | 9.1% | ~200,000* | ~9.6% |
| Home/Contents | 28,619 | 8.3% | 313,058 | 15.0% |
| Life/IP | 14,964 | 4.3% | 96,432 | 4.6% |
| Roadside Assist | 12,957 | 3.7% | ~50,000* | ~2.4% |
| Renters | 5,916 | 1.7% | 42,673 | 2.0% |
| Business | 3,219 | 0.9% | 29,832 | 1.4% |
| Health | 883 | 0.3% | 81,008 | 3.9% |
Approximate based on title keyword matching.
Category Share by Month
| Month | Car | Motorcycle | Pet | Travel | Home |
|---|---|---|---|---|---|
| Nov ‘25 | 18% | 15% | 25% | 27% | 15% |
| Dec ‘25 | 23% | 16% | 21% | 24% | 17% |
| Jan ‘26 | 22% | 16% | 24% | 24% | 14% |
| Feb ‘26 | 22% | 17% | 22% | 26% | 13% |
| Mar ‘26 | 21% | 16% | 25% | 25% | 14% |
| May ‘26 | 21% | 34% | 19% | 14% | 12% |
| Jul ‘26 | 22% | 13% | 26% | 25% | 15% |
May 2026 is the anomaly. Motorcycle jumped to 34% of category share (from ~16% baseline) while travel dropped to 14%. This is a query mix shift different questions were asked that month not a permanent trend. In July, everything returned to baseline.
The takeaway: Don’t overreact to single-month category swings. The relative share of categories is remarkably stable across 6 of 7 months.
Why Monthly AI Search Reporting Must Be Normalised
The observed ChatGPT sample is not evenly distributed by month. This means raw monthly citation counts should not be interpreted as pure brand momentum.
| Month | ChatGPT Citations | Google Citations | ChatGPT Share |
|---|---|---|---|
| Nov 2025 | 50,269 | 247,185 | 16.9% |
| Dec 2025 | 21,165 | 336,238 | 5.9% |
| Jan 2026 | 136,494 | 306,904 | 30.8% |
| Feb 2026 | 4,682 | 313,257 | 1.5% |
| Mar 2026 | 66,130 | 312,171 | 17.5% |
| May 2026 | 11,697 | 304,180 | 3.7% |
| Jul 2026 | 55,735 | 271,000 | 17.1% |
January and February are the clearest outliers. A brand that appears to collapse between January and February may simply be exposed to a different observed query set. For board reporting, compare brands within the same month, use share of ChatGPT rather than raw count alone, and track multi month averages.
Part 5: What ChatGPT Actually Tells Consumers
91% of ChatGPT citations are informational. People research through AI, they don’t buy.
| Intent Type | Citations | Share | Avg Position | Top-3 Rate |
|---|---|---|---|---|
| Informational | 314,970 | 91.0% | 3.9 | 50.7% |
| Navigational | 12,410 | 3.6% | 3.7 | 54.3% |
| Transactional | 12,120 | 3.5% | 4.2 | 46.0% |
| Comparison | 6,672 | 1.9% | 4.0 | 50.0% |
The surprise: Navigational queries (brand specific) get the best positions 54.3% land in the top 3. When someone asks “AAMI car insurance”, ChatGPT surfaces AAMI’s page high. Transactional queries (“buy car insurance”) perform worst avg position 4.2, only 46% top 3.
ChatGPT is a research engine, not a purchase engine. Your content strategy should focus on helping people research, not on pushing them to buy.
The Top 10 Questions ChatGPT Gets Asked
| Citations | Question Pattern |
|---|---|
| 1,021 | How to save money on car insurance |
| 955 | What are you covered for |
| 800 | How to save money on home insurance |
| 752 | What is the online discount for car insurance |
| 711 | Cost of pet insurance |
| 625 | What does motorcycle insurance cost |
| 596 | What does roadside assistance cover |
| 553 | What is roadside assistance |
| 548 | What does pet insurance cover |
| 515 | How to make a home insurance claim |
Part 6: URL Slug Intelligence What Gets Cited in AI Search
The URL slug is the single most controllable element in AI citation strategy. Unlike domain authority or backlinks, you can change a slug today. This section reveals exactly which slug formulas ChatGPT rewards.
Slug Structure Analysis
| URL Structure | ChatGPT Citations | Share | Google Citations | Share |
|---|---|---|---|---|
| category/slug (2 levels) | 133,031 | 38.4% | 854,975 | 40.9% |
| 4+ levels deep | 72,431 | 20.9% | 258,159 | 12.3% |
| single-slug | 67,941 | 19.6% | 486,602 | 23.3% |
| category/subcategory/slug (3 levels) | 62,392 | 18.0% | 456,877 | 21.9% |
| root (homepage) | 10,377 | 3.0% | 34,322 | 1.6% |
Key difference: ChatGPT cites deeper URL structures (4+ levels) at 1.7x the rate of Google (20.9% vs 12.3%). This suggests LLMs are better at navigating and indexing deeply nested content than Google’s crawler prioritizes.
File Extension & Parameter Patterns
| Pattern | ChatGPT % | Google % | Implication |
|---|---|---|---|
| Query parameters (utm_*) | 85.7% | 1.3% | ChatGPT indexes UTM-tagged URLs; Google largely ignores them |
| Anchor links (#) | 0.0% | 56.7% | Google indexes fragments; ChatGPT doesn’t |
| Clean URLs (no extension) | 2.0% | 25.6% | Google prefers clean URLs; ChatGPT doesn’t differentiate |
| .html extension | 11.6% | 15.6% | Comparable |
Critical finding: 97.9% of ChatGPT citations include utm_source=chatgpt.com ChatGPT is tagging its own citations for tracking. This means ChatGPT is actively monitoring which URLs it cites and attributing traffic back. Google results never include UTM parameters.
The Optimal Slug Length
| Slug Word Count | Total Citations | Avg Citations per Unique Slug | Unique Slugs |
|---|---|---|---|
| 1 word | 56,926 | 8.6 | 6,615 |
| 2 words | 70,998 | 26.4 | 2,689 |
| 3 words | 68,451 | 16.7 | 4,099 |
| 4 words | 37,321 | 9.6 | 3,877 |
| 5 words | 30,860 | 8.7 | 3,554 |
| 6 words | 20,181 | 6.8 | 2,958 |
| 7+ words | ~44,000 | ~5.5 | ~8,400 |
The sweet spot is 2 word slugs they generate 3.1x more citations per unique slug than 1 word slugs and 2.5x more than 4 word slugs. Examples: pet-insurance, car-insurance, motorcycle-insurance, roadside-assistance.
Winning Slug Formulas
| Formula | Example | Citations |
|---|---|---|
{category}-insurance | motorcycle-insurance | 21,230 |
best-{category}-insurance | best-motorcycle-insurance | 2,985 |
{category}-insurance-discounts | home-insurance-discounts | 2,637 |
what-does-{category}-insurance-cover | what-does-motorcycle-insurance-cost | 1,769 |
how-to-save-on-{category}-insurance | how-to-save-money-on-car-insurance | 1,996 |
comprehensive-{category}-insurance | comprehensive-motorbike-insurance | 1,700 |
cost-of-{category}-insurance | cost-of-pet-insurance | 1,380 |
{category}-insurance-for-{audience} | motorcycle-insurance-for-australian-riders | 626 |
Top 10 Most-Cited Slugs in ChatGPT
| Rank | Citations | Slug |
|---|---|---|
| 1 | 10,377 | root (homepage) |
| 2 | 7,249 | motorcycle-insurance |
| 3 | 6,349 | pet-insurance |
| 4 | 3,786 | roadside-assistance |
| 5 | 3,228 | cat-insurance |
| 6 | 3,125 | renters-insurance |
| 7 | 2,965 | car-insurance |
| 8 | 2,828 | travel-insurance |
| 9 | 2,487 | landlord-insurance |
| 10 | 2,432 | claims |
Domain Specific Slug Strategies
| Domain | Citations | Top Slug | Strategy |
|---|---|---|---|
| Canstar | 23,464 | home-insurance-discounts (2,041) | Savings/discounts focus |
| Finder | 17,291 | motorcycle-insurance (642) | Category landing pages |
| Choice | 10,537 | pet-insurance (1,031) | Product-specific + money-saving |
| Budget Direct | 7,613 | root (466) | Homepage authority |
| AAMI | 7,357 | comprehensive (506) | Coverage type pages |
| Allianz | 7,434 | adventure (716) | Niche activity focus |
| Moneysmart | 9,736 | pet-insurance (1,306) | Category education |
| PetCover | 3,194 | exotics-insurance (373) | Niche specialization |
Winning Slug Formulas by Category
Car Insurance:
| Formula | Citations | Example |
|---|---|---|
| Savings-focused | 4,503 (10.9%) | how-to-save-money-on-car-insurance |
| How-to guides | 2,907 (7.1%) | choosing-car-insurance |
| Claims-focused | 2,802 (6.8%) | no-claim-bonus-on-car-insurance |
| Exact-match | 3,177 (7.7%) | car-insurance |
Motorcycle Insurance:
| Formula | Citations | Example |
|---|---|---|
| Exact-match | 7,260 (23.1%) | motorcycle-insurance |
| Coverage type | 2,974 (9.4%) | comprehensive-motorbike-insurance |
| Price-focused | 1,253 (4.0%) | what-does-motorcycle-insurance-cost |
Pet Insurance:
| Formula | Citations | Example |
|---|---|---|
| Exact-match | 6,316 (14.5%) | pet-insurance |
| Price-focused | 2,036 (4.7%) | cost-of-pet-insurance |
| Claims-focused | 1,963 (4.5%) | make-a-claim |
Home Insurance:
| Formula | Citations | Example |
|---|---|---|
| Savings-focused | 3,779 (13.2%) | home-insurance-discounts |
| Exact-match | 4,196 (14.7%) | home-insurance |
| How-to guides | 2,164 (7.6%) | how-to-save-money-on-home-insurance |
Travel Insurance:
| Formula | Citations | Example |
|---|---|---|
| Best-of lists | 2,115 (4.3%) | best-travel-insurance |
| Niche activity | 1,140+ | adventure-travel-insurance, ski-travel-insurance |
| Audience specific | 958+ | seniors-travel-insurance, family-travel-insurance |
Action: Slug Optimization Checklist
- Use 2-word slugs for primary category pages (pet-insurance, car-insurance)
- Add savings language to home/car slugs (home-insurance-discounts)
- Use “best-{category}-insurance” for comparison content
- Use “how-to-save-on-{category}-insurance” for guides
- Target niche activities in travel (adventure-travel-insurance, ski-travel-insurance)
- Don’t fear 3-level depth ChatGPT indexes it better than Google
- Preserve citation tracking ChatGPT adds
utm_source=chatgpt.com; analytics and canonical rules should not strip this signal before measurement
Benchmark your slugs and AI citations against your competitors — start with a free brand audit.
Part 7: Supporting Data
Top 10 Domains ChatGPT
| Rank | Domain | Citations | Share |
|---|---|---|---|
| 1 | www.canstar.com.au | 23,464 | 6.78% |
| 2 | www.finder.com.au | 17,291 | 5.00% |
| 3 | www.reddit.com | 13,416 | 3.88% |
| 4 | www.choice.com.au | 10,537 | 3.04% |
| 5 | moneysmart.gov.au | 9,736 | 2.81% |
| 6 | www.budgetdirect.com.au | 7,613 | 2.20% |
| 7 | nationalcover.com.au | 7,540 | 2.18% |
| 8 | www.allianz.com.au | 7,434 | 2.15% |
| 9 | www.aami.com.au | 7,357 | 2.13% |
| 10 | www.nrma.com.au | 6,024 | 1.74% |
Top 10 Domains Google
| Rank | Domain | Citations | Share |
|---|---|---|---|
| 1 | www.canstar.com.au | 111,813 | 5.35% |
| 2 | www.budgetdirect.com.au | 78,417 | 3.75% |
| 3 | www.comparethemarket.com.au | 74,553 | 3.57% |
| 4 | nationalcover.com.au | 74,275 | 3.55% |
| 5 | www.allianz.com.au | 68,200 | 3.26% |
| 6 | www.finder.com.au | 63,486 | 3.04% |
| 7 | www.nrma.com.au | 49,500 | 2.37% |
| 8 | www.aami.com.au | 42,387 | 2.03% |
| 9 | www.qbe.com | 40,472 | 1.94% |
| 10 | www.racv.com.au | 40,030 | 1.91% |
Platform Parity Ratios
| Domain | Google Citations | ChatGPT Citations | Parity Ratio | Verdict |
|---|---|---|---|---|
| moneysmart.gov.au | 32,881 | 9,736 | 29.6% | BEST |
| www.finder.com.au | 63,486 | 17,291 | 27.2% | STRONG |
| www.canstar.com.au | 111,813 | 23,464 | 21.0% | STRONG |
| www.aami.com.au | 42,387 | 7,357 | 17.4% | MODERATE |
| www.suncorp.com.au | 37,501 | 5,447 | 14.5% | MODERATE |
| www.youi.com.au | 37,626 | 5,370 | 14.3% | MODERATE |
| www.nrma.com.au | 49,500 | 6,024 | 12.2% | LOW |
| www.allianz.com.au | 68,200 | 7,434 | 10.9% | LOW |
| www.budgetdirect.com.au | 78,417 | 7,613 | 9.7% | LOW |
| www.racv.com.au | 40,030 | 3,866 | 9.7% | LOW |
| www.comparethemarket.com.au | 74,553 | 3,831 | 5.1% | CRITICAL |
Citation Concentration Tiers
| Tier | Domain Count | Total Citations | Share |
|---|---|---|---|
| Tier 1 (5,000+) | 13 | 126,595 | 36.57% |
| Tier 2 (1,000–4,999) | 46 | 88,705 | 25.62% |
| Tier 3 (100–999) | 251 | 72,355 | 20.90% |
| Tier 4 (10–99) | 1,365 | 39,446 | 11.39% |
| Tier 5 (1–9) | 8,771 | 19,071 | 5.51% |
| Total | 10,446 | 346,172 | 100% |
Brand Trajectory ChatGPT Monthly Citations
| Brand | Nov ‘25 | Dec ‘25 | Jan ‘26 | Feb ‘26 | Mar ‘26 | May ‘26 | Jul ‘26 | Total |
|---|---|---|---|---|---|---|---|---|
| Budget Direct | 1,262 | 463 | 3,206 | 89 | 1,466 | 167 | 960 | 7,613 |
| Allianz | 1,253 | 393 | 2,280 | 84 | 1,230 | 277 | 1,917 | 7,434 |
| AAMI | 1,515 | 457 | 2,981 | 87 | 1,532 | 53 | 732 | 7,357 |
| NRMA | 370 | 375 | 2,326 | 76 | 1,482 | 407 | 988 | 6,024 |
| Suncorp | 1,042 | 361 | 2,474 | 73 | 896 | 171 | 430 | 5,447 |
| Youi | 544 | 322 | 2,491 | 92 | 1,129 | 182 | 610 | 5,370 |
| RACV | 420 | 237 | 1,397 | 54 | 890 | 193 | 675 | 3,866 |
| RACQ | 449 | 206 | 1,308 | 29 | 631 | 94 | 415 | 3,132 |
| APIA | 268 | 112 | 831 | 9 | 195 | 102 | 28 | 1,545 |
| Real Insurance | 224 | 112 | 801 | 24 | 452 | 67 | 15 | 1,695 |
| Stella | 47 | 34 | 229 | 7 | 112 | 11 | 11 | 451 |
| Choosi | 77 | 60 | 394 | 9 | 126 | 17 | 11 | 694 |
Top 10 Growing Domains ChatGPT (Jan → Jul 2026)
| Domain | Jan ‘26 | Jul ‘26 | Change |
|---|---|---|---|
| www.afca.org.au | 22 | 282 | +1,182% |
| www.coles.com.au | 167 | 661 | +296% |
| www.csc.gov.au | 14 | 55 | +293% |
| www.bankofmelbourne.com.au | 25 | 86 | +244% |
| www.petbarn.com.au | 34 | 87 | +156% |
| www.choice.com.au | 2,114 | 5,376 | +154% |
| business.gov.au | 27 | 68 | +152% |
| www.tac.vic.gov.au | 11 | 27 | +145% |
| www.studyaustralia.gov.au | 32 | 76 | +138% |
| www.floodsmart.gov | 36 | 85 | +136% |
Top 10 Declining Domains ChatGPT (Jan → Jul 2026)
| Domain | Jan ‘26 | Jul ‘26 | Change |
|---|---|---|---|
| www.insurancewatch.com.au | 178 | 12 | -93% |
| www.rspcapetinsurance.org.au | 927 | 61 | -93% |
| mymcaa.com.au | 229 | 15 | -93% |
| www.aant.com.au | 205 | 13 | -94% |
| www.freely.me | 358 | 21 | -94% |
| au.trustpilot.com | 242 | 14 | -94% |
| insurance.everyday.com.au | 842 | 45 | -95% |
| en.wikipedia.org | 1,014 | 50 | -95% |
| www.stellainsurance.com.au | 229 | 11 | -95% |
| financialrights.org.au | 565 | 26 | -95% |
Source Diversity Top Domains by Unique Query Breadth
| Domain | Unique Queries | Total Citations | Citations per Query |
|---|---|---|---|
| www.reddit.com | 2,969 | 13,416 | 4.5 |
| www.finder.com.au | 934 | 17,291 | 18.5 |
| www.canstar.com.au | 719 | 23,464 | 32.6 |
| www.comparethemarket.com.au | 441 | 3,831 | 8.7 |
| www.aami.com.au | 421 | 7,357 | 17.5 |
| www.forbes.com | 406 | 4,125 | 10.2 |
| www.choice.com.au | 393 | 10,537 | 26.8 |
| www.budgetdirect.com.au | 379 | 7,613 | 20.1 |
| www.suncorp.com.au | 343 | 5,447 | 15.9 |
Government Domain Citations ChatGPT
| Domain | Citations | Description |
|---|---|---|
| moneysmart.gov.au | 9,736 | Financial literacy portal |
| www.nsw.gov.au | 825 | NSW government |
| www.afca.org.au | 334 | Financial complaints authority |
| www.asic.gov.au | 198 | Corporate regulator |
| www.privatehealth.gov.au | 164 | Private health info |
| www.studyaustralia.gov.au | 141 | Study abroad |
| www.floodsmart.gov | 143 | Flood insurance |
| business.gov.au | 127 | Business portal |
| www.servicesaustralia.gov.au | 109 | Government services |
| www.csc.gov.au | 88 | Superannuation |
| www.tac.vic.gov.au | 55 | Transport accident |
| Total (tracked subset) | 11,920 | 3.4% of ChatGPT |
| All .gov.au domains | ~24,300 | 7.04% of ChatGPT |
Validated KPI Baseline
| KPI | Value |
|---|---|
| Total citations | 2,437,107 |
| ChatGPT citations | 346,172 (14.20% of total) |
| Google citations | 2,090,935 (85.80% of total) |
| ChatGPT : Google ratio | 16.56% |
| Domains in both | 5,522 |
| ChatGPT avg position | 3.93 |
| Google avg position | 5.77 |
| All .gov.au domain share (ChatGPT) | 7.04% |
| New domains in Jul 2026 | 685 |
| 7 month survivors | 279 (2.7%) |
| 1 month wonders | 5,970 (57.2%) |
Formulas and Metric Definitions
Standardized Formulas
1. Platform Share (%)
platform_share = (platform_citations / total_citations) × 100
Example: ChatGPT share = 346,172 / 2,437,107 × 100 = 14.20%
2. Platform Parity Ratio (%)
parity_ratio = (brand_chatgpt_citations / brand_google_citations) × 100
Example: AAMI parity = 7,357 / 42,387 × 100 = 17.36%
3. AI Share of Month (%)
ai_share_of_month = (brand_chatgpt_citations_in_month / total_chatgpt_citations_in_month) × 100
Use this alongside raw citations so monthly query-volume changes do not get mistaken for brand momentum.
4. Period Change (%)
change_pct = ((value_end - value_start) / value_start) × 100
Example: AAMI Jan→Jul = ((732 - 2,981) / 2,981) × 100 = -75.4%. Use cautiously when the monthly sample size changes materially.
5. Tier Concentration Share (%)
tier_share = (tier_citations / total_chatgpt_citations) × 100
Example: Tier 1 share = 126,595 / 346,172 × 100 = 36.57%
6. Content Survival Rate
survival_rate = (pages_cited_in_3plus_months / total_brand_pages) × 100
7. Discount Page Coverage
discount_coverage = discount_pages_published / product_categories_offered
Target: 1.0 (one discount page per product)
8. Question Coverage
question_coverage = coverage_question_pages / top_20_category_questions
Target: >0.50 (answer at least 10 of the top 20 questions)
9. Research Content Mix
research_content_mix = (explainer + FAQ + comparison + discount + coverage_question citations) / brand_chatgpt_citations × 100
Target: >60% of brand AI citations from research-led content rather than quote, claim, and policy-document pages.
Frequently asked questions
What type of content does well for citations in AEO tactics for insurance related content? +
Discount and save pages lead, generating 10.1% of all ChatGPT insurance citations, and domains that use discount language survive in AI results twice as often as those that don't. Coverage-question pages ("Does X cover Y?"), comparison pages, and FAQs are the other consistently cited formats.
In AEO, does ranking well on Google mean ranking well on ChatGPT? +
No. Of 28,725 domains analysed, only 5,522 appeared on both platforms. Some publishers are dramatically over-indexed in ChatGPT relative to Google , Kiplinger by 28.3x and the Wall Street Journal by 20.1x so AI visibility must be measured separately from SEO visibility.
What is the best URL slug length for AI search citations? +
Two-word slugs are the sweet spot. They generate 3.1x more citations per unique slug than one-word slugs and about 2.5x more than four-word slugs. Examples that perform well include pet-insurance, car-insurance and roadside-assistance.
Do 'Does X cover Y' coverage-question pages work in AEO for insurance related content? +
Yes, coverage-question pages attract 10,120 ChatGPT citations, yet only 1.8% of content uses the format, making it the most undersupplied content type in AI search. ChatGPT is a question-answering engine and rewards pages that answer the exact question a user asks.
How many insurance domains get cited only once in AI search? +
57% of domains are 'one-hit wonders' cited in a single month and never seen again. Only 2.7% of domains survive across all seven observed months, and survival correlates strongly with discount, comparison, how-to and FAQ content.
Are comparison pages worth publishing for AEO? +
Yes. Comparison sites earn 19% of their citations from 'X vs Y' content, while insurers earn just 2.4% from it. Almost nobody outside Canstar, Finder and Choice publishes comparison pages, leaving the format wide open.
Why do brand homepages get cited in ChatGPT? +
When ChatGPT cannot find a specific answer page for a query, it falls back to the brand homepage. Homepages account for about 3% of ChatGPT citations overall, and that rose to 12.8% in July 2026 as ChatGPT began citing brands directly.
Do transactional 'quote' and 'buy' pages perform in AI search? +
They underperform. Quote, buy and apply language appears in only 6.7% of ChatGPT citations versus 12.4% of Google citations. ChatGPT behaves as a research engine, so explanatory, comparison and FAQ pages win visibility while transactional pages are better kept for conversion.
About the author
Yajat S
Yajat S is a researcher at Somantra, where he analyses how brands surface in AI search across ChatGPT and Google. He authored Somantra's Australian insurance citation study covering 2.4 million AI search citations, focused on the content formats and URL patterns that get cited by answer engines.
