Foglift
The Foglift AI Citation Map: Four Authority Patterns in Q2 2026
Across 25 industries, 5 production AI search engines, and 75 brand-neutral buyer-intent prompts, every industry falls into one of four citation patterns. Which pattern your category is in determines your strategy, and whether you're the leader or a challenger changes it again.
Last updated Data window 2026-05-18 (375 buyer-intent responses across 75 prompts × 5 engines)Download CSV
Methodology
75 brand-neutral buyer-intent prompts across 25 verticals (10 tech SaaS, 5 consumer services, 10 CPG/retail) were sent to ChatGPT (GPT-5 + web_search), Claude (claude-sonnet-4-6 + web_search), Gemini (gemini-2.5-flash + grounding), Google AI Overview (gemini-2.5-flash + grounding, AIO mode), and Perplexity (sonar) on 2026-05-18. Cited URLs were extracted, normalized to root domains, and classified by structural type (vendor first-party, niche publisher hub, listicle content farm, media business press, marketplace, institutional, video, community, social, and others). Citation share is the fraction of the 15 responses per vertical (3 prompts × 5 engines) in which each domain was cited at least once. Shares are non-exclusive: multiple domains co-occur in the same response, so shares across the top of a vertical regularly sum to more than 100%.
The framework, in one sentence
When we asked five production AI search engines (ChatGPT, Claude, Gemini, Google AI Overview, and Perplexity) the same 75 buyer-intent questions across 25 industries, we expected one distribution. A vendor edging out a publisher by some consistent margin. That isn’t what the data shows. It shows four distinct competitive structures, and which one your category is in determines your strategy.
We call this The Foglift AI Citation Map.
How we measured “citation authority”
For every prompt, we recorded each engine’s cited URLs and normalized them to root domains. For each vertical (3 prompts times 5 engines = 15 responses), we counted the fraction of responses where each domain was cited at least once. We call this the domain’s “share” of the vertical. The “leader” is the domain with the highest share. “Lead over #2” tells you whether that leader dominates or is in a tight race.
Two things about share matter to interpret the tables. First, share is non-exclusive. Each AI response cites multiple URLs (3 to 10 per response depending on the engine), so multiple domains co-occur in the same response. A vertical’s top three domains can each have shares of 60%, 47%, and 47%. They aren’t competing for the same slot. They’re all appearing in overlapping responses, and shares across the top of a vertical regularly sum to more than 100%. Second, “lead over #2” is point difference, not stolen share. When salesforce.com leads techradar.com by 7 points in CRM (47% vs 40%), it means salesforce.com appears in 7 of 15 responses where techradar.com appears in 6. The 7-point lead is the margin in “how often each is cited,” not “how much citation share salesforce stole from techradar.”
We then classified every cited domain by structural type: vendor first-party, niche publisher hub, listicle content farm, media business press, and so on. Lumping a vendor’s own page in with an editorial publisher’s “best-of” article was the central methodological error of the first version of this analysis. The data only becomes intelligible once those types are kept separate. The full numeric reference tables (top 25 cited domains, top 10 per engine, top 5 per vertical, citation-type distribution, URL-depth distribution, cross-engine Jaccard matrix, long-tail concentration curve) live at /research/ai-search-citation-benchmark-2026-q2.
The Citation Map
The 25 verticals fall onto a 2×2 quadrant. The horizontal axis is who occupies the #1 citation slot. The vertical axis is how dominant the leader is.
Loading citation map…
The four named patterns are Contested Lead, Editorial Capture, SEO Capture, and Open Frontier. A fifth cell, Vendor Hegemony, is empty in our Q2 2026 sample. Not a single vertical has a vendor’s own first-party domain crossing the dominance threshold. The biggest vendor lead in the entire dataset is uptrace.dev at 47% share with a 20-point lead in dev tools, which still doesn’t clear the 50%/15pp bar.
The Vendor Hegemony quadrant may be structurally impossible. AI search engines are designed to surface options, not single answers. The response format itself names multiple vendors, compares them, and cites alternatives. A vendor crossing into dominance would require the engine to actively suppress competitors, which runs against the design. Buyers asking “best X” expect a shortlist, not a sole recommendation, and the engines train and tune toward that expectation. Vendor Hegemony may not be a quadrant any vertical ever reaches, regardless of how much content investment any single vendor makes. The citation moat in AI search has a ceiling, and the ceiling is Contested Lead.
Every vertical, ranked
Each vertical’s #1 cited domain, ordered by share within each pattern. The colored bands group verticals by their citation pattern. Use the filter buttons to isolate one pattern at a time.
Loading leaderboard…
Contested Lead
Eight verticals have a vendor’s own first-party domain as the #1 cited source, but the lead over #2 is 7 to 20 percentage points. The citation graph is contested but vendor-led.
| Vertical | #1 vendor | #1 share | Lead over #2 |
|---|---|---|---|
| Mental health apps | growtherapy.com | 60% (9/15) | 13pp |
| Design tools | uxpilot.ai | 53% (8/15) | 13pp |
| HR | gusto.com | 53% (8/15) | 7pp |
| Dev tools | uptrace.dev | 47% (7/15) | 20pp |
| Analytics | pendo.io | 47% (7/15) | 7pp |
| CRM | salesforce.com | 47% (7/15) | 7pp |
| Real estate | libertyhomeguard.com | 47% (7/15) | 13pp |
| Sales engagement | forecastio.ai | 40% (6/15) | 7pp |
CRM is the cleanest example. Salesforce.com is the top-cited domain in CRM responses, but only by 7 points over techradar.com. Folk.app, an upstart, holds the #3 spot at 27%. A challenger with strong first-party content can climb into the top five; an incumbent has to keep producing content or risk losing the lead.
If you’re the leader
Your moat is fragile, and this is as good as it gets. A 7 to 13 point lead disappears in two quarters of content drought. There is no Vendor Hegemony quadrant to climb into. The AI engines are structurally designed to keep a category looking like a shortlist, not a single winner, so the gap to #2 will not widen indefinitely no matter how much you invest. Keep producing deep first-party content (blog posts, comparison pages, integrations documentation). Get media coverage as reinforcement. The work in Contested Lead is to defend the position, not to break out of it.
If you’re trying to break in
The leader is reachable. The play is depth in a sub-niche. Be the deepest first-party authority on one specific buyer use case before generalizing. Folk.app’s path into CRM citations was a focused founder-led-startup narrative, not a Salesforce-killer pitch. Pick your slice, win it, then expand. The 20-point gap between the #1 and #3 spots in most Contested Lead verticals is closable on a 12-month horizon if you commit to a content investment that matches the leader’s depth.
Editorial Capture
Seven verticals have a niche publisher hub (or institutional source) as the #1 cited domain, with shares of 53 to 80 percent and double-digit leads. The brand’s own domain is essentially invisible.
| Vertical | #1 source | Type | #1 share | Lead over #2 |
|---|---|---|---|---|
| Telemedicine | healthline.com | niche publisher hub | 80% (12/15) | 27pp |
| Sustainable apparel | thegoodtrade.com | niche publisher hub | 73% (11/15) | 47pp |
| Travel booking | upgradedpoints.com | niche publisher hub | 73% (11/15) | 33pp |
| Mattresses | ncoa.org | institutional | 73% (11/15) | 13pp |
| Outdoor gear | cleverhiker.com | niche publisher hub | 67% (10/15) | 13pp |
| Athletic shoes | runrepeat.com | niche publisher hub | 53% (8/15) | 7pp |
| Pet care | petmd.com | niche publisher hub | 53% (8/15) | 0pp (3-way tie) |
The shocking part of Editorial Capture isn’t who’s at the top. It’s who isn’t. In athletic shoes, Hoka, Brooks, Asics, Nike, and New Balance all have zero AI citations in their own category. In mattresses, Saatva, Purple, Tempur-Pedic, and Casper all have zero. In outdoor gear, Patagonia, North Face, and Arc’teryx are absent. The publisher hubs have effectively been delegated as the editorial authority for these verticals. AI engines defer to them, and the brand websites are out of the citation graph.
If you’re the publisher
Protect what you have. Keep methodology pages prominent. Preserve named-reviewer bylines (AI engines learn to trust editorial integrity signals over time). Expand vertically. Own the adjacent category before someone else does.
If you’re trying to break in (as a brand)
Stop trying to dominate citation share directly. You can’t out-publish Sleep Foundation. The play is to make the publisher cite you. Provide test samples, original data, expert interviews, methodology documentation. Co-citation analysis on our data shows these publishers travel in packs: when Sleep Foundation cites you, Mattress Clarity and Naplab tend to cite you too. The cluster moves as a unit. Getting into the cluster is the goal, not displacing it.
SEO Capture
Three verticals (all in tech SaaS) have an SEO-optimized listicle or affiliate site as the #1 cited source. These sites outrank both vendors and traditional media through years of head-start SEO content, an affiliate revenue model, and aggressive buyer-intent keyword targeting.
| Vertical | #1 source | #1 share | Lead over #2 |
|---|---|---|---|
| Email marketing | emailvendorselection.com | 60% (9/15) | 7pp |
| Project management | thedigitalprojectmanager.com | 53% (8/15) | 13pp |
| Marketing automation | insiderone.com | 53% (8/15) | 13pp |
The leader here isn’t a vendor you can out-product, and it isn’t a credible editorial publisher you can pitch to. It’s an SEO operation.
If you’re the listicle
Your moat is durable. Keep the SEO machine running. The buyer-intent keywords still convert, and the affiliate economics still pay back the content investment.
If you’re trying to break in (as a vendor)
Two routes, both worth taking. First, get listed on the dominant listicle sites. These operations rank vendors that pay, that have strong feature parity in the category, or that submit detailed product specs (the listicles are content businesses, and good vendor data is their input). Second, own the #2 share via vendor-first-party content depth. In email marketing, knock.app holds 47% share at #2. In project management, wrike.com is at 40%. Being the cleanly-cited vendor that the listicles also feature gives you two routes to citation simultaneously. Don’t try to outrank the listicle for “best X.” Try to be ranked #1 by it.
Open Frontier
Five verticals have no clear leader. The #1 cited domain holds only 27 to 33 percent share, with multiple domains tied or near-tied. AI search authority in these categories is genuinely up for grabs.
| Vertical | #1 source | Type | #1 share | Lead over #2 |
|---|---|---|---|---|
| AI tools / GEO | onely.com | vendor first-party | 27% (4/15) | 0pp (3-way tie) |
| Online learning | youtube.com | video | 33% (5/15) | 0pp (3-way tie) |
| Supplements | fortune.com | media business press | 33% (5/15) | 0pp (tie) |
| Skincare beauty | ulta.com | marketplace | 33% (5/15) | 0pp (tie) |
| Specialty coffee | menehunecoffee.com | personal blog | 53% (8/15) | 13pp |
The signature of Open Frontier: leaders at less than 35% share, three-way ties at the top, or in one case a personal blog winning (menehunecoffee.com in specialty coffee). No real institutional authority has been established.
Foglift’s own category, AI tools / GEO, is in Open Frontier. The leader (onely.com) holds 27%, tied with tryprofound.com and darkroomagency.com. None of the named “AI search visibility tools” in trade press holds a dominant position. Foglift itself has zero citations in this dataset. The race is wide open, but we’re not yet in it.
If you have any citation share in Open Frontier (even 5%)
Invest aggressively now. Citation graphs lock in fast once one player crosses roughly 50% share. Engines start citing that player as a default, the network effect compounds, and Open Frontier becomes Contested Lead. The window is open for one to three quarters, not years.
If you’re trying to break in
Enter via the lowest-cost authority signal. Long-form first-party content. Third-party reviewer outreach. Original research (like this benchmark). Open Frontier rewards commitment more than budget. The winner is often the one who commits first, not the one who spends most.
What this means for “omnichannel vs. dominate”
The natural framing of citation strategy, and the one the first draft of this benchmark fell into, is that AI search authority is a single thing, and the question is whether to dominate it or to diversify across channels.
That framing doesn’t survive the four-pattern view.
In Contested Lead, dominate-the-citation-space works, but margins are thin and challengers can climb. The play is vendor first-party content backed by media coverage. In Editorial Capture, dominate-the-citation- space does not work. Your brand will not out-cite Sleep Foundation. Get into the publisher cluster. In SEO Capture, dominate-the-citation- space is a multi-year project. Get listed by the incumbent and own #2 share. In Open Frontier, dominate-the-citation-space is actively available. The first vendor to invest seriously locks in the #1 slot.
The “omnichannel” prescription is what you get if you average across all four patterns. What omnichannel really means is “use the right channel mix for the pattern you’re in,” and the channel mix varies sharply by quadrant.
There’s also the empty quadrant. Vendor Hegemony, the absence of any vertical with a vendor in truly dominant citation position, isn’t random. AI engines are designed to give buyers options, and the response format keeps a category looking like a shortlist by default. Contested Lead is the ceiling, not the floor on the way to something better. The brand-citation game in AI search isn’t “how do I become Wikipedia in my category.” It’s “how do I stay clearly #1 in a category where the engine itself won’t let any single brand run away with it.” Different game, different planning horizon, different investment math.
Caveats
This is a Q2 2026 snapshot. Live AI search responses drift over time, and specific share numbers will move quarter over quarter. The four patterns are likely durable. The specific pattern-to-vertical assignments may shift. The 25 verticals are buyer-intent-heavy B2B SaaS, consumer services, and CPG/retail. Long-tail B2B (industrial, scientific, legal niche) almost certainly looks different. Brand-anchored prompts (“Hoka vs Brooks for marathon training”) were deliberately excluded because they pre-determine which brands the engine cites. A brand-anchored follow-up arm is queued for a future quarter.
Data
The full Q2 2026 reference tables (overall top 25 cited domains, top 10 per engine, top 5 per vertical, citation-type distribution by engine, URL-depth distribution, cross-engine Jaccard matrix, long-tail concentration curve) live at /research/ai-search-citation-benchmark-2026-q2. Aggregated CSV (one row per domain × engine × vertical × category): /research/citation-benchmark-2026-q2.csv. This page is the analysis of those numbers; that page is the numbers.
Want to know which pattern your category is in? Run a free Foglift scan. The same five engines, the same grounded production models, applied to your URL and your tracked prompts.
Frequently Asked Questions
What are the four AI Citation Patterns?
Contested Lead, Editorial Capture, SEO Capture, and Open Frontier. The pattern is determined by which type of source occupies the #1 citation slot in your industry and how dominant that slot is. A vendor first-party leading by a thin margin is Contested Lead. A niche publisher hub dominating with double-digit leads is Editorial Capture. An SEO listicle farm dominating is SEO Capture. No clear leader is Open Frontier. A fifth quadrant, Vendor Hegemony, is empty by design.
Why is the Vendor Hegemony quadrant empty?
AI search engines are designed to surface options, not single answers. The response format names multiple vendors, compares them, and cites alternatives. A vendor crossing into truly dominant citation share would require the engine to actively suppress competitors, which runs against the design. Empirically, no vertical in the 25 industries we measured has crossed the 50% share plus 15-point lead threshold. The biggest vendor lead in the dataset is uptrace.dev at 47% share with a 20-point lead in dev tools, which still doesn't clear the bar.
How do I find out which pattern my industry is in?
Run the prompts your buyers would actually type through ChatGPT, Claude, Gemini, Google AI Overview, and Perplexity. Record the cited URLs and compute each domain's share of the responses. If your own brand is the leader at 40 to 60% share with a tight margin, you're in Contested Lead. If a publisher hub like Sleep Foundation or Healthline leads at 50 to 80% share, you're in Editorial Capture. If an SEO listicle site leads at 50 to 60% share, you're in SEO Capture. If no domain crosses 35%, you're in Open Frontier. A Foglift scan automates this for any URL you point it at.
What's the difference between Editorial Capture and SEO Capture?
Both have non-vendor third-party dominance, but the third party is structurally different. Editorial Capture means a category-specialist editorial publisher with named reviewers and explicit methodology (Sleep Foundation for mattresses, Outdoor Gear Lab for outdoor, Healthline for health) owns the citation graph. SEO Capture means an SEO-driven affiliate or listicle site (emailvendorselection.com, thedigitalprojectmanager.com, insiderone.com) owns it. The strategic response differs sharply: pitch yourself to the publisher hub in Editorial Capture; get listed on the listicle in SEO Capture.
Does the four-pattern framework apply outside the 25 verticals you measured?
The four patterns are likely durable structural archetypes. The specific assignment of any single vertical may shift quarter over quarter as AI engines retrain. The 25 verticals in our dataset are buyer-intent-heavy B2B SaaS, consumer services, and CPG / retail. Long-tail B2B (industrial, scientific, legal niche) and low-buyer-intent consumer categories are not covered. The framework's predictions generalize; the specific examples may not.
How often will Foglift refresh the Citation Map?
Once per quarter using the same frozen 75-prompt set. The Q3 2026 refresh is the next scheduled run. Cross-quarter comparability is the explicit point of the discipline rule: same prompts, same engines, same extraction logic. Trend data is what makes a benchmark useful over time, and the framework only earns durability through that discipline.