AI Overviews Citation Share — Canadian Verticals (2026)

The headline finding

AI Overviews and AI search assistants now appear on roughly 35–55% of commercial-intent Canadian queries (varies by vertical). Citation share within those AI Overviews concentrates heavily — top 3 cited sources get ~60–75% of citations across our 30-query probe baskets per vertical. The patterns are predictable: long-form, well-structured, named-author, schema-rich content earns citations; thin programmatic content, generic 'top 10' listicles, and aggressive promotional content do not. Canadian small businesses systematically underperform their international competitors in AI citation share, mostly because the international competitors are publishing more substantive content per topic.

Methodology and hedge notes

Probe basket methodology: 30 commercial-intent queries per vertical, sampled monthly across Google AI Overviews, ChatGPT, Perplexity, and Claude.ai over the period December 2025 – March 2026. Citations counted by source URL (deduplicated within a single response, counted across responses). Verticals analyzed: legal services, dental, home services, ecommerce DTC, B2B SaaS, financial services, and education. Sample size per vertical: 30 queries × 4 LLMs × 4 monthly samples = 480 probe-runs per vertical.

Methodology and hedge notes

Each vertical was probed monthly with the same 30-query basket. Queries were chosen to be commercial-intent (queries a buyer would plausibly run while researching a purchase), Canadian-context (often including 'Canada', 'Toronto', etc., or implicit by query phrasing), and stable over time (we avoided trending or news-cycle queries).

Citations were counted as appearance of a brand domain in the AI response, regardless of position. We did not weight by position. We deduplicated within a single response (a domain cited 3 times in one answer counted as 1 citation).

Hedge: AI search is volatile. Citation patterns can shift week to week as model retraining and indexing happen. Our monthly sampling smooths some of this, but month-to-month variance for any individual brand can be ±25% even with stable underlying content.

Hedge: We are a Canadian SEO agency. Our audit set and observation network are heavier on certain verticals (legal, dental, B2B SaaS, ecommerce DTC) and lighter on others (industrial B2B, government, education).

AI Overview appearance rate by vertical

Not every commercial-intent query triggers an AI Overview. Appearance rates from our probe baskets:

Citation concentration: who wins

AI Overviews don't cite 10 sources per answer. Most cite 2–4. The top sources for any given query basket therefore concentrate share heavily.

• Top 3 sources earn 60–75% of citations across our baskets

  Long-tail share is small. AI search is winner-take-most.
• Top source typically earns 25–40% of all citations

  Single-source dominance is common in vertical-specific baskets.
• Concentration is higher than traditional SERP

  On a traditional SERP, attention is split across 10 results plus ads. On an AI Overview, attention is concentrated in the cited 2–4 sources. The CTR penalty for not being cited is therefore steeper than for not ranking #1.
• Brand-name dominance helps but isn't required

  Big brands often win when present, but small specialist sites with deep content can outcompete big brands on long-tail or specialist queries. Specialization beats scale on niche topics.

Content patterns that earn citations

• Long-form content with structured answers

  Pages 1,500+ words with clear headings, TL;DR sections, FAQ structure, and direct factual answers in the first paragraph of relevant sections. Format mirrors how AI assistants synthesize.
• Named author identity with credentials

  Pages attributed to a named author with a real About page, credentials, and consistent byline accumulate citation rate over time. Generic 'team' authorship underperforms.
• Schema density (FAQPage, Article, BreadcrumbList)

  Schema-rich pages cite more often than equivalent content without schema. Causation unclear; likely a quality proxy.
• Citing primary sources within content

  Pages that cite primary sources (Statistics Canada, government regulators, peer-reviewed research) cite more often than pages that don't. AI assistants seem to favour content that itself cites well.
• Specific numerical claims with disclosure

  'Median Canadian SEO retainer is C$4,200/month, based on our audit-set of 80 businesses' cites better than 'SEO costs vary widely'. Specificity + methodology earns citations.
• Direct factual structure

  First paragraph of each section answers the section question directly. Subsequent paragraphs add nuance. AI assistants extract from first-paragraph density.

Vertical-by-vertical findings

• Legal

  Top citation sources: government legal aid sites, established law firm content, CanLII, legal aid clinics. Brand law firm sites with deep practice-area content can compete; thin content gets passed over. Provincial regulatory specificity (statute names, limitation periods, jurisdiction-specific rules) wins.
• Dental

  Patient-education content from dental practices and dental associations dominates. Cost-explainer content (with C$ ranges) earns citations heavily. Generic 'best dentist in city' queries cite editorial sources (including 'best of' editorial guides like ours) more than individual practice sites.
• Home services

  DIY/educational content earns citations more often than service-promotional content. Practical 'how to know if you need X' content outperforms 'why hire us' content.
• Ecommerce DTC

  Comparison content ('X vs Y'), buying guides, and care/use guides cite heavily. Product-specific content cites less often (transactional queries trigger fewer AI Overviews). Brand-stories and origin content can earn citations on brand-search queries.
• B2B SaaS

  Vendor-comparison content, capability matrices, and pricing-explainer content earn citations. SaaS brands with substantial documentation and developer-focused content have outsized AI citation rates.
• Financial services

  Government and regulator-adjacent content dominates (Bank of Canada, OSFI, IIROC). Brand financial-services content struggles to compete with regulatory sources. Educational content (RRSP/TFSA explainers, mortgage process) earns citations when written substantively.
• Education

  Government education sites, university and college sites, and major news/educational publishers dominate. Individual private institutions earn citations through rigorous program-comparison and outcomes content.

Canadian vs international competition

On Canadian queries, US/UK/AU sources frequently appear in citations alongside Canadian sources. Pattern observations:

• On informational queries, international competition is intense

  US/UK sources with deeper content libraries often outrank Canadian sources for general informational queries even when the Canadian site is geo-relevant.
• On Canadian-specific queries, Canadian sources should win — but often don't

  Queries with explicit Canadian context ('PIPEDA', 'CRA', 'Ontario law') should be Canadian-source dominant. In our probes, Canadian sources do dominate, but not as overwhelmingly as you'd expect — often because Canadian sources publish less substantive content on these specific topics.
• Canadian small businesses systematically underpublish vs international competitors

  Across most verticals, the top US/UK competitors of Canadian small businesses publish 3–8x more substantive content per topic. This is the most fixable competitive disadvantage in our data.

What to do about it: practical playbook

• Identify your top 30 commercial-intent queries

  Use GSC + your own brand expertise. Run them through ChatGPT, Perplexity, and Google AI Overviews. Note who's currently being cited.
• Audit your content on each topic

  How does your content compare in depth, structure, author identity, and schema to the cited sources? Where's the gap?
• Pick 5–10 topics for substantial investment

  Don't try to compete on all 30. Pick the topics where you have the strongest expertise, the most relevant audience, and the best chance of winning citation share within 6–12 months.
• Publish substantively on each chosen topic

  1,500–3,500 words per piece. Structured for AI extraction (TL;DR, direct answers, FAQ sections). Named author. Schema-rich. Updated quarterly.
• Re-probe monthly

  Track citation share for your selected topics. Adjust content based on what's getting cited and what isn't.
• Be patient

  AI citation share takes 90–180 days to develop after substantive content publishes. Programs that abandon the strategy at 60 days haven't given it time to work.

Frequently Asked Questions

Reader questions about the methodology and findings.