ChatGPT Atlas, Google AI Overviews, and Perplexity have fundamentally shifted how people search—and which brands get cited. This post explains the new ranking factors for AI answers, why your competitors might be appearing instead of you, and tactical ways to earn citations.
In 2024, we saw the tipping point. By mid-2026, ChatGPT Atlas (OpenAI's real-time browsing model), Google AI Overviews (formerly SGE), and Perplexity have fundamentally changed how people find information. Instead of clicking through ten blue links, users get a synthesized answer **with cited sources** right at the top. I've watched my own clients' traffic patterns shift: fewer exploratory clicks, more direct conversions from people who already trust the brand because an AI tool vouched for it. ChatGPT Atlas browses the web in real time and builds answers from current pages. Google AI Overviews pull from the Knowledge Graph, featured snippets, and indexed pages, then generate a coherent response. Perplexity combines search with large language models and cites academic papers, news, and authoritative blogs. The common thread? All three choose **which sources to cite**. You're no longer competing just for position one—you're competing to be selected as the authoritative voice the AI trusts. That's a different game, and many mid-sized businesses are losing it without realizing why.
I get this question weekly from clients in Ottawa and across North America: "We rank well in traditional search, so why does Perplexity cite our competitor instead?" The answer usually comes down to **recognizable expertise signals** that AI models can parse. Competitors win citations when they have author bios with credentials, published bylines on third-party sites, and schema markup that tells the model "this person is a real expert." If your competitor's blog has an AuthorProfile schema and yours doesn't, the AI sees theirs as more trustworthy. Backlink profiles still matter enormously—if the competitor has links from government sites, universities, or major industry publications, those act as votes of confidence. Recency plays a role too; Perplexity especially favours content updated in the last six months. I've also noticed that **structured answers**—numbered steps, comparison tables, clearly labelled sections—get picked up more reliably than narrative-style posts. Finally, if your brand isn't strongly associated with specific topics in the Knowledge Graph or across many indexed pages, the AI might not see you as a topical authority, even if you rank decently for a handful of keywords.
ChatGPT Atlas browses live web pages and evaluates them on the fly. From what I've observed working with clients who do get cited, it prioritizes **clarity and directness**. If your page rambles for 400 words before answering the question, Atlas often skips it in favour of a competitor who front-loads the answer. First-person expertise helps—"I've done 200+ audits" signals experience in a way generic third-person text doesn't. Atlas also seems to weigh **domain credibility**: sites with strong backlink profiles and consistent publishing in a niche get cited more. I had a client in the HVAC space add detailed case studies with real project photos and first-person walkthroughs; within three months, ChatGPT started citing them for technical HVAC questions in the Ottawa region. One nuance: Atlas doesn't see your internal linking or site structure the way Google does—it evaluates each page more independently. So even a single stellar, well-sourced article on an otherwise modest site can earn a citation if it clearly demonstrates expertise and answers the query precisely.
Google AI Overviews lean heavily on the **Knowledge Graph and entity associations**. If your brand, authors, or key topics aren't recognized entities, you're at a disadvantage. I tell clients: get a Wikipedia page if you can, earn mentions on industry association sites, and use consistent NAP (name, address, phone) everywhere so Google connects the dots. Schema markup is critical—ArticleSchema, FAQSchema, HowToSchema—all give Overviews structured data to pull from. Google also prefers **pages that already perform well in traditional search**: if you own the featured snippet for a query, there's a decent chance the AI Overview will cite you too. One pattern I've noticed in 2026: Overviews increasingly cite **primary research and original data**. A competitor's survey of 500 Canadian SMBs will beat your summary of someone else's survey every time. Freshness counts, but less aggressively than with Perplexity; Google values long-standing authority. Finally, user engagement signals—low bounce rates, high time-on-page—seem to correlate with Overview citations, though that's harder to prove causally.
Perplexity is the most transparent about sources because it lists them inline with every claim. From client work and my own experiments, Perplexity loves **recent, citation-rich content**. If you link out to credible studies, government data, or peer-reviewed papers, you signal that you're building on solid ground, and Perplexity rewards that. Publishing cadence matters—I've seen blogs that update twice a month get cited far more than static sites, even when domain authority is similar. Formatting is huge: bullet points, tables, and clearly labelled headings make it easy for the model to extract structured answers. Author credibility plays a role here too; adding an author bio with LinkedIn, prior publications, or credentials boosts citation odds. One practical trick: if there's a data gap in your niche, fill it. A roofing client published a pricing survey of 40 Ottawa contractors; Perplexity cited it within weeks because no one else had comparable primary data. Perplexity also seems to favour **answer-first content**—don't bury the lead. Start with the direct answer, then provide supporting detail.
Start by auditing your top-performing pages for **demonstrable expertise**. Add author bios with credentials, link to your LinkedIn or industry profiles, and use first-person language where it fits naturally. Implement schema markup—at minimum, Article and Organization schema; if you publish how-tos or FAQs, use those schemas too. Next, identify queries your audience asks and create **direct, structured answers**. Use headings that mirror the question, front-load the answer in the first paragraph, and support it with data or examples. Publish original research if you can; even a small survey or case study series differentiates you. Build or refresh your backlink profile with a focus on **topical relevance and authority**—one link from a .gov or major industry publication beats fifty directory links. Update older content every six months; add new data, refresh examples, and change the publish date so AI tools see it as current. Finally, monitor which competitors get cited (use Perplexity and ChatGPT yourself for your key queries) and reverse-engineer what they're doing: schema, formatting, author signals, link sources. This isn't quick, but I've seen meaningful citation gains in three to six months when clients commit to it.
AI search changing SEO means reallocating budget toward **content quality and authority signals** rather than sheer keyword volume. In 2020, I'd often recommend 15–20 blog posts a quarter to build topical coverage. Today, I'd rather see five deeply researched, expert-authored pieces with original data and strong schema. Expect to invest more in subject-matter experts—paying a credentialed professional to write or co-author content costs more but delivers citations. Budget for ongoing content refreshes; a one-and-done approach doesn't work when Perplexity favours recency. Link building shifts too: generic outreach for backlinks matters less than earning mentions from high-authority, topically relevant sources. For Canadian SMBs, that might mean sponsoring industry research, contributing to trade publications, or getting featured in local news. Practically, an AI-aware SEO program in 2026 might run $3,000–$8,000/month for a mid-sized B2B company: strategy, expert content creation, schema implementation, link building, and quarterly audits. That's higher than five years ago, but the upside—being the cited source in AI answers—can drive more qualified leads than ranking #3 ever did. It depends on your niche and competition, but the ROI is there if you commit to the new rules.
Run your key queries in ChatGPT, Google, and Perplexity and note which brands they cite. If competitors appear repeatedly and you don't, compare your content for author bios, schema markup, backlink profiles, and answer clarity. Tools like Ahrefs or Semrush can show you backlink gaps. I also recommend checking if your competitor's authors have LinkedIn profiles or bylines on third-party sites, signaling recognized expertise that AI models trust.
Yes. AI tools rely on indexed, authoritative web pages to generate answers. Strong domain authority, quality backlinks, and ranking for featured snippets all increase your odds of being cited. Think of traditional SEO as table stakes—it gets you in the pool of sources AI considers. Then expertise signals, schema, and content structure determine whether you're actually chosen for the citation.
Start with Article schema including author and datePublished fields. Add Organization schema with a logo and social profiles. If you publish how-tos, use HowTo schema; for FAQs, use FAQPage schema. Product and Review schema matter for e-commerce. Validate with Google's Rich Results Test. Schema won't guarantee citations, but missing it puts you at a disadvantage because the AI can't easily parse your content's structure and credibility.
Every six months for core pages, more often if your niche moves quickly. Perplexity especially favours recent content. I've seen citation frequency jump after clients refresh a post with new data, updated examples, and a current publish date. For ChatGPT Atlas, recency matters less than accuracy and clarity, but stale information will hurt trust. Prioritize your highest-traffic or highest-value pages first.
Absolutely, especially in niches where you have genuine expertise. Publish original research, case studies, or local data your competitors ignore. Use first-person, credentialed authors. Build relationships for high-quality backlinks—one mention in a regional industry report can matter more than dozens of generic links. I've seen Ottawa-based clients get cited over national brands because they provided specific, locally relevant answers with clear expertise signals. It's harder, but absolutely doable.