How ChatGPT and Gemini Affect Website Traffic

ChatGPT and Gemini are directly reducing website traffic for businesses that rely on Google search. When a user asks an AI assistant a question, they often get a complete answer without ever clicking through to a website. This is called zero-click behaviour, and it is not a future risk, it is happening right now, measurably, across industries. Studies tracking early 2025 data show that informational queries are losing 20-40% of their click-through rates compared to pre-AI-overview baselines. If your business depends on organic search traffic, this shift is already costing you leads.

That said, this is not a death sentence for website traffic, it is a structural change in how traffic is earned. Businesses that understand how ChatGPT and Gemini source their answers can position their websites to be cited, recommended, or linked inside AI responses. That is the new SEO game. The businesses losing traffic are those still optimising for yesterday's rulebook. The ones gaining are building for Answer Engine Optimisation (AEO), structuring their content to be the authoritative source that AI tools pull from.

This post breaks down exactly how these AI tools interact with your website traffic, which content formats survive (and thrive), and what practical steps you can take this quarter to adapt. Whether you are running a SaaS product, a local service business, or an ecommerce store, the implications are different, and specific.

The Mechanics: How ChatGPT and Gemini Pull Traffic Away

To understand the traffic impact, you need to understand how these tools work at a basic level. ChatGPT (when using Browse or GPT-4o with search) and Gemini pull real-time information from the web, summarise it, and present it in a conversational format. Google's AI Overviews, powered by Gemini, appear at the very top of search results for a wide range of queries. Users see the summary, their question is answered, and they do not scroll down to the blue links.

There are three distinct traffic patterns emerging from this behaviour. First, informational traffic, 'how-to' guides, definitions, explanations, is declining sharply. Second, transactional and local intent traffic is largely holding steady because AI tools still point users to businesses for purchases, bookings, and location-based needs. Third, branded and navigational traffic is unaffected, if someone is searching for your company by name, they are still coming to your site. The mistake most businesses make is treating all organic traffic as a single bucket when the AI impact is highly category-specific.

Google AI Overviews vs ChatGPT: Different Threats, Different Opportunities

Google's AI Overviews and ChatGPT behave differently, and the distinction matters for strategy. AI Overviews appear in Google itself, meaning the displacement happens before a user even considers clicking. ChatGPT is a separate destination, users who go there are already opting out of the traditional search funnel. Both reduce traffic, but through different mechanisms. AI Overviews affect your Google ranking visibility. ChatGPT shapes brand perception and recommendation patterns in a way that is harder to track but equally real.

Which Businesses Are Losing the Most Traffic

Not all websites are equally exposed. The hardest-hit categories are those whose primary traffic came from informational content: recipe sites, how-to blogs, news aggregators, dictionary and definition sites, and certain SaaS blogs that built their audience on keyword-driven educational content. If your growth playbook was 'publish 100 informational blog posts and rank for long-tail keywords', that model is under serious pressure.

Service businesses, agencies, consultants, freelancers, face a subtler threat. Potential clients still Google them, but they now also ask ChatGPT 'what should I look for in a web development agency' or 'is Shopify or custom development better for my business'. If your website is not structured to be the answer to those questions, you will not be recommended. This is why building a website that actually sells matters more than ever, it is not just about aesthetics but about structured, credible, citation-worthy content.

Ecommerce: A Mixed Picture

For ecommerce stores, the impact is mixed. Shopify store owners running content marketing to attract informational traffic are seeing declines. But product-level and category-level transactional queries are more resilient because AI tools generally still point users to stores for actual purchases. The bigger threat for ecommerce is AI-powered product discovery inside platforms like Google Shopping and Perplexity. If your product pages are not well-structured, AI tools skip over you entirely in favour of competitors with cleaner structured data.

What AI Tools Actually Cite, and How to Get Cited

This is the most actionable section of this post. AI tools do not cite randomly, there are consistent patterns in what gets surfaced. Understanding those patterns lets you engineer your content to be included. ChatGPT Browse and Gemini both favour authoritative sources with clear structure, specific data points, named authorship, and content that directly answers a narrow question in the first 100 words. Vague, padded content that takes 500 words to get to the point is consistently skipped.

• Structured Q&A sections: AI tools love FAQ blocks. Write them as standalone, complete answers, not teasers that require clicking through.
• Named expertise: bylines, credentials, and 'About the Author' sections signal E-E-A-T (Experience, Expertise, Authoritativeness, Trust) which both Google and OpenAI have confirmed influences sourcing.
• Specific data and statistics: round numbers, original data, and cited research are pulled far more often than generic claims.
• Schema markup: FAQ schema, HowTo schema, and Article schema make your content machine-readable and citation-friendly.
• Direct-answer opening paragraphs: the first 150 words should contain a complete, quotable answer. Don't build up to it.
• Clean page structure: H2/H3 hierarchy, short paragraphs, and scannable lists help AI parsers extract meaning accurately.
• Recency signals: AI tools prioritise recent content. A post last updated in 2022 is at a significant disadvantage.

The New Metric: AI Referral Traffic and Dark Traffic

One of the most frustrating aspects of AI search for business owners is the analytics gap. When ChatGPT or Claude recommends your website and a user clicks through, it typically shows up in Google Analytics as direct traffic or as an unknown referral, not as 'chatgpt.com'. This is what analysts call dark traffic. Your organic traffic numbers may look stable, but a significant portion of previously attributable search traffic has shifted to this dark bucket. If your 'direct' traffic has grown unexpectedly in 2024-2025 without a clear marketing push, AI referrals may be driving it.

The good news: Bing's integration with ChatGPT does pass some referral data, and Perplexity AI has started labelling its referrals. As AI search matures, attribution will improve. For now, the practical approach is to track UTM parameters aggressively on all your owned channels and watch for changes in the ratio of direct-to-organic traffic as a proxy signal. Understanding how SEO is changing for AI search will help you build the right measurement framework from the start.

Cost Reality: What Adapting Actually Requires

Business owners want to know what fixing this costs. The answer depends on where you are starting from. If you have an existing content-heavy site that just needs restructuring for AI visibility, a content audit and restructuring project typically runs between $500-$1,500 depending on the number of pages and the depth of changes required. If you need to build net-new content with proper AEO structure, budget $100-$250 per substantial post. If your website itself needs to be rebuilt with proper semantic HTML, schema markup, and performance optimisation, a full rebuild starts around $1,800-$6,000 depending on complexity.

The mistake businesses make is treating this as a one-time fix. AI search optimisation is ongoing, models update, citation patterns shift, and competitors who start later will leapfrog you if you stop publishing. The businesses I see winning are not just fixing their existing content; they are building a publishing cadence of 2-4 well-structured, expert-driven posts per month and combining that with technical improvements to their site architecture. See our services page for how we approach AEO as part of a broader web development engagement.

What Not to Do: Common Mistakes

• Publishing AI-generated content at volume without expert review: AI tools can detect and deprioritise thin AI content. It needs a human layer of expertise, opinion, and specificity.
• Ignoring technical SEO fundamentals: schema markup, page speed, and mobile performance still matter, AI crawlers use the same signals as Google's traditional crawler.
• Abandoning long-form content entirely: some founders have overcorrected by cutting content investment. Long-form, well-structured content is still the backbone of AI citations.
• Treating all traffic loss as AI-related: check your cannibalisations, Core Update impacts, and technical issues before blaming the AI shift. Analytics hygiene matters.
• Not updating old content: a post from 2021 that once drove traffic needs a 2025 refresh with updated data, schema, and structure, not a new replacement post.

Strategies That Are Working Right Now

Beyond structural content changes, there are broader strategic moves that are working for businesses navigating this shift. The most effective is building what I call a 'citation moat', a cluster of authoritative content around your core topic that gives AI tools no better option than to reference you. If you are a Shopify development agency, you want to own the conversation around Shopify costs, Shopify vs custom development comparisons, Shopify performance, and Shopify app recommendations. Not one post, a cluster of ten well-linked, expert posts that collectively signal deep topical authority.

Another strategy that is working: structured data for local businesses. If you run a physical business or serve a local market, schema markup for LocalBusiness, Service, and Review types is making a measurable difference in AI Overview inclusion. Google's Gemini draws heavily on structured data when constructing local recommendations. This is less glamorous than content strategy but it has a high ROI, a developer can implement comprehensive schema on most sites in 4-8 hours.

Finally, do not underestimate the power of being mentioned in third-party sources that AI tools trust. Being featured on industry blogs, getting press mentions, appearing in comparison roundups, and earning backlinks from authoritative domains all influence how AI tools perceive your brand's authority. This overlaps with traditional link-building but with a different end goal: you are building trust signals that feed into AI training data and real-time citations, not just PageRank. Learn more about how to make your website visible in AI tools for a deeper dive into this approach.

The Bigger Picture: AI Search as a Distribution Channel

The framing of 'AI is killing website traffic' misses a more useful perspective. AI tools are becoming a distribution channel in their own right. When ChatGPT recommends your agency for a Shopify project, or Gemini surfaces your pricing page in an AI Overview, that is qualified, high-intent exposure that costs you nothing beyond the investment in good content and technical structure. The businesses that will win in this environment are the ones who stop fighting the shift and start optimising for it.

This does not mean abandoning traditional SEO. The blue links still exist, people still click them, and Google still drives billions of searches per day. What it means is that the content you publish now needs to serve two audiences simultaneously: the human reader who lands on your page, and the AI crawler that extracts structured answers from it. These two goals are more aligned than they sound, clear, well-structured, expert content serves both. The era of 'write for the algorithm, confuse the human' is over.

What to Do This Quarter

You do not need to overhaul everything at once. Here is a prioritised action list that any business, regardless of budget, can start executing immediately.

1. Audit your top 10 traffic pages: identify which are informational (high AI risk) vs transactional (lower risk). Focus restructuring effort on informational pages first.
2. Add FAQ schema to your top 5 pages: use Google's Structured Data Markup Helper. This is a half-day task that pays off consistently.
3. Rewrite your opening paragraphs: every key page should open with a direct, quotable answer in the first 100-150 words. No preamble.
4. Set up referral source tracking: add ChatGPT and Perplexity as custom channel groupings in GA4 so you can start tracking AI referrals as they grow.
5. Commission 2 expert-written, AEO-structured posts per month: not AI bulk content, real expert content with data, specifics, and proper H2/H3/FAQ structure.
6. Check your page speed: AI crawlers penalise slow pages just like Google does. A slow website actively costs you revenue in both traditional and AI search.
7. Build internal links between related content: AI tools use internal linking to understand your topical authority. A cluster of well-linked posts outperforms isolated pages.

The businesses that treat this as a crisis will react slowly, make expensive mistakes, and lose ground. The ones that treat it as a structural market shift, like mobile search was in 2015, will invest methodically, adapt their content strategy, and come out with stronger visibility than before. AI search is not the end of website traffic. It is a filter that rewards expertise and punishes mediocrity. If your content is genuinely useful and expertly structured, that is an advantage, not a threat.