What ChatGPT shopping queries are doing to affiliate creators.

Between December 2024 and June 2025, shopping queries on ChatGPT grew faster than any other query category on the platform. Comparison queries (which espresso machine under €500), deal-finding queries (cheapest place to buy a tripod right now), and review-checking queries (is the Sony XM5 worth it) are now routinely answered inside the chatbot. The recommendation surfaces. The user often doesn't click out at all.

For affiliate creators this is the most consequential shift in the channel's history. Affiliate marketing as a business model depends on a click happening on your link. When the buying decision is made inside a chat interface that doesn't link out (or links out to manufacturer sites instead of affiliate publishers), the click that triggers a commission never occurs.

This post is an honest read of what changed, what's measurable, and where the channel still works. No "AI is finished" doom, no "AI changes nothing" denial. For the broader 2026 picture across Perplexity, Google AI Overviews, and the actual revenue numbers, see is AI killing affiliate marketing? The 2026 data.

The mechanic that broke

Affiliate marketing works because of a chain of three events:

• A buyer searches for product information.
• A creator's content surfaces in the search results.
• The buyer clicks the creator's affiliate link and converts.

Each step has its own attribution mechanism. Search engines log impressions. Creators log clicks. Affiliate networks log conversions. The chain is observable end to end.

ChatGPT and similar AI search interfaces collapse step 2 and remove step 3 from many sessions. The buyer asks the question; the chatbot synthesizes an answer from millions of pages of training data plus live web retrieval. The buyer gets the recommendation. The buyer goes directly to the merchant. The creator whose content shaped the recommendation isn't in the link chain at all.

OpenAI's own data and third-party measurement studies converge on a rough number: roughly 80% of consumers now rely on zero-click results in at least 40% of their searches, contributing to an estimated 15-25% reduction in organic web traffic for content sites in shopping-adjacent categories.

What this actually looks like in the numbers

Three patterns show up in creator analytics in 2026:

1. Top-funnel queries lose traffic faster than bottom-funnel

"Best gaming headset under €100" loses meaningful traffic. "Where to buy the SteelSeries Arctis Nova 7 cheapest" loses much less. The AI is confident on broad comparison; it's less confident on real-time stock and price. Bottom-funnel commercial queries still surface human-written content in the source citations.

2. Conversion rate on remaining traffic goes up

The visitors who do click through from AI-influenced searches arrive with the comparison done. They've already decided. Conversion rate per click rises 30-60% on the click volume that remains. The total commission number goes down because click volume goes down faster.

3. Citation visibility becomes the new ranking

On Google, "you ranked first" was the measurable goal. In AI search, "you got cited as a source" replaces it. Some chatbots show citations; some don't. Some show them only on certain query types. The pattern that's emerging: a site cited frequently in AI responses gets a small but real trickle of brand searches and direct visits, which then convert. Sites that don't get cited disappear.

What you can actually measure today

AI search visibility tracking is roughly where SEO tracking was in 2008: crude, partial, but useful. Concrete things you can measure in 2026:

Brand-name searches

Set up Google Search Console alerts on direct searches for your site name + product categories you cover. When AI cites you, some users follow up with a brand search. A rising brand-search trend on a flat or falling overall-traffic chart is the most reliable AI-cited signal available without paid tools.

Direct-traffic referrals from chatgpt.com and similar

Some AI platforms now pass a referrer when a user does click a citation. ChatGPT (chatgpt.com), Perplexity (perplexity.ai), and Claude (claude.ai) all show up in analytics referrer data when users click out. The absolute volume is small. The trend matters.

Affiliate-tag-verified conversions per click

Click-to-conversion rate is now a better health metric than total clicks. A flat conversion rate with falling click volume is the AI-search pattern. A falling conversion rate is a different problem (a broken affiliate tag, a sales-page issue, see affiliate dashboards lie by omission). Knowing which one you're in matters.

Paid AI-citation tools

A growing category of paid tools (Profound, Otterly, AthenaHQ, and others) crawl AI search interfaces with synthetic queries and report how often your site or brand is cited. Useful if you can afford the €100-€500/month they cost. Less useful for creators with under €10k MRR.

What still works

Trust-bridge content

A specific kind of content survives the AI summarization filter: content that depends on the writer's identity, opinion, or testing. "Best espresso machine" gets synthesized away. "What I learned after using the Lelit Bianca for 18 months in a small apartment" doesn't. The chatbot can quote it, but it can't replace the writer.

Owned audiences

Email lists, paid communities, podcast subscribers, YouTube channel followers. Distribution you control doesn't pass through a chatbot interception layer. The creator who has 8,000 newsletter subscribers and earns affiliate commission from a Tuesday-morning send is unaffected by AI search. The creator who has 80,000 monthly visitors from Google but no email list is the one absorbing the hit.

Bottom-funnel commercial intent

Queries with explicit purchase intent (buy, coupon, discount, cheapest, in stock) still favor human-written content because AI is less reliable on real-time inventory and pricing. Optimizing existing content toward bottom-funnel intent is one of the highest-ROI moves in 2026. See also reviving old affiliate content.

Multi-platform presence

Affiliate revenue split across Google traffic, YouTube descriptions, Pinterest, newsletter, and a podcast is far more durable than the same revenue concentrated on Google search. AI is eating one of those channels harder than others. Diversification is no longer about hedging against algorithm changes. It's about distributing across surfaces AI summarization can't collapse.

The honest bottom line

Affiliate marketing isn't dying. The model still works on platforms AI can't replace (YouTube, Instagram, podcasts, newsletters) and for query types it can't synthesize (bottom-funnel, locally-scoped, identity-driven). What's ending is the era where a SEO-optimized comparison page could be the entire business. The creators thriving in 2026 own their distribution, write content that depends on their voice, and treat their affiliate catalog as monitored infrastructure (because every remaining click matters more, so silent leaks hurt more).

The biggest concrete shift to make this quarter: start treating your email list as your top-of-funnel acquisition target, not your bottom-funnel monetization channel. The chain becomes: content surfaces (somewhere, anywhere, including AI citations) → email subscriber → newsletter recommendation → affiliate click. The chatbot can intercept step one. It can't intercept the rest.