US advertisers will spend $13.81 billion on affiliate marketing in 2026, up 11.3% from $12.42 billion the year before, according to eMarketer’s 2026 forecast. That is faster than US retail ecommerce is growing overall. If you have read a dozen headlines this year announcing that AI killed affiliate marketing, that number should stop you.
The money did not leave. It moved. And it moved away from exactly the kind of site that most beginner affiliate guides, including some of the ones on this site, taught people to build: a niche review blog that ranks on Google and collects commissions on tracked links. That model is not slowing down. It is coming apart. The channel is healthier than ever and the classic playbook is dying at the same time, which is the part almost nobody is explaining clearly.
I have a stake in getting this right. I own a portfolio of content sites, some of them affiliate-monetized, and I have watched buyer-guide traffic slide across the whole portfolio over the past year. So this is not a theory I read about. It is a spreadsheet I look at.
What actually collapsed
The damage is concentrated in one place: organic search traffic to review and buyer-guide content.
New York Times Wirecutter, the most cited affiliate publisher in the world, lost more than 60% of its Google search visibility between May and August 2025. Independent tracking tools measured the same fall from different angles: Sistrix put it at 69%, Ahrefs at 72%, Semrush at 65%. Wirecutter is not a thin-content farm. It runs 4,500-word articles built on real product testing. If depth and authority were enough to survive AI Overviews, Wirecutter would have survived. It did not.
The pattern is industry-wide, not one publisher’s bad quarter. Google’s March 2026 core update knocked rankings down for 71% of affiliate sites. Ahrefs found that AI Overviews cut click-through rates for the number-one organic result by 58% in late 2025. SparkToro measured that 58.5% of US Google searches already ended without a single click to the open web back in 2024, and AI answers have only widened that gap since.
Talk to the people running publisher commerce teams and the revenue picture matches the traffic picture. Paul Cunliffe, a longtime affiliate consultant who ran affiliates at Time Inc UK, described the hit bluntly: “a huge drop in traffic, a corresponding drop in affiliate revenue from 20-40% in some cases, which is really a massive hit.” Roger Lynch, the CEO of Condé Nast, told publishers to plan “as if search is zero.” Close to seven in ten publishers now say they are worried that Google’s changes will damage their affiliate business. That is not panic. That is people reading their own analytics.
I covered the broader version of this story earlier this year in the piece on AI Overviews erasing publisher clicks. The affiliate slice of it is sharper, because affiliate economics are almost purely click-based. No click, no commission. A newsletter or a course can survive a traffic cut by monetizing a smaller, warmer audience harder. A review page that earned $4 per thousand visitors has no such lever. When the visitors stop arriving, the revenue is simply gone.
Where the commissions went instead
Here is the reframe that matters. Total affiliate spend is up. So if review sites are collecting less of it, someone else is collecting more. Three groups are.
Creators are the fastest-growing category in the channel. On Awin’s network, the share of affiliate revenue flowing to content creators rose from 15.9% to 19.5% in a single year. The reason is structural, not trendy. A creator does not depend on a Google ranking that an AI summary can intercept. They own the relationship. Buyers typically need three to four exposures to a product before they buy, and by the eighth time a creator features the same product, click-through rates hit 1.8 times the first mention. That compounding trust is something an AI Overview cannot replicate and cannot steal, because it lives inside an audience the creator controls.
Cashback and loyalty platforms are quietly winning. They already command roughly 35% of affiliate ad spend, the single largest slice, and Julian Henrichs, an affiliate consultant with 14 years in the channel, explained why they are almost immune to the AI shift: “users have to visit their platforms to activate rewards before purchasing. The click survives, and so does affiliate attribution.” An AI assistant can tell you a product is good. It cannot activate your Rakuten cashback for you. The transaction still routes through the platform, so the commission still gets paid.
Coupon and deal publishers are still huge but exposed. Discount and coupon publishers captured 42.4% of US affiliate revenue in the first half of 2025, up from 39.7% a year earlier. But their moat is thinner than the cashback platforms’. An AI answer can surface a working discount code directly in the response, cutting the publisher out of the loop entirely. Big today, structurally vulnerable tomorrow.
There is one more twist that most affiliate marketers have not priced in. Your affiliate content may be worth more as AI training fuel than as a click generator. When ChatGPT talks about the eyewear brand Zenni, nearly 70% of the sources it cites trace back to affiliate marketing content. The reviews still shape the buying decision. They just do it inside the model now, uncredited and uncompensated, instead of on a page where a commission fires. That is why generative engine optimization, getting your content cited by AI systems, has gone from a curiosity to a survival skill in about eighteen months.
What I changed in my own affiliate content
I stopped treating rankings as the asset. The asset is the audience and the attribution path.
Concretely, three shifts. First, I moved effort out of net-new “best [product] for [use case]” pages, which are the exact format AI Overviews eat, and into content that answers a question an AI cannot fully resolve in a paragraph: long-term ownership reports, real teardown comparisons, edge-case use I have actually tested. Depth over breadth, which is the same advice Cunliffe gives, and it is working better than defending twenty thin pages.
Second, I started routing readers toward monetization paths where the click survives contact with AI. That means leaning into recurring-commission programs and high-ticket offers where one converted reader is worth months of small commissions, and it means building an email list so I am not renting my entire audience from Google’s algorithm.
Third, I got honest about which pages to let go. Armando Roggio at Practical Ecommerce put the operational version of this well: the publishers that make it will know their revenue per article, kill the deadweight, and personalize the follow-up. I now look at earnings per session, not just traffic. A page with strong earnings per session and weak distribution is a promotion problem worth solving. A page with heavy traffic and near-zero earnings was never a business.
The playbook that still works
If you are starting affiliate marketing in 2026, the honest version of the advice is different from what this site published in 2024, and I would rather say that plainly than pretend the old guide still holds. The beginner affiliate guide still gets the fundamentals right, disclosure, choosing programs, building trust. What has changed is where the leverage lives.
Build an audience you own before you build a library of pages you rent. Pick a monetization model where the click cannot be intercepted: creator-led recommendation, cashback-style attribution, high-ticket or recurring offers, or a real email relationship. Write for depth and lived experience, because that is both what humans still click on and what AI systems still cite. And measure revenue per piece, not traffic, so you know which content is actually a business and which is just a chart that used to go up.
Affiliate marketing did not die. $13.81 billion does not die. The comfortable version of it, the one where you rank a review page and wait for Google to send buyers, is what died. The people treating that as a crisis are the ones who built their whole business on the rented land. The people treating it as a reshuffle are already collecting the same commissions from ground that AI cannot take away.