2026 was the year publishers got less traffic but better shoppers.

Search referrals fell and fell some more. AI referrals stayed too small to fill the gap. Yet when we broke down partner performance across the StackCommerce network this year, the deals themselves didn’t get weaker. The decline almost always traced back to fewer sessions.

Holiday shopping is still ahead of us, so we’re working from last year’s results and this year’s forecasts. Here’s what the rest of 2026 showed us, and what we expect from 2027.

Search Fell Faster, and Sessions Became the Constraint

Google Search traffic to publishers fell 40.2% year over year from July 2025 to July 2026, according to Chartbeat’s 2026 Publisher Playbook, nearly double the 21.9% drop the year before. Google Discover’s decline trended the same way, from 6.6% to 34.3%. Smaller publishers took the hardest hit: Over two years, Chartbeat found small publishers lost 60% of their search referrals, compared with 47% for mid-size publishers and 22% for large ones.

Inside our network, what fell was the number of people arriving, not how they shopped. A regional newspaper group saw sessions drop by more than half, while its conversion rate and profit per session both improved. A B2B tech publisher took fewer orders than last year, but each one earned about 10% more, and its effective payout rate went up.

The clearest proof came from a national news publisher. When its traffic came back over the summer, its deals converted at the same rate as before. The deals hadn’t changed. The volume had.

So the real question is where sessions come from. The partners holding steadiest leaned on distribution search doesn’t control: newsletters, social deal drops, push and app placements, video, and syndication and repost programs.

Prediction 1: Sessions stay the scarce resource

Search referrals will continue to decline in 2027, and the publishers who win in commerce will be those who control their own distribution. Expect commerce teams to get measured on reach per placement instead of article count.

Fewer Placements, More Profit

When sessions are scarce, more articles don’t add up to more revenue. A top tech and culture publisher in our network cut its daily deal cadence from 5 articles to 3, and profit per article nearly tripled. Volume paid off only for the very largest publishers. Adding placements still brought in more overall, but profit per article went down.

Fewer placements means each one needs a clear job. Even within last holiday season, different placements did different work: Conversion across our network nearly doubled from Black Friday through Cyber Monday, while weekend deal drops brought big session spikes at lower efficiency. One was a conversion play, the other a reach play. We covered more of last season in our 2025 holiday recap.

What goes into the placement matters as much as when it runs. Evergreen productivity software drove most of the biggest days of 2026.

Prediction 2: Quality over quantity becomes the default plan

More publishers will run fewer, better-planned deal placements throughout the year and protect their conversion rates instead of flooding feeds.

AI Traffic Is Small, and It Converts

AI chatbots still send well under 1% of publisher pageviews. Nobody should count on them to replace lost search traffic. But more shoppers are starting there: PwC’s holiday research found 29% of consumers plan to use AI in their shopping this year, up from 22%. Most use it for research, and over half use it to compare prices.

The shoppers AI does send convert better. In September, AI traffic to U.S. retail sites converted 90% better than non-AI traffic and brought in 83% more revenue per visit, according to Adobe, which expects that revenue gap to double this holiday season. Our own shop data shows the same pattern, including on publisher storefronts. On StackSocial, visitors who arrive from AI assistants like ChatGPT and Perplexity convert at roughly twice the rate of the site’s overall traffic. On our publisher partners’ storefronts, they convert about 25% better than the rest of their traffic.

Meanwhile, the in-chat checkout experiment stalled. OpenAI launched Instant Checkout in ChatGPT, then ended in-chat checkout in March 2026 and started sending shoppers back to merchant sites to pay. Google’s Universal Commerce Protocol keeps the retailer as merchant of record. For commerce partners, that’s the detail that matters: The sale still happens on the merchant’s site, where affiliate tracking works, as long as the shopper arrives through a tracked link.

Prediction 3: AI becomes a measurable commerce channel, and attribution becomes the fight

AI referrals will keep growing from a small base. The open question is credit when there’s no tracked link. When a shopper researches in an AI tool that drew on publisher content, then clicks straight from the answer to the merchant, who gets paid? Expect more pressure on networks and brands for clean, verifiable attribution. We covered more of this in our post on AI discovery and attribution.

Relying on One Retailer Got Riskier

Google wasn’t the only platform that changed the terms this year. This spring, Amazon cut affiliate commissions by as much as 50% for some publishers, eliminated milestone bonuses, and stripped down the reporting affiliates used to optimize, according to Adweek. None of it was publicly announced. On both the traffic side and the revenue side, the lesson is the same: When one source carries the business, someone else sets your terms. Most publishers already know that. Acting on it is harder. In October, we looked at why diversification still stalls even when everyone agrees it’s needed.

The affiliate channel itself is still growing, and the growth is spread out. Adobe expects affiliates and partners to grow their share of U.S. online retail revenue by 12% this holiday season, more than any channel except social. Cashback and rewards sites make up 36% of affiliate traffic, and creators 22%. Moving away from a single retailer isn’t a retreat. It’s where the channel is growing.

Prediction 4: Diversification stops being optional

After a year of commission cuts and reporting changes from a single large retailer, partners will plan around the risk of concentration. They’ll spread revenue across more merchants, more categories, and more affiliate types, including cashback, rewards, and deal communities.

Commerce Is Reaching New Publishers

This year we saw more interest from local and regional media groups and broadcast-owned sites looking for commerce revenue that doesn’t depend on search. The regional newspaper group above shows why it can work: Its sessions fell by more than half, and its conversion rate and profit per session still went up.

Prediction 5: Commerce spreads to new publisher types

Local, regional, and broadcast-owned media, along with creator and newsletter businesses, will add commerce as a revenue line that doesn’t depend on Google. The infrastructure is ready for them: storefronts, curated deals, and repost programs.

Before You Plan 2027

Before 2027 planning locks in, look at your own commerce numbers as sessions times conversion rate times profit per session. If conversion held up and traffic didn’t, the fix is distribution, not different deals.

That’s the work we do with partners every day at Stack: helping publishers and affiliates turn the traffic they have into commerce revenue, with curated deals, storefronts, and data on what converts. If you’re mapping out 2027, let’s talk: pubams@stackcommerce.com