Google Search traffic to publishers fell 40.2% year over year from July 2025 to July 2026, according to Chartbeat’s 2026 Publisher Playbook. That’s nearly double the 21.9% drop it measured the year before.

None of that will surprise anyone running a publishing business. When we talk to partners at StackCommerce, almost all of them say they need to diversify beyond search. Then we ask what they’re doing about it, and the answer gets fuzzier. Budget goes back to the content that’s always worked, the new channel gets a pilot, and the pilot turns into “we’ll revisit next quarter.”

We don’t think that’s a knowledge problem. We think it’s a reporting problem, and it’s worth unpacking.

Search Isn’t the Only Point of Failure

Will AI pick up where search left off? So far, it hasn’t. AI referrals grew eightfold over two years to make up 0.01% of traffic, and they slipped in 2026. GEO matters for how content shows up in AI answers, but it won’t replace the search volume publishers have lost.

The commerce side has its own version of this problem. This past spring, Amazon cut affiliate commissions by as much as 50% for some publishers, eliminated milestone bonuses, and stripped down the performance data affiliates used to optimize, according to Adweek.

Put those together, and the pattern is the same on both sides. Leaning on one traffic source or one retailer means someone else sets your terms, and both of those someones changed their terms this year. Across our network, a wider mix of products and affiliate partners is what protects revenue when that happens.

The Retreat Is a Rational Bet

Some partners are already moving, and a couple of major tech publishers are building formal GEO strategies. But plenty of outlets are sticking with what’s worked before, and honestly, we get it.

A publisher’s proven content has a known return. Its domain authority, its testing, and the trust it’s built with readers over years all show up in the numbers. A new channel doesn’t have that yet. If you’re judged on this quarter’s results, backing the proven thing is the reasonable call.

That’s where reporting comes in. Diversification works like insurance, and insurance doesn’t show up in a QBR. It pays off years from now, when a traffic source or a retailer changes the rules and you’re not the one scrambling. A quarterly dashboard can’t see that risk, so it keeps rewarding the retreat.

Where Partners Are Finding Traffic Instead

The partners making progress didn’t give up what works. They took it to places search doesn’t reach.

Newsletters are the clearest example. One publisher runs deal content across 7 newsletters, and another puts deal posts in a digest that reaches 25,000 subscribers. Social is picking up too. One major tech publisher has spent two years building a social deal strategy, and an audio entertainment partner has asked for weekly social deal drops. Others are putting more into Apple News and push notifications.

Then there’s video. One science publisher embeds YouTube Shorts inside its long-form articles and has found they perform better in LLMs. It didn’t build a video business from scratch. It added video to the articles it was already good at. Several other publishers are investing in video now too.

Chartbeat’s numbers back this up. Direct traffic grew from 13.5% to 15.9% of pageviews over the same two years that search fell. Readers who already know where to find you are a growing share of the audience, and newsletters, apps, and push are how you reach them.

Why Q4 Is a Good Time to Test

Holiday shopping season makes one thing plain: BFCM is a conversion event.

Conversion rates across the StackSocial network last year nearly doubled from Black Friday through Cyber Monday.

That’s the case for planning traffic and conversion separately. Where your traffic comes from is shifting fast. When it converts best hasn’t changed. A new surface tested this quarter gets its first real read at the moment readers are most ready to buy.

Nobody needs a plan to rebuild distribution from scratch, and most partners don’t have the bandwidth for one anyway. The realistic version is smaller: one new path, tested from inside an existing strength, judged over more than a single quarter.

If you’re stuck between the plan and the execution, let’s talk before 2027 planning locks in another round of “we’ll revisit.”