I’ve always found predictions posts slightly embarrassing to reread a year later, so I’ll keep these modest. But 2025 moved more in search than any year I can remember since mobile, and it’s worth writing down where I think it goes next.
A quick look back
In March, Google put AI Mode into Labs for US subscribers. By May it was rolling out to everyone in the US, and in late July it reached the UK. Last month Gemini 3 went into AI Mode on day one, which tells you how central Google now considers it.
Meanwhile the evidence on clicks piled up. Ahrefs found AI Overviews cut click-through for the top result by about a third; Pew found people clicked a traditional result in 8% of visits when an AI summary showed, against 15% without. I wrote about measuring that in your own data back in May.
The reporting rules shifted under our feet too. Google required JavaScript for search in January and dropped the num=100 parameter in September, and each time a lot of SEO data broke with no announcement.
On privacy, Chrome kept third-party cookies and then retired most of Privacy Sandbox. The UK passed the Data (Use and Access) Act, which should loosen consent for some analytics.
And OpenAI started pushing into commerce and browsing: Instant Checkout in ChatGPT in September, its own Atlas browser in October.
We’re also in the middle of the December core update as I write, the third of the year. I’ll leave that one alone until it finishes.
Six things I expect in 2026
1. AI Mode gets more prominent, slowly. I don’t think Google makes it the default results page next year. Too much ad revenue depends on the current layout, and the ads model inside AI answers is still being worked out. But expect it to be pushed harder: more entry points, more queries routed there, more of it on mobile.
2. Informational organic traffic keeps falling, and reporting finally catches up. The sites that rely on “how to” and “what is” traffic will see another hard year. I expect more businesses to stop reporting sessions as the main SEO measure and move to clicks, conversions and revenue from the queries that matter. That shift is overdue regardless of AI.
3. “AI visibility” becomes a budget line, and most of the tools stay rough. A lot of vendors launched trackers this year that check whether your brand appears in ChatGPT, Perplexity or AI Mode answers. I expect marketing teams to start paying for them. I also expect the data to stay noisy, because AI answers vary by prompt, by user and by day. Treat these tools as research, not as a KPI. I made a similar point in June and I still think it holds.
4. Agentic shopping is slower than the hype. Checkout inside ChatGPT is a real product and Google has its own payments protocol for agents. But buying something is where trust, returns, fraud and payments regulation all bite at once. I expect lots of announcements and pilots in 2026, and very modest actual sales through AI agents by the end of it. Product data quality is the bit worth working on now; clean feeds and accurate stock and pricing help whatever form this ends up taking.
5. Google keeps clamping down on scraping. Two breaking changes for SEO tools in one year is a pattern. I’d expect at least one more, and rank tracking to get more expensive or less granular as a result. Search Console and your own analytics become even more important as the record.
6. First-party measurement stops being a project and becomes the default. Server-side tagging, enhanced conversions, Consent Mode done properly, and some form of incrementality testing. Very little of this is new. What’s new is that the alternatives (cookies, deep rank tracking, clean referrer data) keep getting less reliable.
What I’m not predicting
The end of SEO, the end of Google, or the replacement of marketing teams by agents. I’ve read all three this year, more than once. Google’s search volume is still enormous, people still click on things they want to buy, and a fast, clear, useful site still matters whoever is reading it.
I’ll revisit these next December and see how badly I did.