I read Google’s AI Search Playbook so you don’t have to

It’s a pretty comprehensive guide to being the most relevant answer in the new era of AI search.

Clock icon
3
min read

Marketing Week and Google have just released the second edition of The AI Search Playbook. It’s a pretty comprehensive guide for brands and agencies to be the most relevant answer by creating and capturing demand in the new era of search.

Here’s what matters, what to take with a grain of salt, and what it means for anyone who makes content.

‍

The big idea: rankings are out, relevance is in

The report opens with a pretty blunt line: in 2026, Google Search is AI Search. People aren’t typing “marathon shoes” anymore. They’re asking something closer to “I’m training for a marathon, my feet hurt, which shoes will help and how long should I break them in?” AI Mode queries now run three times longer than traditional searches on average.

That changes what visibility means. AI needs a reason to recommend you, not just to rank you in a list of search results. One contributor sums it up as the gap between being found and being selected.

Being found is nice. Being selected pays the bills.

‍

AI doesn’t create authority. It recognises it.

If you’re going to highlight one line in the report, make it this one: Google says the signals that get a brand surfaced are expert perspectives, customer reviews, consistent brand information, and structured product data. 

In other words, AI rewards brands that have put the work into making sure their marketing and content is high-quality, consistent, and something only that brand can produce (i.e. original data and expert perspectives). It can amplify good marketing, but it can’t save a weak value proposition or thin content.

But it’s this question from the report’s checklist that deserves a sticky note on every content team’s monitor:

Does every asset we publish help customers make a decision?

Not “fill a slot.” Not “hit a word count.” Help someone decide.

‍

Keywords can’t keep up with how people talk

When the same need can be phrased a thousand ways, keyword lists lose their importance. Google’s answer to the myriad ways your buyers can ask their question is AI Max for Search, which matches ads to the intent behind a query rather than to exact words. Google reports an average 27% more conversions at a similar CPA or ROAS. Trustpilot’s six-week pilot showed a 171% higher B2B conversion rate and a 61% lower cost per qualified lead.

The broader lesson applies even if you never touch AI Max: build content around topics and real customer problems, not keyword lists.

‍

Last-click is a flatterer

The measurement chapter is the most useful one for anyone who has to justify where the money goes.

Last-click attribution gives all the credit to the final step. It’s a bit like thanking the waiter for the meal and ignoring the chef. Content usually does its best work long before that final click, which is exactly where last-click isn’t looking. WPP Media’s Marcus Reynolds calls last-click ROAS one of the most misleading “good-looking” metrics on today’s dashboards.

The report recommends three things:

  • Agree with finance on what “value” means, whether that’s revenue, profit or lifetime value. (Yes, an actual meeting with finance. On purpose.)
  • Prove incremental improvement with controlled tests, geo-tests and marketing mix modelling.
  • Combine ROAS with lifetime value and profitability, rather than chasing one number into a corner.

Lenovo had a polished Search program built around its own brand name, and it worked well for people already looking for Lenovo. But growth was flattening, because more and more shoppers weren’t naming brands at all. They were describing problems: what they needed a laptop for, not which one they wanted.

Rather than guess, Lenovo ran a controlled test in the UK aimed at those problem-first searches. Within two weeks the new approach was pulling ahead. By the end of the pilot it had delivered 34% more conversion value and a 15% lift in revenue that wouldn’t have happened otherwise.

For content teams, the lesson has always been that your next customers might never type your name. They’ll describe their problem, and the brands that show up with genuinely useful content are the ones that answer it.

‍

Your budget might be the bottleneck

This chapter gives us two new phrases I suspect we might see pop up on LinkedIn.

The efficiency trap is when rigid cost-per-acquisition targets keep AI from reaching valuable new customers. The saturation illusion is when rising average costs look like a tapped-out market but actually reflect profitable growth into new demand.

Rising costs don’t always mean you’ve run out of road. Sometimes they mean you’ve found a new one.

The practical advice in the report is specific: set daily budgets at about four times expected spend, and watch for campaigns that hit 25% of their daily budget unusually early. But treat these as Google’s recommendations rather than gospel, and test before you adopt.

‍

Agentic AI is coming, and it reads your data

The final chapter looks ahead to AI agents that search, compare and buy on a customer’s behalf. While it’s particularly focused on B2C, it does have a lesson for B2B marketers too.

Whether an agent can buy from you comes down to your data. Structured data gives it the facts (price, stock, delivery). Unstructured content gives it the context (expertise, reviews, credibility). If an agent can’t work out whether your product is in stock, it won’t wait around. It’ll go buy from someone else.

What is relevant to B2B marketers is Google’s takeaway from this. As AI takes over the execution, our work moves to what machines can’t fake: brand voice, authority, credibility and authenticity.

‍

The grain of salt

This is a Google co-produced report, and it reads like one. Most of the recommended tools are Google products, and the headline performance numbers come from Google’s internal data and hand-picked case studies. That doesn’t make it wrong, but read the AI Max chapters as a pitch as well as a playbook.

The strategic core holds up regardless of which platform you use: be the most credible, useful answer to your customer’s real question, and measure the value that creates beyond the last click.

‍

The TL;DR

  • AI Search rewards relevance and authority, not rankings.
  • Quality content and consistent, structured data are now performance levers.
  • Build around customer problems, not keyword lists.
  • Measure incrementality and business value, not last-click.
  • Don’t let rigid budgets cap profitable demand.
  • Get your data ready for AI agents now.

‍

Read the whole report for yourself.

Sar is the CEO of Campfire Labs

Want more stories like this?

Subscribe to our newsletters for insights, ideas, and perspectives from the brightest minds in marketing, delivered straight to your inbox.

Need credibility that converts?

We help software companies produce their best performing content.

Get in touch