May’s State of Search & AI unpacks why organic traffic is only part of the picture and what marketers need to measure — and build — differently to keep up.
May’s State of Search & AI unpacks why organic traffic is only part of the picture and what marketers need to measure — and build — differently to keep up.
A trend is happening in search traffic that’s easy to misread. Organic clicks are softening for many brands, and the instinct is to treat that as a performance problem. But in many cases, it’s actually a behavior shift.
AI tools are handling more of the research and evaluation work before a user ever reaches a website, which means the visit itself occurs later, with more intent behind it. May’s signals are largely about understanding that dynamic and adjusting both measurement and content strategy to account for it.
April was about fragmentation. Discovery spread across platforms, AI Mode interceded in branded queries, and Google Maps evolved from a directory into a recommendation engine. The broader pattern was that visibility is becoming harder to predict and measure. It’s no longer centralized on a single platform, tied to a stable ranking position, or guaranteed even for well-established brands.
May narrows the lens. Where April raised the question of where brands show up, May is about what happens when they do, and how to tell whether it’s working.
This month’s developments reinforce that the traditional click-through is no longer the primary unit of value in search. Discovery is happening earlier, intent is arriving later, and the infrastructure that determines AI inclusion is becoming just as important as content quality.
When AI tools answer a question directly in the interface, users often don’t click through. However, that doesn’t mean the brand went unnoticed. What’s happening instead is a rise in direct traffic: users who encountered a brand through search or AI earlier, then came back on their own terms. This conversion path is different from what marketers are used to tracking.
The challenge is that most attribution models weren’t built for this path. A session that starts with a Google AI answer and ends three days later as a direct visit looks, in the data, like it came from nowhere. It didn’t.
What this means for marketers: Before drawing conclusions from softening organic click volume, look at trends in direct and branded traffic alongside it. If those are holding or growing, the discovery engine may be working fine, just delivering value through a different door.
In local AI search especially, the website has taken on a new function. AI systems read sites to decide whether a business is worth surfacing at all. Location details, service descriptions, pricing signals, and trust indicators are all parsed before a recommendation is made. A site that’s clear and well-structured is more likely to be cited accurately; one with buried or ambiguous content may be skipped.
This reframes what “local content” is for. While it still serves as copy written for a human to read on a screen, it’s also structured data that needs to be legible to a machine making a recommendation.
What this means for marketers: Audit local pages for AI readability. Look for clean schema markup, descriptive H2s and H3s, short sentences, and FAQ content written the way customers actually ask questions. The goal is to enable AI systems to accurately summarize what you do and where you do it without any ambiguity.
Users still go through the same basic stages of awareness, consideration, evaluation, and action. What’s shifted is that AI tools are increasingly handling the middle stages (comparison, research, shortlisting) before a user ever visits a website. When someone does land on a page, they’re often already further along in their journey than their traffic source would suggest.
This isn’t entirely new. Zero-click search has been compressing informational traffic for years. But AI tools are doing it more completely and across a broader range of query types. The visit that used to come after reading three blog posts may now come after a single AI conversation.
What this means for marketers: Expect the volume of low-intent, top-of-funnel visits to continue softening. Optimize landing pages and conversion paths for users who arrive more informed and ready to act. If a page is still written for someone who knows nothing about the category, it may be losing visitors who already know plenty.
Organic click data was already an incomplete measure of search’s contribution. It didn’t account for brand searches, direct returns, or offline influence. AI widens that gap. When an AI summary drives consideration, and the eventual visit comes through a branded search or a direct URL, the organic channel gets no credit. The AI gets no credit either.
Rand Fishkin and others have been making this case for a while, and May’s data reinforces it. You’re most at risk of misreading search performance when you measure it through the narrowest possible lens.
What this means for marketers: Build reporting that connects organic visibility, direct traffic trends, branded search volume, and assisted conversions into a single view of discovery health. Any one of those metrics in isolation will tell an incomplete story.
AI systems can’t recommend what they can’t read. That’s the practical reality behind a consistent theme in this series: technical health is an eligibility requirement. Crawl errors, rendering issues, thin or duplicate location pages, and broken internal linking structures can prevent a brand from being surfaced in AI-generated responses.
For multi-location brands in particular, the stakes are high. An AI tool building a local recommendation needs to be able to understand, without ambiguity, what each location offers, where it is, and how it differs from others. If that information is inconsistent or hard to parse, the AI will route users elsewhere.
What this means for marketers: Treat technical SEO as infrastructure for AI inclusion. Prioritize crawlability, structured data, and content clarity on the pages that most directly represent your business, such as service pages, location pages, and key landing pages.
Google confirmed a bug in Search Console that inflated impression counts for many sites. For teams reporting impressions as a primary visibility metric, this led to confusing data. But the more interesting implication is what it reveals about the fragility of top-line visibility metrics in general.
Even without the bug, impressions measure exposure, not engagement, intent, or outcome. In a search environment where an impression can mean anything from a traditional blue-link result to an uncited AI summary, the metric was already losing clarity. The bug just made that more visible.
What this means for marketers: Use the moment to revisit what’s being reported upward. Impression volume as a headline number is a weak signal of business impact. Qualified traffic, conversion rate, and downstream engagement tell a more honest story about whether discovery is working.
The headline from May is that the signals marketers have been using to measure it are losing accuracy. Organic clicks, impressions, and ranking positions were always proxies for something more important: whether your brand is being discovered, considered, and chosen. AI is widening the gap between those proxies and the underlying reality.
Brands need to build visibility across the full discovery path and measure performance accordingly.
For marketers, that means:
The click comes later now. The focus is on making sure the impression happens at all.
May’s State of Search & AI unpacks why organic traffic is only part of the picture and what marketers need to measure — and build — differently to keep up.
Turn ancient archives into modern gold! Learn how updating old blog content for AI search improves SEO and gets your site cited with fresh facts and structure.
Discovery is fragmented across search, AI, and social. Here’s how to build a brand presence that gets found — and cited — no matter where customers look.
