AI search makes traffic a weaker KPI

TL;DR: AI search makes raw traffic less reliable as a success metric, so teams need to measure business impact before they rewrite the whole content strategy.

Is the traffic decline actually hurting the business?

Search Engine Journal’s “The State Of Search In An AI World: What Marketing Leaders Need To Know Now” points to its State of Search 2027 and asks the question most dashboards still avoid: is a traffic decline actually hurting your business?

That is the right starting point.

A lot of AI search panic treats every lost click as equal. It is not. A lost visit from a vague informational query is different from a lost visit from someone comparing vendors, checking pricing, or looking for implementation help. One is often top-of-funnel noise. The other can be pipeline.

The hard part is that most SEO reporting still rolls those together. Sessions, impressions, click-through rate, average position. Useful, but blunt. In an AI answer world, some queries will produce fewer site visits because the answer is satisfied on the results page or inside an assistant. That does not automatically mean the brand lost. If the brand is cited, remembered, searched later by name, or included in a buying shortlist, the path just got harder to attribute.

That is uncomfortable for marketers because it weakens the clean story SEO has told for years: rank, get clicks, convert. The new version is messier: be present in the answer layer, earn enough trust to be named, then capture demand when the user becomes specific.

Ashe runs Lucky Domains, which works on SEO and search visibility, and this is exactly where the measurement problem shows up in real site work: traffic can fall while the leads that remain get better, or traffic can hold while commercial intent quietly leaks away.

What should marketers measure instead?

I would not throw away traffic. I would demote it.

The better question is whether organic visibility is still producing qualified outcomes. That means separating pages and queries by job. Educational content should be judged partly by visibility, citations, branded search lift, email capture, and assisted conversions. Commercial pages should be judged by qualified form fills, calls, trials, demo requests, quote requests, and sales conversations.

This is where AI search forces better discipline. If a page exists only to answer a simple definitional query, AI systems are good at compressing that value into one paragraph. That page may still matter as proof of expertise, but it probably should not be the center of the strategy.

The durable work is different. Original examples. Comparison pages with real tradeoffs. Calculators. Local or niche data. First-party experience. Screenshots. Implementation notes. Strong product and service pages. Clear authorship. Pages that answer the next three questions, not just the first one.

The catch is that many teams will respond to AI search by publishing more generic content. That is backward. If the answer can be generated from ten existing pages, publishing the eleventh page is not a strategy. It is inventory.

Where should the investment go now?

Search Engine Journal’s framing, “where to invest next,” is the useful part because it shifts the conversation from anxiety to allocation.

My bias: invest in assets that make your company harder to summarize without naming you. That might be proprietary data, expert commentary, tools, templates, customer stories, category definitions, or content tied directly to a product workflow. Not because AI cannot ingest those things, but because they create reasons for users and systems to treat your site as a primary reference rather than a rewrite of everyone else.

Also invest in brand search. This sounds boring, but AI discovery often increases the value of being the known option. If a user sees your name in an answer, hears it in a podcast, finds it in a comparison, then searches directly, that branded query may be the conversion event your generic keyword report never credits.

Practitioner’s take: pull your top organic pages into three groups this week: pages that drive revenue, pages that create qualified assists, and pages that only drive volume. Protect and improve the first group, add stronger next-step paths to the second, and be ruthless with the third. The catch most teams miss is that AI search does not just reduce clicks. It exposes which clicks were never worth much.