210: AI AGENTS ARE HERE (AND THEY HATE YOUR WEBSITE) WITH JES SCHOLZ

The Metrics Trap

Marketing didn't neglect brand building over the past decade. It corrupted it. When every click became trackable and every conversion attributable, the entire discipline shifted toward what could be measured in 24-hour windows. Teams that once thought in years started thinking in days. The branding department got million-dollar budgets while SEO scraped by, and nobody questioned why.

The uncomfortable truth? Most potential customers aren't in market at any given moment. They don't want to visit your website. But ignoring them until they're ready to buy means competing on price at the moment of purchase, which is exactly why CPC costs spiraled upward. Brand salience requires consistent touchpoints over time, and that's precisely what short-term optimization undermined.

A New User With No Patience

There's a user on your website right now that doesn't see your carefully designed interface. It reads your code from top to bottom, starting with whatever sits in the header. If that header contains thousands of lines of navigation links designed to create internal linking opportunities, this user will scroll through all of it before reaching anything useful. Often, it gives up.

AI agents operating through tools like ChatGPT visit websites in text-based mode nearly half the time. No CSS. No JavaScript execution. Pure semantic HTML. That navigation mega-menu optimized for SEO? It's now a barrier. Those interstitial popups that humans reluctantly close? Agents abandon immediately. The cookie banner gets handled, but the email capture overlay causes instant departure.

These agents know they have thousands of alternatives. They feel no loyalty to any particular website. Every friction point becomes an exit point.

What Agents Actually See

Pull up your website in a text-based browser. Tools like Links, originally from 1995, reveal exactly what AI agents encounter. The experience is often shocking. Headers designed to maximize internal linking become walls of text. Semantic meaning disappears into div soup. Buttons built with span tags and Tailwind classes might render correctly for humans but confuse agents attempting to complete tasks.

The irony cuts deep: accessibility best practices that existed for decades, ignored because humans would "push through," now determine whether AI agents can use your site at all.

Beyond Your Domain

Entity optimization matters more than website optimization for AI visibility. Large language models don't just crawl your site. They synthesize information from knowledge graphs, Wikipedia, news sources, and community platforms. How your brand appears across these sources shapes how AI systems understand and recommend you.

The Reddit Temptation

Some marketers have noticed that LLMs reference Reddit discussions when making recommendations. The obvious hack? Seed Reddit with positive mentions. Create accounts. Post favorable comparisons. Upvote strategically. This approach creates a ticking time bomb. But what happens when those fake posts get exposed? Reddit communities are notoriously hostile to astroturfing, and the evidence lives forever in searchable archives. The short-term visibility gain converts into long-term reputation damage.

The Correction Ahead

Younger marketers have never worked without perfect attribution. Every campaign had a dashboard. Every decision had data. Learning to operate with uncertainty, to invest in brand building that won't show returns for months or years, requires a fundamental mindset shift. It won't be comfortable.

The good news? AI agents are forcing exactly the improvements that should have happened years ago. Cleaner code. Better accessibility. Reduced manipulation. Websites will actually improve because the tricks that worked on humans fail completely on bots. Sometimes progress requires a correction.