The Future of Business Websites in the AI Era

What follows for how you build, today

Current best practice Server-render what must be found. The no-JS-execution fact makes this non-negotiable for AI visibility, and Google’s own guidance already recommended it. Every mainstream platform does this by default; client-rendered applications are the exception that pays the price.

Current best practice Make business facts machine-legible. Structured data (Google Search Central, structured data documentation) and unambiguous on-page facts — who you are, what you do, where, for whom — serve every machine reader at once. Cheap, durable, documented.

Current best practice Decide your crawler policy deliberately. Allowing or blocking GPTBot, ClaudeBot, and peers is a real business decision with a real trade (representation vs. protection). Make it explicitly in robots.txt rather than inheriting a default.

Current best practice Invest in citable substance. AI answers are assembled from sources worth citing: specific expertise, first-hand detail, original information. The content that wins AI citations is the content that was already winning search — the era raises its premium and deepens the discount on commodity pages.

What the trendlines support — labeled as such

Reasonable prediction The website’s audience keeps broadening to machines. Every measured curve above points one direction; planning for human-plus-machine readership is trend-following, not clairvoyance.

Reasonable prediction Direct traffic undercounts value. As more answers resolve inside AI surfaces, some visits that would have happened convert to influenced-but-unattributed outcomes. Measurement will lag the behavior; businesses that only value what analytics attributes will underinvest in the asset doing invisible work.

Reasonable prediction Agentic browsing arrives unevenly. Vendor roadmaps all point toward assistants that research and act across sites. Machine-friendly structure and low-friction flows will matter more; the timeline and transaction depth remain genuinely uncertain.

Speculation Beyond that, honesty requires the label. Whether AI surfaces disintermediate whole categories of sites, whether attribution models emerge that repay publishers, whether the crawl-consent economy gets formalized — plausible futures, all unproven. We build for the facts and practices above and hold the speculation loosely; we recommend you do the same, especially when someone selling something presents tier-four claims wearing tier-one confidence.

Keep going