AI Rankings Aren’t Real

If you’re trying to “rank” in AI like it’s 2013 Google, you’re aiming at a moving target. Large language models don’t think; they predict words. Answers vary run-to-run, learn from a user’s history, and are stitched together from multiple behind-the-scenes sub-queries. That means a single query could fan out into several others before an answer is composed so what exactly are you “tracking”? On top of that, tools now peddle “AI visibility,” but citation sets rotate, sometimes within weeks. Screenshots aren’t strategy.

Personalization, Variability, and “Query Fan-Out”

Greg Gifford points out three dynamics that break the idea of AI ranking: personalization based on user behavior, built-in variability across attempts, and the way assistants decompose a question into multiple sub-queries you never see. Track “rank” on one prompt and you’re measuring an illusion. Use AI data for directional insight, not for KPIs.

SEO Fundamentals Still Matter

AI changed user behavior and answer assembly, not the fact that great content and trustworthy sites win. Gifford notes that what reliably correlates with citations is useful, fresh content plus a brand that’s present where assistants look. Research he cites shows that about half of AI-cited content is less than 13 weeks old and roughly three-quarters is less than 16 weeks old. In plain English: recent, high-quality material gets picked up more.

Fresh, Useful Content Beats Hacks

The web is flooded with “AI slop” unedited auto-generated articles that echo errors and clichés. Don’t add to the noise. Publish concise, human-edited answers to real buyer questions, and keep them current. Thin press-release spam or gimmicks like white-text-on-white-background may trick a model today and burn your domain tomorrow. Play long games with content people actually need.

Mentions > Links for AI, but Your Site Still Matters

Traditional Google algorithms care about links; AI systems care more that you’re mentioned widely in credible places, from industry sites to Reddit. That doesn’t replace your website it raises the bar for it. Your pages still supply the substance assistants summarize, and they still need clean architecture, fast load, and clear CTAs to convert the traffic that does arrive.

Revenue Beats Visibility

Traffic isn’t the trophy; revenue is. Marketers often celebrate traffic spikes or “visibility,” while owners ask the only question that matters: are we selling more? Gifford’s advice is blunt optimize for leads and sales, not vanity metrics. Expect more of your research-phase audience to stay inside AI assistants and arrive on your site later, ready to convert. That’s fine. Losing low-intent pageviews is not losing pipeline.

The Traffic Trap

It’s common to see traffic double while sales don’t budge. That’s not success; that’s misalignment. Tie content to clear intents, point every surface to a consistent CTA, and judge success by qualified conversations and closed revenue. If a tactic can’t be connected to those numbers, it’s a distraction.

What Dealers and Local Brands Should Do Next

1. Keep publishing substantive, up-to-date answers to buyer questions and service needs.
2. Show up where assistants look: credible industry sites and authentic communities. Mentions matter.
3. Standardize one CTA and align landing pages so you can trace story → session → signup → sale.
4. Experiment with AI tactics carefully; don’t pay twice for “AI SEO” that’s just SEO in a new wrapper.
5. Report like an owner: leads, close rates, revenue influenced not just rankings and reach.