How I think about SEO and AI search.

Most SEO consensus is consensus for good reason. But “best practice” is where the thinking starts, not where it ends. Some of it is right in general and wrong for your situation. Some was true two years ago and no longer applies. And some guidelines come from platforms (hi, Google!) that aren’t always forthright about how their own systems work, or whose advice suits their own agenda above your brand’s interests.

That’s why I read the source material and focus on what demonstrably works, rather than taking any of it at face value. It’s the only way to tell what actually matters for your brand from what everyone’s just repeating.

Generating traffic is actually pretty easy. Generating traffic that sticks and makes money is the tricky part. The quick-and-easy traffic plays rarely work for long — they can create some “line go up” charts that look nice in a pitch deck, but often those lines go back down a few months later.

Sometimes the best strategy for your business is the one that takes six months to start looking good in reports. That’s often the work that actually sustains over the long term.

If the measurement’s wrong, everything after it is wrong. Too much marketing runs on numbers nobody has ever vetted, or vanity metrics that never touch the bottom line. Clean tracking and honest baselines come before anything else.

But even good measurement is getting harder. Cookie deprecation, privacy changes, a search landscape splintering across a dozen new surfaces — the full-funnel picture we got used to is degrading, and some of it isn’t coming back. In a lot of ways, it’s a return to old-school marketing: fuzzier signals, more triangulation, reading things directionally instead of chasing decimal precision.

That’s fine, as long as everyone’s honest about it. I’ll tell you what we can actually measure, what’s a reasonable estimate, and when the numbers moved for reasons we can’t fully attribute. And if a quarter goes badly, you’ll hear that from me, too — a report that spins poor performance is simply wasting time that could be spent fixing the problem.

A lot of SEO tools and vendors provide “audits” that are just laundry lists of minor fixes that don’t add up to much. “All these title tags are too long! Look at all these 404s to fix!” That stuff’s worth fixing, but it’s table stakes, not strategy.

“Ten blog posts a month” isn’t a plan, either, unless everyone’s clear on why that’s the right play for your brand. (These days, it’s probably not.) Some of what gets recommended exists mainly because somebody needed to fill a retainer.

My job is telling you which handful of things will actually move the needle, which are nice-to-haves, and which you can safely ignore, even when that means less work for me.

I use AI tools daily for work and they’re genuinely useful. But they have real limitations and failure modes. They’ll hallucinate. They’ll fuse two correct pieces of information into one incorrect answer. (AI once informed me that James Brown wrote “Enter Sandman” — who knew?)

The trouble is, the hallucinations and stale advice arrive in the same confident tone as the accurate stuff. AI closes the knowledge gap fast, but it doesn’t close the judgment gap, and subject matter expertise is more important than ever. The risk isn’t that your team does nothing useful with AI; it’s doing something confidently wrong, at scale, that nobody catches until it bubbles up in reports.