SEO is SO Old (Do What AI Can’t)

I did some thinking about SEO with my team this morning, and I wrote this:

People are using AI for search because of the better results given – compared to a list of links to go try (as I write that, seeing a page of links seems kind of old and silly already). There’s just a new audience/tool in the middle now. Instead of a list of pages to try, something more personalized is doing the work. Websites need to reach that tool and be optimized for that tool now.

But isn’t that true? How long have we put up with search engines like Google giving us a list of web pages to go try? That’s what search engines have been – for a long, long time. Search engines and AI have combined, in a way. Google has had AI results for a while, and now you can dive deeper.

You’re Living in the Past Right Now

With each generation, what happens is, you get reactions like, “You used to have to do… what?”

Yeah, growing up, there were like 4 TV channels – that’s it. If you had a show you wanted to see, you had to be in front of the TV at a specific time to see it, or you’d basically never see that episode again – maybe 20+ years later when they started making channels showing re-runs. Or, when the VCR finally came out, you could try to record something, but even then you’d run into issues (it didn’t happen for some unknown reason, there wasn’t enough room on the tape, etc.).

I had the thought about swimwear and Facebook this morning. Yeah, girls today don’t want to wear the same swimsuits their mothers wore, so that has changed. Kids don’t want to use the same social platforms their parents used, so they don’t get on Facebook. They have their own style and versions of things they’ve attached to. It was the same when I was a kid – we wanted something different than our parents had, too. We wanted to progress and be what we thought was better – more evolved, if you will.

“You used to have to type something in a box and then go try to find the answer yourself?”

Right now, or as of a year ago, you had to type something into a box and just hope that you got something that maybe gave you the right answer. Yeah, sounds silly, right? And worse yet, there’s a whole industry out there of people getting paid to make sure certain results get shown – the SEO industry.

Is AI Really Taking Over Search?

Yes and no. Like I told my team, it’s just a different tool – or interface. It’s more personal. Instead of a page of links to websites that might have what you need, you’re being told and you’re given explanations. It’s better – like, way better. That’s the interface now. You still need to do SEO – to help AI find your website.

And that’s the bottom line – you need to help AI find and recommend your website. That’s mostly going to come down to content. The focus now is…

“Do What AI Can’t”

Can AI experience something? Can it do original research? No – not right now anyway.

Can it take a few resources and combine them into something new? Kind of, but not really.

AI can tell you what it has found, like what other people think. That’s really what it’s doing. It can tell you what people think of a restaurant, but it can’t go there for the experience – the feel of the place, the staff, the taste of the food, and the service they received. Nope. That’s a human-only thing. AI can’t do that.

Using AI to write content that AI can already produce almost doesn’t make sense to do. It does in the respect that if you build good content on a website, you look like an authority on the topic. But, what will make your website stand out is doing what AI can’t.

Examples of What AI Can’t Do

That is, besides try to put together something from Ikea…

Field Experiments and Hands-On Tests

• Build-it logs: Document a project with dated steps, failures, and fixes. Include photos, commits, and changelogs.
• Before/after trials: Show speed, UX, or conversion lifts with exact settings, timing, and raw numbers.
• Stress tests: “What breaks first?” testing with measured thresholds and video proof.

First-Party Data and Benchmarks

• Original benchmarks: Repeatable test rigs with published methodology and sample data downloads.
• Price/availability trackers: Time-series charts you collected (not copied).
• User cohort stats: Aggregated, anonymized insights from your own customers or subscribers.

Access-Based Content

• Expert interviews you conducted: Publish audio, transcript, and your takeaways.
• Factory/shop/facility tours: Your photos, process notes, and staff quotes.
• Shadowing a pro: Day-in-the-life documentation with timestamps.

Local and Physical-World Coverage

• On-the-ground guides: You drove the route or visited the site. Include your photos and practical notes.
• Photo surveys: “Every example on Main St.” posts with a map and attribution.
• Regulatory walk-throughs: Call the clerk, pull the form, and show exactly how to do it.

Original Tools and Calculators

• Calculators/checkers you built: Publish inputs, formulas, and edge cases.
• Datasets + mini-APIs: Let others reuse your data. Provide a CSV and schema.
• Embeddable utilities: Widgets with a public changelog and usage stats.

Investigations and Audits

• Mystery-shopper audits: Document calls, emails, and outcomes. Redact PII, keep receipts.
• Dark-pattern spotlights: Screenshots, flows, and proposed fixes.
• FOIA/public-records pulls: Source PDFs, your extraction sheet, and summary findings.

Opinionated Takes with Skin in the Game

• Predictions with scorecards: State predictions, set review dates, publish hit rate later.
• Unpopular opinions backed by tests: Take a stance, show data, invite rebuttals.
• Design/code critiques in the wild: Annotated screenshots with specific suggestions.

Community-Driven Pieces

• Reader challenges: “Try X for 30 days,” collect results, publish a roundup.
• Panels and AMAs: Curate multiple expert answers to one hard question.
• Case-study swaps: Feature readers’ setups with hard numbers and configs.

Proprietary Lists and Directories

• Curated, verified lists: Validate each entry by phone or email; note verification dates.
• Scorecards: Create a rubric, publish it, and rate consistently.
• Living docs: Show diffs over time with update notes.

Time-Bound, On-Scene Reporting

• Live notes from events: Timestamped takeaways, your photos, and links to slides.
• Same-day reg change explainers: Read the rule, call an official, translate to actions.
• Supply-chain or outage diaries: What happened today, who said what, next steps.

Comparative “Taste Tests”

• Blind comparisons: Same task, multiple tools/services, measured outputs.
• Bitrate/ad-load tests: Measure minutes of ads per hour and show clips.
• Theme/plugin bake-offs: Fresh installs, Lighthouse runs, and CLS logs.

Build-in-Public and Transparency Reports

• Monthly numbers: Traffic, revenue, costs, what changed, what you’ll try next.
• Post-mortems: What failed, why, and what you changed.
• Roadmaps with votes: Publish priorities and invite user votes—report what shipped.

Negative Results and Dead Ends

• What didn’t work: Tactics you tried, time spent, and why you’re shelving them.
• False positives: Myths you tested and debunked with data.
• Opportunity cost diaries: What you chose not to do and the logic.

Original Visuals and Artifacts

• Annotated photos and diagrams: Your marks, your notes, include source files.
• Screen recordings of real workflows: No stock B-roll—cursor, keyboard, mistakes.
• Scans of receipts or sketches: Redacted but real; shows you were there.

SOPs and Checklists You Actually Use

• Internal playbooks: Step-by-step with owners, tools, and timing.
• Incident runbooks: Exact steps you follow when something breaks.
• Printable field checklists: One-page, no-nonsense, field-tested.

Longitudinal Follow-Ups

• 6- and 12-month check-ins: Re-measure and report decay or compounding effects.
• Cohort comparisons: What changed for groups that adopted vs. didn’t.
• Maintenance costs over time: Not just install—track upkeep.

Proof Signals to Add to Any Piece

• Methodology section: Assumptions, tools, dates, versions, test rigs.
• Raw files: CSVs, screenshots, audio, and redacted invoices.
• Timestamps and locations: EXIF-intact photos, signed notes, repo commit hashes.
• Named sources: With permission and role (not “a source says”).
• Change log: What you updated and when.

Helpful Schema Types

• Dataset, HowTo, Review, Product, SoftwareApplication, QAPage, Event, FAQPage — choose the one that matches the primary intent and link to your methodology.

Quick Ideas Tailored to My Properties

RankCheck Pro

• Monthly SERP volatility index: Publish from your data with method notes.
• Theme speed bake-off: Standardized test harness and repo.
• Cold-start SEO experiments: Full logs, timelines, and outcomes.

ServiceAreaMaps

• Case studies: Lead changes after adjusting radius or copy.
• County-level coverage guides: Screenshots and local notes.
• Embed performance: Hits, CTR, and time-on-page deltas.

CUE Streaming

• Ad-load timing studies: Minutes of ads per hour by channel.
• Bitrate comparisons: Peak vs. off-peak quality measurements.
• Support response audits: Mystery-shop transcripts and outcomes.

Webstix

• Build-in-public plugin series: Weekly diffs with Git tags.
• Client change logs: “What we changed and why,” with before/after data.
• Live teardown (with permission): Annotated local site reviews.

The Bottom, Bottom Line

I just used AI to come up with a list of things AI can’t do – how crazy is that?

So, to conclude this, first make sure you concentrate on posting content on your website focusing on what AI can’t do. I just gave you a whole list of that. Doing that will set your website apart. That means not only people will notice, but AI itself will notice, and when you stand out from the crowd, you get the referral.

But, you can still AI to do things like:

• Polish your content – check spelling, grammar, content style.
• Ultra-polish your content – act like a professional copywriter, crafting a message in a way that will get you more conversions.
• Help build authority on your website so you look like the best source to choose.
• Take care of the grunt work. For example, create relevant images, or come up with TL;DR summaries – that kind of thing.
• Help you do research that you add to your own research. I just did that in this article.

AI is new, it’s cool, it’s fun, and we’re enjoying it right now, but really, it’s a tool. We’re still trying to figure out ways to use this tool, but we need to keep common sense in mind and not go overboard. A phrase I always think about is, “Good Things Take Time” and that’s the cold, honest truth. Quick things can be good, but they tend to not last the test of time. It’s OK to use tools that give you things quickly, but in the end, people tend to respect handcrafted items.

📄 Download a PDF of This Article