Notes on building the AI-native web.
Page 2 of 17
Writing a headline that survives a skeptical buyer
Most headlines are a category description with adjectives. Four tests that separate the ones that work from the ones that were approved.
The backup you have never restored
An untested backup is a belief, not a control. The restore drill takes an afternoon and is the only thing that turns one into the other.
Landing pages vs your main site
Separate landing pages win on campaigns and quietly build a second, worse website. Where the line belongs.
What an AI employee actually costs to run
The API bill is the smallest line. Four costs that don’t appear on any invoice and decide whether it pays back.
Do video and podcasts help you get cited?
Only in text form. Which is an argument for publishing the transcript, not for abandoning the medium.
What AI search does with your pricing page
“Contact us for pricing” reads to a model as no information. What that costs you, and what to publish instead.
Keeping an AI feature working for three years
Nothing about it stays still: the model, your data, your business, and the prompt. Four recurring commitments that keep it honest.
Why your best page isn’t the one getting cited
The page you are proudest of is often the least quotable thing you publish. Four reasons the model picked something else.
How to buy AI from a vendor
Every product has AI in it now, which makes the label useless. Six questions that reveal what you are actually purchasing.
Your traffic is down and your enquiries aren’t. What now?
The pattern most sites are seeing: fewer visits, similar business. It means your reporting is measuring the wrong thing.
Which trust signals actually work
A wall of logos is decoration. What buyers are checking is narrower and easier to satisfy than most sites assume.
How much of your data does an AI feature actually need?
Less than the vendor implied, and less than your first architecture assumed. Scoping the corpus is the real design work.