Musings
Things WPM is thinking about: digital, ecommerce and AI, written in the open.
Sixty Years of Marketing Research Says Growth Comes From Strangers, Not Loyalty
The Ehrenberg-Bass Institute's decades of research keep landing on the same uncomfortable point: brands grow by reaching people who barely know them, not by deepening relationships with people who already buy.
What the Research on Diverse Teams and Creative Work Actually Holds Up To Scrutiny
McKinsey's diversity research is widely cited and genuinely contested. The honest version of the story is more interesting than either side of the argument alone.
What Actually Predicts a Website Underperforming, According to Google's Own Case Studies
Real published case studies keep landing on the same two Core Web Vitals thresholds, and the conversion gains from clearing them are bigger than most redesigns manage.
AI Generated Images Tested Just As Trustworthy As Real Photography, and That Should Worry You a Little
Nielsen Norman Group's own recent study found no meaningful trust gap between AI generated and real stock photography, when readers didn't know which was which.
Why the Same Codebase Doesn't Mean the Same App on iOS and Android
A shared codebase in React Native or Flutter is not the same thing as identical behaviour across platforms, and the gap shows up in the parts that are hardest to test.
The Codebases We Inherit Now Come With a New Kind of Technical Debt
AI generated code is increasingly part of what gets handed over during a migration, and Stack Overflow's own data shows developers don't fully trust it either.
What Actually Predicts Whether an Ecommerce Rebuild Pays For Itself
The case studies with real numbers behind them don't point to a redesign. They point to load speed and checkout friction, which are much smaller problems to fix.
The Checkout Mistakes That Have Cost the Same 70 Percent for Fourteen Years
Baymard Institute has tracked cart abandonment for over a decade, and the average rate hasn't moved much from 70 percent. Most of what's driving it is still fixable.
Developers Are Using AI More and Trusting It Less, and That's Not a Contradiction
Stack Overflow's own 2025 survey found AI tool use among developers climbed to 84 percent while trust in the output dropped to 29 percent. That gap tells you more than either number alone.
Why "How Many Businesses Use AI" Depends Entirely on Who's Asking
Ask how many Australian small businesses use AI and you'll get an answer anywhere between 11 and 44 percent. Both figures are real. Neither is the number that actually matters.
Nothing in that category on this page. The full archive is at the link above.
Page 1 of 3 · 25+ articles archived
Let's build something that lasts.
Or just send the URL of the thing that is bothering you.