Blog

Thinking in public

AI CONSULTANCY · 8 MIN

The AI maturity model I actually use with clients

Most published AI maturity models have five tiers and are mostly useless for deciding what to do next. Here's the three-question version that actually is.

AI AUTOMATION · 9 MIN

Error handling in automations: the pattern that actually works

Most workflows are built for the happy path and patched for failures after they happen. The reverse order produces more reliable systems.

MEDIA BUYING · 9 MIN

Ad account structure: consolidation vs fragmentation

Too many campaigns starve the algorithm of data. Too few campaigns blend signals you actually need separated. Most accounts lean one way by accident.

WEB DEVELOPMENT · 9 MIN

The browser support decision nobody wants to make (and how to make it anyway)

"Support all browsers" isn't a real decision, it's an avoidance of one. Here's how to actually make the call.

SEO · 8 MIN

Why site speed and SEO are the same conversation now

These used to be two different specialists' problems. Treating them separately increasingly costs both rankings and revenue.

AI CONSULTANCY · 9 MIN

When 'we already have ChatGPT' is the wrong objection to an AI project

A general-purpose chat tool and a purpose-built workflow solve different problems. Conflating them kills projects that would have actually paid off.

AI AUTOMATION · 9 MIN

The automation that broke because nobody owned the API key

A surprising number of automation failures trace back to a credential, not a workflow bug. Ownership is a design decision most teams skip.

MEDIA BUYING · 9 MIN

The post-purchase survey data everyone collects and nobody uses

"How did you hear about us" surveys sit in a spreadsheet at most companies. They're one of the few genuinely incrementality-adjacent signals available.

AI AUTOMATION · 9 MIN

The automation you should kill, not fix

Not every failing workflow deserves another debugging session. Some should just be turned off.

MEDIA BUYING · 7 MIN

The creative brief template that actually produces testable ads

Most creative briefs produce one ad concept. A good brief produces several genuinely different hypotheses worth testing against each other.

Next step

Prefer the applied version?

A 30-minute discovery call. You bring the numbers, I'll bring the questions. If I'm not the right fit, I'll tell you who is.

Book a discovery call See the work