Reviews
What people wrote after the readout
We keep the mild reservations in. A studio that only collects praise is marketing, and we already have a journal for that tone.
Module four’s QA scripts are now the first comment I leave on a tracking pull request. Engineering stopped asking me what “success” meant for the save event.
Amelia Hart, staff product designer
★★★★☆
Solid on naming. The Identity, Sessions, and Deduping clinics assumed more warehouse access than our contractor setup had. I still finished, but week two was awkward.
via cohort feedback form · March intake
Cardiff. Sat Taxonomy Studio while we rebuilt onboarding. The retirement dates on old funnel events were the unglamorous win; finance stopped quoting a metric we had unofficially killed.
Rhys · Cardiff
Client in broadcast media. We wanted Helen to bless a 200-event dump. She refused and made us pick twelve. That annoyed the growth lead for a week and then became policy.
Anonymous, UK broadcaster
Case study: a marketplace checkout that lied about “placed”
A GB marketplace arrived with three events that all claimed an order was placed: one on button press, one on 202 from payments, one on email dispatch. Board packs used whichever looked healthiest that month.
In Event Taxonomy Studio they collapsed the trio into order_confirmed with a trigger on the payment success callback and a property for attempt count. Button presses moved to a separate checkout_submitted that product could still watch without polluting revenue.
Ninety days later they reported a 19% drop in “orders” on the old dashboard and a matching rise in finance’s actual settlement count. The drop was the point. They still argue about refunds, which the studio never claimed to solve.
Case study: a public-body form with twelve screen views
A local-authority digital team in Wales sat Screen and Gesture Instrumentation after a supplier left them with a view event on every partial scroll. Storage costs were the complaint; trust in completion rates was the real one.
They kept one form_step_viewed with a step index, dropped pixel-depth events, and added a single form_abandoned with last completed field. Completion reporting moved closer to what case workers observed on the phones. The limitation they documented: they still cannot see why someone stopped without a follow-up interview, and we did not pretend otherwise.