Cases
What needs to change, what needs to hold, and why.
Find the decision closest to yours. These first-person accounts explain the problem, the choices I made, and what happened next.
-
WordPress payment plugin audit before refactor
A payment plugin was missing charges and charging customers twice. The next refactor depended on understanding how money actually moved.
- The decision
- Describe the payment behavior and risks before committing to a gateway and data-model refactor.
- The outcome
- The discovery pass surfaced 28 production bugs and gave the client a documented basis for the next modernization phase.
- Reading the evidence
- The $3.4M figure is annual payment volume through Stripe, one of three gateways, at the time described. The 28 findings belong to the discovery pass; they do not establish that the rewrite is complete.
-
The founders needed a subscription platform in a market where a competitor was already live. The launch also had to support years of operation.
- The decision
- Build a subscription platform that could compete at launch and remain operable as the business grew.
- The outcome
- Paid subscribers rose 20% within 12 months of launch. The platform crossed 500,000 monthly users during 4.5 years of engineering ownership.
- Reading the evidence
- The engagement ran from February 2017 to September 2021. These are outcomes reported in this first-person case; raw analytics and the uptime measurement method are not published here.
-
AI-assisted PowerPoint brand QA case study
An AI-assisted slide review flagged so much noise that the reviewer could not trust the findings. Each issue needed evidence the reviewer could inspect.
- The decision
- Make a PowerPoint review workflow useful by separating deterministic brand checks from evidence-gated AI judgment.
- The outcome
- Findings fell from 110 to 31 on the annotated sample deck, while retaining 20 of 22 detectable reviewer-noted issues.
- Reading the evidence
- Recall and precision were evaluated against reviewer speaker notes on the annotated sample. Two decks of 24 and 33 slides ran end to end on May 18, 2026; this is not a general accuracy benchmark.
-
Pragmatic product trade-off under deadline pressure
An enterprise buyer needed an add-on purchase flow the booking product did not support. A separate checkout would have pulled the team away from its roadmap.
- The decision
- Meet an enterprise add-on purchase requirement while preserving the existing booking and payment model.
- The outcome
- A working flow was ready within one week. The client signed, and the flow later handled thousands of transactions.
- Reading the evidence
- The week describes this implementation window. The later transaction volume is qualitative; an exact count and measurement period are not published here.
Facing a similar decision?
Tell me what your system does, what needs to change, and what makes the decision difficult.
