I turn messy operations into systems that just work.

Churn diagnosis and recovery system

Sparkle Stories had a rising churn problem. Before touching the data, I researched subscription pricing principles from Recurly, ProfitWell, Chargebee, and Simon-Kucher, so the recommendation would rest on what actually works rather than instinct.

I rebuilt the model from fifteen months of raw Stripe payment data, then split the real churn into two separate problems: about 29 percent was failed or expired cards, a billing issue with no discount required to fix it, and about 71 percent was members genuinely choosing to leave.

I built the financial model and a decision deck that weighed each option leadership was considering, with numbers behind every verdict, then implemented the fix directly in Stripe: smart retries, a card updater, and expiring-card emails to recover the involuntary churn automatically.

On top of that, I built a personalized automation layer in n8n that watches for members showing signs of trouble and routes them differently depending on value: a direct, personal outreach nudge for high-value members, and an automated win-back sequence in SendGrid for anyone who fully cancels.

Churn fell for four straight weeks after the fix, from 6.2 percent down to a new low of 4.9 percent. The model sized the total opportunity at close to $35,000 a year, with member lifetime value projected to nearly double.

Ongoing research, migration strategy, and automation

Beyond the churn project, this has been a running relationship built on the same skill applied to different problems.

Access control and payment system

Jen needed a way to sell access to an internal tool without handing it over openly. I built a full web app from scratch, Stripe for payment, Google Sheets as the database, Apps Script and n8n running the logic underneath. When someone pays, the system creates their account, sends credentials automatically, and checks their status, active, expired, or unpaid, every time they log in. I later adapted the same system for a second product, so it was built reusable rather than one-off.

Building it lean instead of with custom infrastructure saved an estimated $5,000–$10,000 in development cost.

Data entry and sheet protection interface

The team was entering data directly into the management spreadsheet, which was causing frequent errors. I built a clean web interface with controlled input forms, dropdowns synced to the sheet's own validation, required fields, that write to the sheet automatically through Apps Script. The team gets a foolproof front end, the sheet stays the single source of truth underneath.

Data accuracy is now around 96 percent, and the team saves about 20 hours a week.

Revenue analyst and operations

I managed pricing across a large vacation rental portfolio using Wheelhouse, KeyData, and PriceLabs. I found that 23 units at one property existed only as individual listings rather than a consolidated building group, which was masking real performance problems: occupancy well below the market benchmark, minimum stays that were too restrictive, and no forward bookings past a certain point. I consolidated the group and built a Wheelhouse optimization strategy around it. I also served as the team's default technical resource for the internal revenue dashboard, since the other analysts focused purely on pricing.

The optimization strategy projected a 25–35 percent revenue increase for the consolidated units.

Product manager, AI outreach

TempShift builds an AI outreach product with one clear mandate: make automated outreach feel human, not templated. As product manager, I drove product strategy and worked across engineering, design, and marketing, while also building and managing the underlying n8n automations myself, more hands-on than most product roles. This is where I built most of my depth in prospecting, lead management, and campaign design, the instincts that still shape how I run my own outreach today.