Fractional Data Executive · Advisory
Compounding the edge —
with data science
and agentic systems.
The fractional practice of Gavan Acton. Fifteen years shipping ML, data, and agentic systems against EBITDA, at scale — pointed at whichever process in your business hides the next quarter’s margin.
On the name
stivot /ˈstī.vət/ · verb
A ski-racing term: to transfer weight mid-turn and take the line a course actually gives — when the terrain shapes the run, or the gate comes faster than the plan. Turn by turn, the saved tenths compound into a large margin at the finish.
The best companies work the same way — stepping off the line they planned, attacking the line the market actually gives. They redirect without losing velocity. Process by process, the margin stacks, the wins multiply.
The ninth wonder
Compound interest, the saying goes, is the eighth wonder of the world. Call it a ninth when machine learning, data science, and agentic systems get pointed at a single business process.
A churn model that reclaims a percent of revenue this quarter does so next quarter, and the one after that. An agent that shaves minutes from an underwriting decision does it on every decision that follows. A pricing experiment that lifts margin by a hundred basis points compounds across every transaction the business will ever run.
Margin isn’t earned in heroic quarters. It’s stacked — one well-aimed process at a time. The work is picking the right process, pointing the right toolkit at it, and knowing when to stack the next.
Why engagements stall
Most “AI initiatives” don’t fail.
They diffuse.
“Let’s use AI” is a mandate, not a mission — it produces pilots that never ship, POCs that never generalize, and spend that never compounds. The failure mode isn’t a bad model; it’s a vague one.
The fix is ruthless specificity: a single operational lever, a measurable ROI target, a credible first result inside one quarter, not one fiscal year. Not a strategy deck. A shipped model, a live agent, a number on the board. Specificity is the whole game.
Practice
Three levers,
one thesis.
Engagements are selective and senior. Every one comes back to the same three moves — the anatomy of a modern stivot, shaped to the gates in front of you, not a packaged offering.
- 01
Machine Learning & Data Science for Margin
Machine learning applied to the operational levers that move EBITDA — pricing, retention, fraud, forecasting, attribution — paired with the data science rigor that measures each gain for signal and keeps each risk bounded. Small, well-aimed improvements designed to compound across the lifetime of the business.
- 02
Agentic Systems & AI Workflows
LLM- and agent-based decision support that compresses decision latency, multiplies expert leverage, and removes operational drag from day-to-day work. Scoped to your actual workflow and bounded by the measurement that catches its wrong answers early — not a vendor demo, not a hype cycle.
- 03
Foundation & Fractional Leadership
The foundation behind reliable models and trustworthy agents: warehouse, governance, pipelines, the knowledge base an agent can actually ground in. Most stalled initiatives blame the model; the real culprit is the layer below it, and half the work of any real initiative is quietly rebuilding that layer. Plus the fractional executive seat that carries the rest — forecasting, revenue operations, M&A diligence, the call on which lever to point at which process, and when.
Track record
Built.
Grown.
Scaled.
An AI programmer turned data leader. Wrote AI in C++ for AAA games at Digital Extremes, eventually leading a Halo 4 programming team. Then pivoted into data — where the work would compound. Recruited as the company’s first analyst. Built the warehouse, the BI platform, and a seventeen-person team running RevOps, experimentation, and forecasting — the commercial engine behind Warframe and its eighty-plus million players. Led the data story across three M&A events. Left as Director of Data, Insights & Revenue Operations.
What that adds up to is pattern recognition. I arrive knowing which levers move EBITDA in your kind of business, where these initiatives fail, and what a first real win looks like — so the conversation starts at “which process, by when,” not “what could we try.”
- 15+
- years architecting data and revenue engines
- 80M+
- customers served by the data org I built
- 3
- M&A events as the data and revenue lead
Contact
Ready to stivot?
Let’s talk.
Available for selective fractional and advisory engagements. The fastest path is a short note describing the decision in front of you.