What the work looks like.
One promo calendar, rebuilt. $2M in found revenue.
Promotions run on gut feel, with no baseline for what a good one looked like. Every promo scored against a benchmark, and a $2M lift in the gaps.
One product page, rebuilt. +28% conversion.
Traffic arrived on the Nude Mood Lip Kit page; too little converted. I rebuilt the page around how people actually shopped it, and CVR rose 28% over two months.
Who comes back, and what brings them.
Four years of customers, one clean cohort read: how many first-timers came back, and whether the hero product was the thing bringing them.
Which creative actually works? +73% CVR, answered.
Spend spread across paid-social ads on gut feel. I found the attributes that actually converted, using video, text, and image analysis, and lifted ad CVR 73%.
An unauthorized code, quietly discounting 136 orders.
A coupon that wasn't in Shopify kept discounting influencer kits. I traced it to the source, matched every use, and shut it off before it compounded.
Board-ready data, six hours faster.
Every board cycle, an executive lost hours assembling the numbers. I took the prep off her plate and handed back a board-ready pack, insight first, not raw tables.
The whole year, read in one place.
At year-end the exec team needed to look back and plan forward, but the data was scattered. I built a 360° read of the year they could actually set strategy from.
We forecast their biggest sale before it ran.
The birthday sale is the biggest revenue event on the calendar. We modeled it in advance, benchmarked it against three years of history, and handed over concrete moves to lift it.
A new CMO, fluent in the data in days.
A new CMO needed to get up to speed fast. I built the briefing, report, underlying data, and a working call, so she could start forming strategy on day one.
A wall of spreadsheets, replaced by one dashboard.
Weeks-of-stock lived in a stack of hand-updated Excel files. I built one dashboard the ops director refreshes weekly, every product, every warehouse, at a glance.
Heatmaps and session replay, live in a week.
The team was optimizing a store it couldn't watch. I stood up Microsoft Clarity site-wide, heatmaps, session replay, rage-click detection, in a week.
Raw survey responses, turned into decisions.
Survey data sat in a spreadsheet, too messy to act on. I cleaned, coded, and distilled it into a decision-ready read the team could actually scan.
Thousands of open-ended replies, turned into a barrier map.
Post-purchase survey answers were free text nobody could act on. Topic modeling plus LLM summarization turned them into a ranked read on what almost stopped the sale, and what drove high-LTV buyers.
A test backlog that ranks itself.
Experiment ideas came from everywhere and got prioritized by whoever argued hardest. I built a scoring framework that ranks every test on impact, effort, and how well the evidence backs it.
Analytics that answer the business, not the tool.
Before a line of tracking code: a measurement plan that maps every business objective to the KPIs and metrics that prove it, and flags what's tracked today versus what's in scope to build.
Eleven weeks. $1.4M of leaked revenue, recovered.
Six platforms saying six different things about CAC. One rebuilt attribution model. $1.4M they'd been writing off as channel overlap.
Sixteen weeks. Checkout completion plus twenty-four percent.
Twelve A/B tests run by an in-house team that didn't know how to size them. Four winners. One framework they can keep running without me.
Eleven dashboards in. One decision per week out.
A Series B SaaS team had eleven dashboards and one open question: 'where should we focus this week?' Eight weeks later, one screen. One question answered.
Sixty workflows audited. Three shipped. Eighty hours back.
A regional insurance broker wanted 'AI everywhere.' We audited sixty candidate workflows, shipped three, and reclaimed eighty hours a week, without firing anyone.
A leaky checkout at scale, sealed.
At high volume, a point of checkout friction is real money. A mobile-first checkout teardown and test program lifted completion double digits.
Turning a returns problem into a margin lever.
Returns were treated as a cost of doing business. A SKU-level returns model turned them into a merchandising signal, and protected real margin.
A size model that cut fit-based returns.
Most returns were fit, not taste. A size-recommendation model learned from purchase-and-return history nudged shoppers to the right size before they bought.
Churn, predicted before the cancel button.
Retention was managed after members left. A cohort-and-churn model surfaced the at-risk signals early enough to act, and lifted save rates on the members worth saving.
Subscription health, on one executive screen.
Subscription metrics were spread across tools and decks. One executive dashboard put MRR, churn, cohort retention, and box economics on a single screen leadership actually runs on.
A member lifecycle, redesigned with the founder.
Beyond the numbers: an advisory partnership with the founder to redesign the member lifecycle, onboarding, engagement, and the moments that decide whether a member stays.
A growth command center for a beauty DTC.
Sales, paid acquisition, and retention lived in separate tools and weekly screenshots. I built one growth command center that put executive overview, Meta acquisition, and LTV on a single screen the team runs the brand from.
An executive KPI command center for a DTC.
Leadership was flying on channel-level ROAS with no line of sight to profit. I built an executive KPI command center that ties revenue to contribution margin, channel efficiency, and unit economics on one screen.
Turning a community into a growth engine.
A loud, loyal community drove most of the growth, but nobody could measure it. I built a command center that quantifies referral and UGC-driven revenue next to product velocity and repeat behavior.
Reading every drop before it sells out.
Launches sold fast but planning was blind. I built a drop-performance command center that ties launch revenue to sell-through and unit economics, so restock and production decisions stop being guesswork.
A conversion lab that pays for itself.
Traffic was strong but the funnel leaked and baskets stayed small. I stood up a conversion and AOV lab: one funnel view, one basket view, and an experiment log where every test ships, gets measured, and gets a verdict.
Predicting the reorder before it lapses.
A replenishment brand lived or died on reorders, but planning was reactive. I built a model that forecasts reorders a cycle ahead and scores each subscriber for churn risk, wired into one intelligence screen.
Letting real shade demand plan the buy.
An inclusive range was a brand promise, but the buy plan did not match real demand. I built shade-level intelligence that ties sell-through by shade to acquisition and LTV, so inventory follows the customer.
I work with founders and teams who want clearer decisions, better systems, or sharper growth. One slot open for Q3 2026.