I fix operational data, and the process that produced it, where the errors cost money.
I find where money leaves through your data and name the field it leaves from. Chargebacks traced to their cause, deductions ranked by what’s recoverable, launches modeled before they eat your cash. You get a remediation plan your team can run without me.
Then I fix the data and the process that let it leak, so the reports you run the business on start from records that are right.
- 25 years in operational data
- Fixed fees, no hourly billing
- Written intake, no call
- NDA, private upload, destruction certificate
Three kinds of data I work on
The method is the same in each. I fix the records, check the process that produced them, and put the gap in dollars.
Two ways to start, one of them free
The Deduction Scan finds what your retailers took and what’s recoverable. If nothing is recovered, there is no fee. The Snapshot puts your three most expensive data problems in dollars in seven business days, and its fee credits in full toward an audit.
I’ve spent twenty-five years in operational data, nineteen of them at a bootstrapped incentive-fulfillment company, seventeen as Director of Operations. The job there was checking that the data coming in was right before anything downstream relied on it. It is the same job here, in retail and deduction data, campaign and web analytics, and rebate and incentive programs.
In CPG & Specialty Food
The Problems I Take Off Your Desk
Behind every engagement: audit pipelines in Python and R, decision tools your team actually uses, and validation engineering that runs without me. The capability list is on the Services page. The point is what it does to your P&L.
The Quiet Cost
Nothing about this money is hidden. It leaves on every remittance, line by line, in amounts too small to argue with one at a time. A shortage on a pallet your warehouse counted twice. A label fee for a case that scanned fine on your dock. A promotion you ran in March, deducted again in June. Chargebacks like these run 5–15% of gross sales in specialty food, against net margins of 3–5%. The deductions can outweigh the profit and never once appear as a single number anyone has to sign off on.
That is the quiet part. The loud part is whoever does cash application, at month end, with a remittance PDF on one screen and the invoice on the other, trying to tie a Kroger deduction back to a shipment from March. The controller sees it a quarter later, as a clearing account that won’t close and a write-off to approve. The reports that would show the pattern are built from the same data that caused it, so the pattern shows up only to the person going line by line, and by then the dispute window on the oldest lines has closed.
Trade spend is the same story with more zeros. A brand puts 15–20% of revenue into promotions and asks, reasonably, which ones worked. The answer is split across three files that have never met: the plan, the deduction, and the scan. Cinderhaven, a brand I built from realistic test data to see where the money goes, runs a disciplined 11% all-in and still leaks, which shows where the leak is: in the reconciliation, not the spending.
Follow the leaks back and most of them end at twelve product-master fields that disagree across three systems, each consistent with itself. No single screen shows the disagreement, so it passes every check that looks at one screen at a time.
I work upstream of all of it, on the data layer between your records and the systems that trust them, so that what flows through is already right. The first look is free: twelve months of remittances in, a ranked list of what’s recoverable out.
Why your ship rate and Walmart’s don’t match
Cinderhaven’s own records say it ships 99% complete. Walmart’s scorecard says one of its shipments in six arrives late or short. Both numbers are computed correctly from the records each side keeps. Only one of them writes chargebacks against the remittance every week.
That gap is what unclean data and loose process look like on a P&L. Closing it means reconciling the two systems, finding which orders the retailer’s scorecard counts differently, and fixing the process that creates the difference, rather than arguing about whose number is right.
How I Work
Written Intake
A short form, no call. You describe the data problem; I tell you in writing whether it's a fit and what a scoped engagement looks like.
Frame the Problem
You and I settle the scope: what data, what systems, what question needs answering. I send a written proposal with the deliverable, the timeline, and a fixed fee.
Deliver
I do the work and hand you the deliverable: a written report, an interactive tool, a triage workbook, whatever the scope calls for. If the data needs ongoing attention, I'll propose a retainer. If not, you have what you need.
For a deduction engagement, that means a ranked list of every disputable deduction, the evidence to file each one, and the dollar value of what’s recoverable.
One dataset, written findings in seven business days
The Data Health Snapshot is the short version of an audit: send one dataset, get written findings with the annualized cost and a top-10 fix list. The fee is a fixed $950, credited toward a full audit within 60 days.
Start with a written form, not a call
Describe the data problem in the form. I’ll reply in writing with whether it’s a fit and where I’d look first.