Real-world examples.
I spent 4.5 years running marketing and supply chain for a 27-store RV network, and I got a lot of chances to make a real impact on the business. Here are three examples that show how I work.
The thing I care about most is finding the right problem to solve. AI and technology are just the tools — once you know the real problem, you usually need to build something to fix it, and that's where they come in. That's what these stories are really about.
These describe work at a former employer, so they focus on the approach more than the proprietary details, and I've left the company unnamed.
The inventory was always a step behind demand
Carrying used inventory was a heavy drag on the business's cash — and in a business this seasonal, every dollar tied up in the wrong unit at the wrong time works against you. We were always reacting too late: by the time anyone noticed we had too much, weeks of units were already in the pipeline heading our way. Demand could cut in half from one month to the next, so we'd end up overstocked with aging units, fire-sell them at crushed margins, then run short when the new season started. We lost money on both ends of the same cycle.
The analysis was the starting point — I worked out, month by month, what we actually needed on the ground, accounting for the real lag between deciding to buy and units landing. But an insight in a spreadsheet changes nothing. The real work was building it into the tools and processes used to make buying decisions every day — so a store could see exactly how much to carry and when, and know when to slow buying down and when to turn it back up. The timing logic stopped living in my analysis and started living in the daily workflow.
Getting the right inventory on the ground at the right time drove 23% growth in used revenue and a 54.5% improvement in margin — mostly by killing the annual fire sale and never running dry at the start of the season.
Most inventory businesses manage by looking at what's happening now. The money is in reading the signal early enough to act — and building that into the daily decision, not a report nobody opens.
The industry was measuring the wrong thing
The whole industry judged products by the same standard metric for sales velocity. It sounds reasonable, but it's misleading — it only reflects what already sold, and it flatters products you chronically under-stocked. Decisions were being made on a number that pointed the wrong way: telling us to celebrate products we were actually starving, and to overlook real demand we weren't meeting.
We developed a better way to read true demand. But the metric alone changes nothing — the real work was operationalizing it: using AI and our own tooling to fold that signal directly into buying and selling decisions, so the right read on demand showed up at the moment someone decided what to stock and how much. Once it was in the workflow, the gaps jumped out — steady demand we'd been quietly missing because we never carried enough.
We made deliberate bets on key product lines and stocked them heavier than anyone was comfortable with. One store manager called in February to say he had more of a unit on the ground than he'd sold the entire prior year. We told him to trust it. He sold through — every time. Across the product class where we rolled this out, sales grew 13.6%.
Almost every business runs on a metric everyone trusts that's quietly pointing the wrong way. The value isn't just spotting it — it's getting the better signal built into the decisions people make every day.
Reinventing how orders got made
Ordering inventory was slow, and the decisions behind it were often under-informed — not for lack of skill, but because the data was scattered across dozens of dashboards and spreadsheets and was genuinely hard to pull together. On top of that, ordering ran through a recurring review process between the stores and the inventory team. It kept the company's money safe, but it was a bottleneck: by the time a buy was discussed, justified, and approved, the order often went out weeks too late.
We reinvented the ordering process from the ground up. First, we consolidated the data — pulling all those scattered reports into one place so the full picture behind any buy was visible in seconds, not assembled by hand. Then we simplified the decision and the ordering flow itself, with the right checks built in rather than handled through meetings. It also surfaced inventory across all locations, so a store could pull a unit from another location instead of buying new.
A great weekend turned into an order the same day instead of weeks later — made with the full picture in front of you. And because the process scaled instead of the headcount, the business went from 16 stores to 27 with a smaller inventory team, and cost per store dropped by about half.
The win wasn't fancy technology — it was consolidating scattered data and collapsing a slow, multi-step process into one fast, well-informed decision. Good build work doesn't just automate a task; it changes what the business can do.
Your business has a constraint like these.
The question is where. That's exactly the conversation worth having.
Start a conversation