The work splits cleanly into two arcs. The first was a diagnostic story — "why is direct revenue collapsing, and can we even trust the comparison?" The second was a productization story — "we keep doing these one-off analyses; how do we make them recur themselves." Each gets its own case study, in two versions.
For EquipmentShare's internal use
Detailed walkthroughs with the real numbers, internal table names, and platform-side implementation. This is the working version EquipmentShare's team can take with them.
Trust the numbers — from decline diagnostic to recovery story
Direct Shopify revenue looked like a collapse. Before we could explain it, we had to separate real customers from internal branch / employee / phone-entry orders that were polluting the totals. Once we did, the story turned out to be a phone-channel collapse and recovery, plus an honest answer about whether the Jan 8 Shopify theme change hurt anything (yes — but not where it was supposed to).
From ad-hoc analysis to productized tooling
Four workstreams that started as one-shots and turned into things EquipmentShare can keep running: orders-by-reporting-location in BigQuery, an 80/20 SKU competitive-intelligence tool with a quarterly checklist, a documented Walmart channel evaluation (we recommended against deeper investment — context included), and an end-to-end inventory × sales pipeline wrapped in a one-command skill.
export__bigquery__report__order_lines_by_location
dbt model · competitive analyzer + battlecards + whitespace report · Walmart
executive summary + strategic context · April inventory pipeline + Claude skill
For external marketing — needs EquipmentShare approval
Story-first versions written for publication — Tromml's website, LinkedIn, and sales conversations. Specific dollar figures, percentage changes, internal table names, and platform implementation details are stripped. Quote callouts are placeholders that will be filled from the engagement review recording.
About this draft
- Audience: EquipmentShare leadership review. Numbers and dates are from the production work; happy to correct anything that doesn't match how the team remembers it.
- What's missing: the verbal-only stories — including a draft-orders-to-Klaviyo automation we built that didn't get adopted for opt-in reasons. We left those off the page for review-call conversation.
- Source code & data: all the work lives in
Tromml/dbt-pocunderad_hoc/equipmentshare/(analysis scripts) anddatamonster/datamonster_dbt/(platform models). Output CSVs, PDFs, and charts that were delivered live in the local DataMonster mirror.