The Situation
Collectors Auto Supply didn’t start as a drop-ship business. It started as something much more familiar to anyone who’s run a physical-goods operation: a company with warehouse inventory, an eBay store, a website, and a sales team running outbound phone calls to move product.
That model worked — until it became clear it was the wrong model to scale.
The phone sales operation was high-touch and high-cost. Physical inventory tied up capital and created the usual carrying risks. The business had real revenue and real customers, but the operating model put a ceiling on how far it could grow without adding proportional headcount and overhead.
The decision was made to rebuild from the ground up. We wound down the phone sales operation entirely. Cleared the physical inventory. Rebuilt the sales channels around eBay and a direct e-commerce storefront. And shifted to a pure drop-ship model — no warehouse, no inventory on hand, every order fulfilled directly by suppliers to the customer.
On paper, that’s an operationally lean model. In practice, it created a new and very specific problem.
Orders came in. Suppliers needed POs. Invoices came back. Tracking numbers had to be extracted from those invoices and sent to customers. At 300–400 orders a month, the manual workload was already at its limit — and the business had clear runway to grow well beyond that.
The decision to automate didn’t come from crisis. It came from clarity.
We could see exactly where the ceiling was. Every additional order added more manual handling time, more room for error, more customer service follow-up. Hiring people to absorb that volume was the obvious move. It was also the wrong one — because the work itself shouldn’t have required a person in the first place.
So we automated first. Built the order processing, invoice extraction, and supplier communication workflows before the volume forced our hand. That’s what allowed us to scale from 300 orders a month to 800 without adding a single person to handle it. The team that was buried in order processing got their time back — and redirected it to customer service, returns, and the operational work that actually required human judgment.
That’s the sequence that matters: automate the repetitive work first, then scale into the capacity you’ve created.
The Diagnosis
Before building anything, we mapped the workflows and put real numbers on what manual processing was actually costing. The pattern was the same across every process: high frequency, repetitive tasks running on human attention that had no business requiring it.
Order Processing
Each order required manual validation, supplier routing, PO generation, and confirmation — averaging 5–7 minutes per order. At 300 orders a month, that was 25–35 hours of ops time on work that followed identical logic every single time. Manageable — but only just. And clearly unsustainable at the volume we were planning to reach.
Invoice Processing & Tracking Extraction
Every supplier shipment generated an invoice. Every invoice contained the tracking number that had to go back to the customer. Every supplier formatted their invoices differently — different layouts, different file types, different fields in different places.
The process: receive invoice → open file → locate tracking number → return to order management system → match to order → send tracking confirmation to customer. At 300 invoices a month, averaging 4–6 minutes per invoice, that was already 20–30 hours a month on a task with zero strategic value. At 800 invoices a month — where we were headed — that number would nearly triple.
Supplier Communications
PO generation and order routing ran through manual workflows with no systematic tracking of acknowledgments, shipment status, or delays. Supplier issues surfaced reactively — when a customer asked where their order was, not when the problem first occurred. At higher volume, that lag becomes a customer experience problem you can’t manage your way out of.
Total quantified waste at 300 orders/month: 50–75 hours per month.
Projected waste at 800 orders/month without intervention: 130–180 hours per month.
At a burdened labor cost of $25/hour, that forward projection was $3,250–$4,500/month — a number that made the build decision straightforward. The question wasn’t whether to automate. It was whether to do it now, while we still had the capacity to do it properly, or wait until the volume made it an emergency.
We didn’t wait.
