Ask most fleet operators what it costs when a driver quits. They'll say something like "a few hundred bucks in job postings." They're off by about 97%.
Replacing a single last-mile driver — when you account for recruiting, onboarding, training, and the productivity hole they leave — runs between $4,000 and $10,000. For a fleet with 20 drivers and 60% annual turnover, that's $48,000 to $120,000 walking out the door every year. Most of it invisible on the P&L.
The productivity gap makes it worse. New drivers are 15 to 20% less efficient for their first 90 days. They're slower on routes they don't know, rougher on equipment they haven't learned, more likely to generate exceptions that cost you service penalties. A driver who stays three years amortizes their onboarding cost across thousands of trips. A driver who leaves in month two is a pure loss.
Here's what most operators get wrong: they assume the fix is pay. Bureau of Labor Statistics (BLS) data confirms delivery wages have been rising across the board — but turnover rates haven't dropped proportionally. Because pay isn't the primary reason drivers leave. Predictability is.
The operators who've figured out retention aren't paying the most — they're running the most predictable operations. Predictability is free. The cost of ignoring it isn't.
Frequently Asked Questions
What does it actually cost to replace a DSP driver? Industry estimates range from $3,500 to $10,000 per driver replacement event, accounting for recruiting costs, onboarding time, training hours, and the productivity gap during the 3–4 week ramp period when new drivers complete 60–70% of a full stop count.
What is the average driver turnover rate for Amazon DSPs? Industry-wide DSP driver turnover runs 80–100% annually, meaning the average 20-van fleet replaces most of its driver roster every 12 months. High-performing operators achieve 40–60% annual turnover through retention programs and equipment reliability.
What actually reduces driver turnover in last-mile delivery? The most consistent predictor is operational predictability — drivers knowing their schedule in advance, equipment that works reliably, and consistent dispatch behavior. Pay increases alone have not moved the needle in documented DSP case studies.