How an LA-Based Grocery Delivery Company Cut Failed Deliveries by 38%

Last year we worked with a grocery delivery operator running out of three hubs across the Los Angeles metro - Culver City, North Hollywood, and Alhambra. They were doing around 1,100 deliveries per week, with a fleet of 14 vehicles and a dispatch team of three. Their first-attempt failure rate was sitting at 18.4%. They knew it was high. They did not know exactly why.

Four months after switching their dispatch operations to DeliverLoop, that number was at 11.4%. A 38% reduction in failed first attempts. This is a breakdown of what actually changed.

Where the Failures Were Happening

The first thing we did was pull three months of delivery records and tag every failed attempt with a failure type. The breakdown was not what the operations manager expected:

• Customer not present: 44% of failures. Driver arrived, knocked, no answer.
• Late arrival outside committed window: 31% of failures. Customer gave up waiting and left.
• Access issue: 14% of failures. Apartment intercom failures, gated complexes, building security restrictions.
• Other (wrong address, order issue): 11%.

Three-quarters of their failures were either a notification problem or a time window problem. Nothing to do with drivers. Nothing to do with the product. Entirely a dispatch and communication issue.

Change 1: Advance Notifications With Real ETAs

Previously, customers got one notification when their order was confirmed - the night before. No day-of updates unless they checked the app manually. Most did not check.

We added two automated notifications: one at approximately 60 minutes before predicted arrival (based on current route progress), and one at 15 minutes out. Both were SMS, not push notifications only. SMS open rates for delivery alerts run around 94% - significantly higher than push.

The 60-minute notice alone accounted for an estimated 7-9 percentage point drop in "customer not present" failures. People who were out running errands came home. People who had forgotten about the delivery window made sure someone was present.

Change 2: Tighter Window Commitments Built at Routing Stage

The second issue was the gap between committed windows and actual arrival. Customers were being promised 2-hour windows that the routes structurally could not keep. Stops 20-28 in a 28-stop route were consistently arriving 40-70 minutes after the committed window end time, because the route was designed for ideal traffic, not actual Tuesday-morning LA traffic.

By building routes that account for realistic traffic patterns - using rolling 90-day traffic data by day-of-week and time-of-day - the committed windows became achievable. On-time rate for the final third of each route improved from 64% to 88%.

Change 3: Access Notes Integrated Into Driver Workflow

The access issue failures were tackled separately. We built a structured access notes field into the driver app that captures gate codes, intercom unit numbers, building security contact info, and any special instructions. Drivers who successfully complete a delivery can update the record in 10 seconds. The next driver assigned to that address sees the notes before they arrive.

Within six weeks, 73% of their recurring delivery addresses had populated access notes. Access failure rate at those addresses dropped by over 60%.

The Cost Math

Before the changes, with 1,100 weekly deliveries and an 18.4% failure rate, they were running approximately 202 failed first attempts per week. Each reattempt cost them roughly $9.80 fully loaded (driver time, fuel, scheduling overhead). That is about $1,980 per week in reattempt costs - over $100,000 annualized.

At 11.4%, failed attempts dropped to around 125 per week. Savings: approximately $750 per week, or just under $39,000 per year. Against a mid-tier DeliverLoop subscription, the payback period was 11 weeks.

What Did Not Change

The fleet size did not change. The number of drivers did not change. The dispatch team size did not change. Three people are still running dispatch across 14 vehicles. The operations manager told us the biggest difference he notices day-to-day is that his team spends less time fielding inbound calls from customers asking where their delivery is - because the automated notifications answer that question before customers have to ask.