Abstract 

Choosing an ASRS starts with understanding how your own operation behaves, not with comparing vendor spec sheets. Vendors publish throughput figures measured under conditions that will not match your operation. The gap between a rated number and a delivered one comes down to four things you can measure from an order file you already hold: the shape of your SKU velocity curve, how concentrated demand is in your peak hour, the bins your same-day cutoff consumes as work in progress, and the headcount each option needs to reach the same rate. 

This guide explains how to size an ASRS and what to evaluate before you commit, with a worked example running through every section. It closes with how each option holds up as your catalog and volume grow, and what a throughput guarantee should contain. Use it as a checklist to evaluate any automated storage and retrieval system against your own order profile. 

Terms used in the guide

Term	What it means
Line	One SKU picked for one order. An order with four different items is four lines.
LPH	Lines per hour. The rate at which an operation or a station completes picks.
Deep dig	In a stacked storage system, the work of lifting bins out of the way to reach one buried underneath.
Port or station	The position where a person actually picks. One person usually staffs one station.
Order bin	A container holding an order until it is released to pack.
Goods-to-person	Any system that brings inventory to a stationary picker instead of sending the picker to the inventory.
95th percentile hour	The hourly demand level your operation stays at or below 95 percent of the time.

What this guide covers 

Velocity: does your SKU curve make bin access a throughput problem? 

Cube-based systems stack bins, so retrieval time depends on how deep a bin sits. That works when a small set of fast movers drives most picks. When demand spreads across the catalog, deep retrievals become routine and the port’s rated line rate stops holding. The guide shows how to measure where your catalog sits, with a sensitivity curve you can read your own profile against. 

Peak: should you size on the average hour or the 95th percentile hour? 

Size on the 95th percentile hour. The daily average undersizes the system, and the busiest hour on record oversizes it for a spike that may not return. The guide shows how to build the distribution from your order history and how to check a vendor’s nominal station rate against its effective rate. 

Cutoff: how many bins does your same-day promise consume? 

Every order picked before cutoff needs a bin until it ships. Orders arriving after cutoff need one until the next shift. That work in progress can take a significant share of a system’s capacity before any inventory is stored. The guide shows how to simulate the buffer and two levers that reduce it. 

Cost: what does a throughput shortfall actually cost? 

Vendors publish throughput in whichever unit suits the architecture. The guide shows how to convert every quote to one unit using your own lines per order, then compare options by the headcount each needs to reach the same rate. System price is paid once. Labor at the station is paid every year. 

Checklist preview 

The full guide ends with a 12-point checklist you can answer from your own data before you talk to a vendor. Four of the questions: 

1. What share of your picks comes from the bottom 90 percent of your catalog? 
2. What share of your orders contains at least one of those items? 
3. What is your 95th percentile picks per hour, not your daily average? 
4. Have you converted every vendor quote to the same unit using your own lines per order?