Tom,
I'd go for the following:

Productivity:
Moves/warehouse worker
Order lines picked/(employee)/hour

Throughput:
Trucks (or cars, or containers)/hour (or day) in combination with Waiting times for trucks (cars/containers)

Lead times
In-time performance for picking, loading, truck departure, etc.
An indicator measuring speed to update inventory moves

Kind regards,
Robert Steenhof

posted August 24, 2007

Tom:

In any operation we default to the three most fundamental physical units of measurement. The rest are all converted measures to suit the P & L methodology. Whether it is 4, 5 or more is purely based on how one takes these and converts them into secondary measures that are really multicollinear in nature (Can be related to each other).

The three most critical metrics are related to:

(a) Length (Includes Distance, Area, and Volume)
(b) Mass (Throughput or material flowing in and out)
(c) Time (Cycle Time) The time taken for completing a cycle of activity

These three measures can be related to anything that the finance or business function needs in order to quantify their business objectives.

Have a nice day.

Shree

posted August 24, 2007

Tom,
I would manage using more than 3 metrics, but here are the top ones:
1) Inventory accuracy, including location accuracy; total $$$ adjusted upo and down; net $$$ accuracy;
2) Order defects - from the perspective of the customer - usually tracked as a parts per million;
3) Productivity, including transactions per hour; throughput; cycletime.

Be glad to discuss further.
Carl

posted August 24, 2007

Order fill rate (1-%stock outs)
Order Fill Lead time (Order receipt to ship confirmation)
Return Rate (lines returned/lines shipped)
On Time Delivery (to customer need date)
Cost per line shipped (payroll/lines)

posted August 25, 2007

Productivity of direct labor
Productivity of entire operations including support labor
% Overtime

Dock to Stock
Ontime Shipment

Order Accuracy

posted August 25, 2007

From a warehouse ops perspective, excluding inventory related ones like fill etc. I'd go for

Binning/Put away time (% binned in x hours after arrival)
Nil Picks
Minutes worked per line shipped (or lines shipped per total hour worked)
Shipping Service level - % shipped within target deadlines
Quality (measured both by random sampling internally and by customer claims)

Cheers

posted August 26, 2007

Tom,

Two metrics will tell you most of what you need to know operationally. together, they will usually drive high performance on the typical performance metrics (Inventory turns, Stock Outs, Obsolescence Write-Offs, and others).

To drive higher turns, lower write-offs, "leanness," and contain quality concerns escaping to customers: Inventory-Dollar-Days (you may substute other currency or time units to match organizational scale)---For each SKU, multiply the purchase price of each item by the number purchased and still on hand by the number of days the lot has been in stock and sum the totals by SKU. Bigger numbers are worse and lower numbers are better with zero as the ideal for perfect JiT performance.

To drive infrequent backorders, stock outs, production shut downs, lost sales, and the like: Throughput-Dollar-Days---Selling price of the order containing each delayed/lost delivery minus the purchase price of the materials (all SKUs) needed to satisfy the delivery multiplied by the number of days of delay caused by the missing SKU that caused the delay (or lost order [for lost business, a large, artificial time interval like 30 days or 90 days may sometimes be appropriate even if the SKU was actually unavailable for only a few hours]). Again, bigger numbers are worse, smaller numbers are better, and zero is the "never mised a promise date" ideal.

Pareto Lists on IDD and TDD will indicate potentially overstocked items (IDD) and potentially understocked items (TDD). Statistics on seasonally adjusted average daily (weekly, monthly, etc. according to how fast good move) sales, average time to replace sold stock, and standard deviations on both sales and replacement time will help define target inventory levels, reorder frequencies, reorder sizes, and shipping frequencies.

For help integrating these operational metrics with your processes, contact one of the three experts I've mentioned (or me). I could have recommended others, but LinkedIn cut me off before I was done.

:-)

Brian Potter

voice: 248/661-2556
e-mail: jmbpotter<at>sbcglobal<dot>net

Brian P also suggests these experts on this topic:

• mark W
• Rick D
• Dave S

posted August 29, 2007

The % of use (average number of used locations/ total number of locations) is an indicator not mentioned above , but quite interesting.
Usually when a warehouse has a % of use over its design % of use productivity goes down, broken items go up.
Besides you can open this indicator by kind of storage. For instance if you have a drive in secor and the % of use is way under 70% then you should think of moving some SKUs away from the drive in into another kind of storage let's say a selective storage.

posted August 30, 2007

Tom,
Interesting answers so far!

Number one would have to be service. And in this you have to account for claims, of course.
After that it's all about cost to serve. This might be split into warehouse and transport and expressed as pence per unit.
In an automated place, capacity utilisation would be important rather than productivity, but in a highly direct pick and replen operation keeping a lid on labour hours and agency costs is key (watch the argument for trading off labour versus service!)
Overheads will also feature large on the P&L - I've seen many good operations be flawed by the combination of crippling rates, insurance and maintenance (both of kit and building fabric).
Incidentally, if you have any third party contracts you ought to review these as a first port of call in taking out costs.

If you're after a quick and dirty 'list my DC's in order of performance', then the only caveat is to ensure that the cost to serve is measured correctly and fairly (for example, I've seen companies use such comparisons but use different 'sale' units in the calculation, much to the exasperation of DC managers involved)

posted August 27, 2007

Tom -

In addition to the metrics listed above - we closely monitored any request for a Hot Pick. That is, when any DC employee can not locate an item at a specified location (either due to lack of inventory or a lost item), they would mark the first location for audit and ask for the next pick location.

As an Exception to the pick process this metric and the underlying root-cause analysis will demostrate key areas for improvement. Simple issues like theft and spoilage, poor put-away, and lazy pickers can be spotted with basic analysis. Inventory shortages by vendor can be caught at this point if you do not run 100% QA at the inbound door.

As an example: our DC had manual picks on racks with five (5) shelf levels. The top jevel was just out of reach for many of our employees - and that level had a slightly higher "error" rate. Many pickers would return thier pick tour and the missing item would be assigned a Hot Pick to some other picker. QA would have to audit the location further adding to the cost.

Turning the statistic towards Pickers, we were able to see those that frequently missed picks. Allowing those that chose lower performance to "chose" to be unemployed. Seasonal employees were most notorious.

The ability to program a WMS to generate an efficient pick tour is destroyed when inventory is not in the expected location and in the expected quantity or when the picker undermines the process with laziness.

Root cause analysis takes some time, but once idendified the remedy can be enacted quickly an effectively.

~Tom.

posted August 27, 2007