Warehouse data: what to measure

On many warehouse managers' computers there are dozens of reports: balances exported from 1C, a forty-column Excel sheet, charts refreshed every week. Yet at the end of the month one question remains: is the warehouse working well or not? The problem is not a lack of data but a lack of clarity about what to look at. Trying to measure everything is the surest way to end up measuring nothing.

One filter question

Ask every metric the same question: if this number changes, what exactly will I change? If the answer is "I don't know", the metric is not needed yet.

Say the household-goods warehouse of a wholesale depot tracks "number of dispatches" every day. The figure went up — is that good? Perhaps there are more customers, or perhaps one order was simply split in two. No decision follows. But when "share of orders dispatched late" rises, both the cause and the person responsible are clear: you check receiving, picking or the supply link. Keep numbers like that one.

Four groups to start with

For most warehouses these four are enough:

• Stock accuracy — how closely the recorded balance matches what is on the shelf. It is the foundation of every other figure: if the balance is wrong, the rest of the calculations will be wrong too.
• Turnover and days in stock — how many times stock is replaced in a year and how many days an item sits in the warehouse. These show where cash is tied up.
• Order completeness (fill rate) — what part of the quantity a customer asked for was actually supplied. It measures service quality.
• Losses — shrinkage, damaged goods, expired and spoiled items, in so'm and in percent.

Together they give a full picture: whether the data is right, where the money sits, what the customer gets and what is lost on the way. You do not need complicated software to calculate them: balances, sales and receipt and issue documents are enough, and a plain export from 1C or Excel will do.

Every metric needs an owner, a period and a threshold

A number changes nothing by itself; someone has to watch it. For each metric write down three things: who is responsible, how often it is reviewed and at what value a signal is raised.

For example: stock accuracy — the warehouse manager, weekly, and if it falls below 95 percent a cycle count is run. Fill rate — jointly with the sales team, monthly, and below 90 percent supply is reviewed. These thresholds are only samples; set yours according to your goods and customers. What matters is that "good" and "bad" are agreed in advance, so you do not argue about it every time.

Common mistakes

• Calculating differently each month. At cost this month and at selling price next month makes comparison meaningless. Choose the method once and write it down.
• Looking only at the overall average. 96 percent accuracy across the whole warehouse looks fine, yet one zone may sit at 80. Look at zones and categories too.
• Measuring what is easy to measure. Counting pallets is simple, but it is not what the decision needs.
• A metric without an owner. Everyone sees the number and nobody answers for it.
• Mixed units. When pieces, cartons and kilograms share one column, no formula will work correctly.
• One-off measurement. If a figure is calculated for a single month and then forgotten, there is no trend, so improvement or deterioration goes unnoticed.

A plan for the first month

1. Week one: choose four metrics and write the formula for each.
2. Week two: export the last three months of data from 1C or Excel and calculate the starting values.
3. Week three: set thresholds and name the people responsible.
4. Week four: hold the first short meeting — for every figure, either one decision or the answer "no change for now".

Without a starting value you cannot judge any change, so do not skip step two. After that, review the list every quarter: dropping a metric that does not work matters as much as adding a new one.

Conclusion

Good warehouse analytics starts not with many numbers but with a few reliable ones that lead to decisions. If you do not trust the numbers, first check the balances against a real count. CountLab works in exactly this direction: it turns stock-count results into figures and charts and pulls them together into a short management report.