Comparing branches: rules for a fair comparison

An owner with several branches is usually given one table: how many so'm of shortage each branch has. The branch with the biggest number is declared the "worst", and its manager is reprimanded. Often this is the wrong conclusion: in a large branch the shortage in so'm is naturally bigger, because it sells more and holds more goods. For a fair comparison, branches have to be viewed relative to their size.

A ranking in so'm misleads

A hypothetical example: annual data for four branches, in millions of so'm.

• Tashkent, Chilonzor: sales 900, shortage 8.1, average inventory 180.
• Tashkent, Yunusobod: sales 600, shortage 7.2, inventory 150.
• Samarkand: sales 450, shortage 3.6, inventory 90.
• Andijan: sales 250, shortage 4.0, inventory 50.

In so'm, Chilonzor is the worst (8.1) and Samarkand the best (3.6). Now divide the shortage by sales: Chilonzor 8.1 ÷ 900 = 0.90 percent, Yunusobod 1.20 percent, Samarkand 0.80 percent, Andijan 4.0 ÷ 250 = 1.60 percent. The ranking changed completely: Andijan was second best in so'm but comes out worst in percent, and Chilonzor went from the worst branch to only the third worst. For the network as a whole it is 22.9 ÷ 2 200 = 1.04 percent.

Look at inventory too

The second measure is shortage as a share of average inventory: Chilonzor 8.1 ÷ 180 = 4.5 percent, Yunusobod 7.2 ÷ 150 = 4.8 percent, Samarkand 3.6 ÷ 90 = 4.0 percent, Andijan 4.0 ÷ 50 = 8.0 percent. Andijan stands out here too: on both measures it is above average. When two measures agree, the conclusion is more reliable. When they do not, the cause is often the range: a branch with expensive, small items looks different relative to inventory.

Conditions for comparison

Before comparing the figures, check for uniformity:

• Is shortage defined the same way: are spoiled, expired and written-off goods counted identically everywhere?
• Counting frequency: if one branch counts monthly and another once a year, shortage becomes "visible" at different times.
• Range: cosmetics and bulk sacks have different kinds of shortage.
• Counting method: a branch that counts with scanners finds more differences than one that counts by hand.

To explain part of the gap, break the branches down by category too: compare like categories with each other.

Chance or real difference

In a small branch a single event moves the figure sharply. In Andijan a one-off loss of 1.5 million so'm raises the rate by 1.5 ÷ 250 = 0.6 percentage points, while in Chilonzor it raises it by 1.5 ÷ 900 ≈ 0.17 points. So do not judge a small branch on one month: look at six months or a year and mark any large incident separately.

The network average is 1.04 percent. Chilonzor (−0.14 points), Yunusobod (+0.16) and Samarkand (−0.24) are within the usual range. Andijan stands out at +0.56 points: the check should start there. Set the threshold in advance: for example, a branch that deviates from the network average by more than 0.5 percentage points goes onto the review list.

What to do after comparing

Use the ranking not for punishment but for learning. Ask what the best branch does differently: are counts more frequent, is goods receipt done by two people, are the racks labelled. Carry those practices over to the others. In a branch with a high figure, break the shortage down by category, zone and shift to find the cause.

A one-page card per branch is enough: shortage rate, three-month trend, the three categories with the largest shortage, the date of the last count and the change against the previous period. Such a card lets managers speak the same language and reduces objections like "our situation is quite different".

Conclusion

A ranking in so'm measures size, a ranking in percent measures quality. CountLab runs stock counts and discrepancy analysis in all branches by the same method, brings the results to the same measures, and gives the comparison by category and zone in the management report. The owner then gets an answer not to "who is to blame" but to "where and why".