Closing the Shop for a Day to Count It Was the Wrong Idea

The annual full count is dreaded, disruptive and usually wrong by the time it ends. Counting continuously in small pieces is more accurate and costs no trading days.

Every March, Gopal closed his Vijayawada shop on a Sunday and counted everything with his two staff. It took eleven hours, cost a trading day, and produced a number he did not entirely believe by Tuesday.

The last time he did it, the count was ₹1.9 lakh below his records. He had no way to know whether that represented a year of theft, a year of unrecorded write-offs, or a year of small counting errors accumulated in an eleven-hour marathon at the end of which nobody was concentrating. So he did what most shops do with a number they cannot explain: he adjusted the records to match and moved on.

Why the Annual Count Fails

It is not laziness. The method has problems the effort cannot fix.

• A variance covering twelve months cannot be traced to a cause, so nothing is learned and nothing changes
• Counting accuracy falls sharply after the first few hours — the last aisle is the least reliable
• A shop closed is a day of revenue gone, plus customers who went elsewhere and may stay there
• By the time you finish, the early aisles have been trading for hours if you did not close
• It happens once, so the count that is wrong stays wrong for a year

I was spending a Sunday every year to produce a number I did not trust and could not act on.

Counting Small, Continuously

Gopal now counts about fifteen minutes a day, before opening. Roughly twenty product lines, chosen so that everything gets counted on a cycle, with his fastest-moving and most valuable stock counted far more often than the rest.

• Top-value and fast-moving lines: monthly
• Middle group: quarterly
• Slow, low-value lines: once or twice a year
• Anything that produced a variance last time: again next week

The effect is not mainly the counting. It is that a variance found on Tuesday covers about four weeks, is small, and is traceable — the delivery that was short, the damaged packets removed without a write-off, the staff member who sold a loose kilo without billing it. A cause you can name is a cause you can fix.

A count taken while the shop is selling is wrong before it is finished, and the discrepancy it produces is meaningless. Fifteen minutes before opening, with the counter closed, is worth more than two hours during the day.

Counting It Properly

The mechanics matter more than people expect, and most of the classic errors are procedural rather than arithmetic.

• Count blind — the counter should not see the expected number, or they will find it
• Count by location, shelf by shelf, not by walking a product list around the shop
• Include the back room, the loft, the display near the door and anything in the delivery area
• Record zero explicitly for lines you expected to find and did not, since a blank is ambiguous
• Recount variances the same morning, before investigating — around half of them are counting errors

Blind counting is the one shopkeepers resist and the one that matters most. Shown an expected figure, a human being counting quickly will confirm it. It is not dishonesty, it is how attention works.

What the Variance Is Telling You

A clean count with a real gap is information, and the four usual explanations have different fixes.

• Receiving errors — you were billed for more than arrived, and the fix is at the door
• Unbilled sales — small items handed over without a bill, the most common cause by volume
• Unrecorded write-offs — damaged and expired goods binned without an entry
• Theft — real, and usually the smallest of the four, though it is the first one people assume

Gopal has not closed the shop for a count in two years. His variance on any given cycle now runs under half a percent, and when it moves, he knows within a week what moved it.

Mapis schedules cycle counts by value and movement, hides the expected figure while you count, and records each variance with the date and the person, so a pattern is visible long before it becomes a ₹1.9 lakh mystery.

Frequently asked questions

How often should a small shop do a stock count?

Continuously in small pieces rather than annually in one marathon. Count your highest-value and fastest-moving lines monthly, a middle group quarterly, and slow low-value lines once or twice a year. Around fifteen minutes before opening covers a typical kirana shop on that cycle.

Should I close the shop to do a stock take?

No. A closed day costs revenue and customers, accuracy collapses after a few hours of counting, and a variance covering twelve months cannot be traced to any cause. Short counts before opening are more accurate, cost no trading time, and produce variances small enough to investigate.

What is blind counting and why does it matter?

Counting without seeing the expected quantity. Shown an expected figure, a person counting quickly will tend to confirm it rather than count independently — this is ordinary human attention, not dishonesty. Blind counts are the single largest accuracy improvement available in stock verification.

What causes a difference between counted stock and recorded stock?

Four things, in rough order of frequency: unbilled sales where small items are handed over without a bill, unrecorded write-offs of damaged or expired goods, receiving errors where you were invoiced for more than arrived, and theft. Theft is usually the smallest of the four and the first one assumed.