When your inventory tracking lives in one person's head

Ask whether you have it, and listen to what happens

Inventory tracking in a small business usually means one person who knows what is on the shelf. It is accurate and fast and it costs nothing, until the volume passes what one person can carry. After that the same question gets two answers depending on who you ask, and the stock figure stops being something anyone can plan against. What replaces it is a single count that every channel reads from and writes to.

A customer asks whether the black one in medium is available. What follows is not a lookup. It is a shout across the shop, or a call to the storeman, or a pause while somebody thinks about what they saw on the shelf on Tuesday. Most of the time the answer is right, which is exactly why nobody treats it as a problem.

The bill arrives in two forms. Occasionally a customer pays for something that turns out to be gone, and somebody has to make a phone call nobody wants to make. More quietly and more expensively, you reorder forty of something you already had thirty of, and that money sits on a shelf for a season.

Neither of those is a failure of the person holding the count. They are usually extraordinarily good at it. It is a failure of a business that has grown past what one memory can hold, without anyone noticing the moment it happened.

Stock leaves without a record more often than it arrives without one

Goods coming in almost always get written down, because there is an invoice attached and somebody has to pay it. Goods going out are a different story, and that asymmetry is where nearly all the drift comes from.

Something goes out for a photoshoot and comes back three days later, or does not. A staff member buys something at cost and the note gets made later, which means never. A breakage gets swept up. A return comes in and goes straight back onto the shelf without touching any record. And the worst offender is the reserved pile, meaning items promised to a customer who is coming on Saturday, which are neither sold nor available and are counted as both.

On top of that, the shop and the online listing usually sell from the same physical shelf while looking at two different numbers, and neither of them decreases when the other one sells. In a market where the same business runs a counter, an Instagram page and a marketplace listing at once, one shelf is being promised in three places.

• Samples and photoshoot items that leave with no line written
• Staff purchases recorded later, which means not recorded
• Returns put back on the shelf without touching any record
• Breakages and damage that adjust the shelf but not the sheet
• Reserved items counted as available and as sold at the same time

Three questions that tell you where you stand

Answer these honestly and you will know whether this is your problem without buying anything.

First, can you say right now, without asking anyone, how many units of your best selling item you hold. Second, the last time something left the premises that was not a sale, what got written down and where is that piece of paper. Third, when was your last real count, and how big was the gap between what you counted and what the records said.

Then add one measurement over the coming month: how many times you told a customer something was available and then had to go back on it. That number is the one that costs you reputation rather than money, and it is the one people forget to track because each instance is quietly resolved and never written anywhere.

The free fix: five minutes at closing, on the items that matter

Do not start by counting everything. A full stock take is a project, it happens once, and nothing about the daily process changes afterwards, which is why businesses run one every year and stay exactly where they were.

Instead, pick the small set of items that carry most of your revenue. For most businesses it lands somewhere around twenty lines. Count those, and only those, at closing time. It takes five minutes once people are used to it.

The important part is what you do with a discrepancy. Do not silently correct the sheet to match the shelf. Write the difference down, with the date, and leave both numbers visible. Two weeks of written differences will show you exactly where stock is leaving without a record, which is information no annual count has ever produced, because an annual count tells you the total damage and nothing at all about the cause.

Alongside it, put one pad by the door and one rule next to it: nothing leaves without a line. Sample, photoshoot, staff purchase, repair, replacement for a complaint, all of them. And give reserved stock a physical shelf of its own with a dated label, so that a promise to a customer stops being invisible and expires on a stated day rather than sitting there until somebody wonders about it.

• Count the twenty or so lines that carry the revenue, not the whole catalogue
• Count at closing, every day, and keep it to five minutes
• Write the difference rather than correcting the sheet quietly
• One out pad by the door, and nothing leaves without a line
• A separate labelled shelf for reserved items, with an expiry date on the label

What a system adds, and where it is the wrong purchase

The useful part of a stock system is not the counting. It is having one number that the counter, the online listing and the person answering messages all read from, so the same last piece cannot be sold twice. Add an alert when a fast moving line drops below the level that covers your supplier's lead time, and the ability to answer an availability question in a chat without anyone walking to a shelf, and you have most of the value.

The blunt prerequisite is that a system holds whatever you put into it. If nobody counts today, a system will hold wrong numbers, and it will hold them with more authority than a person who at least knew they were guessing. Stock software installed on top of an uncounted shelf produces confident availability answers that are wrong, which is worse than a shout across the shop.

Two cases where you should leave this alone. If everything you sell is made to order, or every piece is genuinely unique, there is no count to keep and a stock system adds admin for no return. And if you know your numbers perfectly well and still run out, your problem is supplier lead time or cash tied up elsewhere, and a system will simply inform you more precisely that you have nothing to sell.

One shelf promised in three places is not an inventory problem yet. It becomes one on the first busy Saturday.

After the five minutes at closing, what a shared number involves

Run the five minutes at closing with your own team for a month before buying anything. If the count on the items that matter holds, keep it in-house and spend the money elsewhere, because the discipline is the product. What breaks it is the second channel. The moment the counter and the online listing can both sell the same last piece, no closing routine is fast enough, and the problem stops being a counting problem at all.

There is no price on this site. What moves the number here is how many places a sale can happen, whether your till or your storefront will permit anything to read and write a stock figure, how many lines actually matter, and who keeps it honest afterwards. The audit takes the first week and most of it goes on establishing whether the register and the shelf currently agree. One shared number is live inside a fortnight, on the lines that matter rather than on everything in the building.

The system holds one number and makes no decisions about it. A discrepancy hands it to a person rather than adjusting itself, because a stock figure that quietly corrects is a theft report nobody read. Any write-off, any price change and any refusal to sell wait for human approval. The stock record, the movement history and the automations sit in accounts under your own logins, so you own the system and the history outlives the person who used to hold it in their head.

The published work nearest this is RM Gulistan Engineers in Karachi, where three purpose built ERP systems cover accounts, HR and inventory, and Emraan Rajput, where Shopify operations run inside a full operating system. Ibrahim owns build and workflows at Wobble, which works from Karachi, bills month to month and carries 25 engagements across six countries.

Common questions

How often should a small business count stock?

Daily on the small set of lines that carry most of your revenue, and less often on everything else. A short daily count catches the cause while it is still recent, while an annual count only measures the total damage after the fact and tells you nothing about where it came from.

Do I need inventory software or is a spreadsheet enough?

A sheet is enough while one person updates it and one channel sells from it. The point at which it stops working is when two channels sell the same physical item, because a sheet cannot decrease itself when a sale happens somewhere else, and that is the failure that costs you a customer rather than an hour.

How do I stop selling items that are already gone?

Separate available stock from reserved stock physically and in the record, and give reserved items an expiry date. A large share of oversold items are things promised to somebody who never came back, sitting in a state that is neither sold nor for sale, and being counted as both.

What causes stock discrepancies most often?

Goods leaving without a record rather than theft, in most cases. Samples, photoshoot items, staff purchases, breakages and returns put straight back on the shelf account for a great deal of it. That is why writing the difference each day is more useful than correcting the sheet, because the pattern of differences names the cause.

Can an AI assistant answer availability questions from stock data?

It can, and it is one of the more useful things to automate, because availability is among the most repeated questions any seller gets. It is only as good as the count behind it, so decide what happens when the number is uncertain. Saying that you will confirm in ten minutes is far better than a confident answer that turns out to be wrong.