Most store owners can tell you to the cent what the register took yesterday.

Almost none can tell you what was on the shelves.

That gap is the most expensive thing in retail, and it stays invisible because nothing about it looks broken. The day closes, the deposit balances, the store feels fine. Meanwhile margin leaks out somewhere between the delivery truck and the customer's bag: stock that was never counted, a best seller that sat empty for nine days, a dead line you kept reordering out of habit.

First, what a POS system actually means

POS is short for point of sale. It is the system that records the moment a sale happens: software, plus whatever hardware you attach to it, such as a terminal, a barcode scanner, a receipt printer and a cash drawer.

The hardware is not the important part. This is: a POS records a sale as a list of products, not as a total.

That one difference carries the entire argument. A register gives you $3,200. A POS gives you $3,200 and the 84 items that produced it, each with quantity, price, tax and timestamp. Every useful thing downstream, every stock figure and every buying decision, is assembled from that list.

What the gap costs

These are industry figures, not our estimates. They describe retail at every scale, but small stores feel them hardest, because a small store has no buffer.

6.5%
of global retail sales
lost every year to out of stocks and overstock
$1.77T
lost worldwide to inventory distortion
IHL Group, 2025
~63%
average inventory record accuracy
Auburn University RFID Lab
1.68%
of revenue lost to shrink in the US
highest rate in over a decade

Read the middle figure alongside one more: 43% of small businesses do not track inventory at all, or track it manually. Roughly a third of inventory records are wrong at any moment, and nearly half of small businesses have no system that would tell them.

You are not running out of stock because you are unlucky. You are running out of stock because nobody is counting.

Where the money actually goes

When retailers finally measure the leak, it is never one dramatic loss. It is four or five medium ones that nobody was watching.

Global inventory distortion by cause (IHL Group, 2025)
Empty shelves
$690.9B
Weak customer service
$165.6B
Product not findable
$145.2B
Price or offer mismatch
$77.4B

Look at the biggest bar. It is not theft. It is empty shelves, which means a customer who walked in wanting to give you money and left because the item was not there.

Shrink at least leaves a gap you eventually find at count time. A lost sale leaves nothing at all, and it is the largest line in the chart by a wide margin.

The same study splits the total into $1.2 trillion of out of stocks and $572 billion of overstocks. Both are the same disease: you do not know what you have.

The four things a POS does

Everything a vendor demos sits on top of these four.

  1. Records every sale as a line item

    Not a total, a list. This product, this quantity, this price, this tax, at this time. The total becomes a by product.

  2. Moves stock automatically

    Every sale takes the count down, every receipt of goods takes it up. You stop walking the aisle to find out what you have, because the number was never lost.

  3. Produces a correct, taxed receipt

    The right rate for the right jurisdiction and category, applied from the product record rather than recalled at the counter, and the same record feeds your filing.

  4. Turns that history into decisions

    Once a year of itemised history exists, questions that used to be guesses become arithmetic. What actually sells? What should I reorder Thursday? Which vendor's line is dead?

Step four is the one that pays for the system, and it is the one you cannot skip to. It works only because steps one to three ran honestly, every day, for months. A POS is not a report you buy. It is a record you accumulate.

Spreadsheet versus system

Neither column is a caricature. Plenty of good stores run on the left one. It simply has a ceiling.

How the same day goes
Manual or spreadsheet
A POS
Receipt totalled at the counter
Total, tax and rounding computed per line
Tax rate recalled or looked up per sale
Rate attached to the product once
Stock known by looking at the shelf
Stock known before you walk to the shelf
Reorder decided by what looks empty
Reorder decided by what sold, and how fast
Staff sales and cash handling on trust
Every ticket carries the name of who rang it
Month end is a reconstruction
Month end is a report that already exists
A missing item is found by a customer
A missing item is flagged before it runs out

The specific failure of a spreadsheet deserves naming. A spreadsheet lets you overwrite a number. Somebody corrects a stock figure in March, nobody records why, and from then on the file is confidently wrong. That is the trap: it feels like a system while behaving like a guess.

Shrink is a measurement problem first

US retailers lost an average of 1.68% of revenue to shrink in the most recent National Retail Security Survey, roughly $112 billion industry wide and the highest rate in over a decade.

Here is the part that usually gets skipped. You cannot have a shrink number at all without inventory records, because shrink is the difference between what your records say you should have and what you actually counted. With no records, that difference is undefined. The loss still happens. You simply never learn its size.

Types of POS systems, and which one a store needs

The category is wider than it looks, and most of it is not aimed at you.

TypeWhat it isWho it suits
Legacy on premiseSoftware on one terminal, data stored locallyStores with poor connectivity, but backups become your problem
Cloud POSRuns in a browser or app, data synced to a serverMost retail, and anyone who wants reports from off site
Mobile POSSelling from a phone or tabletSmall counters, pop ups, markets, line busting
Self checkout or kioskCustomer scans and pays unattendedLarge format stores, rarely justified below that
Restaurant POSTable management, kitchen tickets, course timingFood service, not retail

For a grocery, convenience, apparel or specialty store, the practical answer is a cloud POS that keeps selling when the internet drops and syncs when it returns. Treat continuous connectivity as something that will fail, because it will.

The reorder problem is the real prize

Everything above is hygiene. This is the part that changes what the store earns.

Every reorder decision is a forecast. You are already predicting demand. You are just doing it from memory, under time pressure, with a rep waiting.

Memory is biased in a specific direction. You vividly remember the item you ran out of last week. You completely forget the twelve slow items in the back, financed by you, tying up the cash you needed for the first one.

Once itemised history exists, the forecast becomes measurable:

And the discipline that keeps it honest: stock should be an append only ledger. Every movement, whether sale, receipt, return, damage or correction, is recorded as an event, and the current count is derived from those events rather than typed by a person. The moment somebody can overwrite the stock number you are back to ghost stock, a system that is confidently wrong, which is worse than a spreadsheet because you trust it.

Be honest: do you need one yet?

Not every store does. Here is a straight test rather than a sales pitch.

You are probably still fineYou have outgrown manual
Under roughly 50 SKUs you know by heartHundreds of SKUs, with variants or sizes
You are the only person who rings salesStaff ring sales when you are not there
One locationMore than one, or planning a second
One or two vendors, ordered weeklySeveral vendors on staggered lead times
Counts never surprise youCounts regularly disagree with expectation
You never wonder what sold last monthYou are making buying decisions from memory

If you are reading mostly the right hand column, manual tracking is no longer saving you money. It is costing you margin you cannot see.

What to look for, and what to ignore

And the things that sell POS systems but rarely change a store's economics: loyalty programs before you have the traffic to use them, a dozen dashboards nobody opens, and any AI claim made without the sales history to justify it. Intelligence comes after the record, never instead of it.

Common questions

What does POS stand for?

Point of sale. It refers both to the moment a transaction happens and to the system that records it.

What are the main types of POS systems?

Broadly: on premise, cloud, mobile, self checkout and industry specific systems such as restaurant POS. Most small retailers are choosing between cloud and mobile, and often use both on the same counter.

What are the disadvantages of a POS system?

A real cost per month, time to set up your catalog properly, and a dependency you did not have before, which is why offline capability matters. There is also a discipline cost: a POS only stays accurate if receiving and returns are actually entered.

Is a card reader a POS system?

No. A card reader takes payment. A POS records what was sold. Some products bundle both, but taking payment alone leaves you with the same blind spot you started with.

How long before it pays for itself?

The compliance and speed benefits arrive immediately. The expensive problem, buying the wrong stock, needs roughly one season of history before the system can say anything useful about it.

The one line version

A POS does not make you money the day you install it. It starts a record, and three months later that record is the only thing standing between a decision and a guess.