A cycle count is a count of a small slice of your inventory on a regular schedule, checked against what your system says you have on hand. Instead of counting the whole store in one annual marathon, you count a manageable piece of it every day or every week, so every SKU gets verified on a cadence that matches how much it matters.

You know the other way of doing it: a Sunday in January with the doors locked, the heat barely on, a box of donuts going stale by the register, and a stack of clipboards getting passed around. Someone draws the fastener aisle, and by midafternoon nobody on the crew can say with a straight face whether bin 214 holds 847 wood screws or 850. You pay a day of overtime or give up a selling day, and what you get for it is one accurate number on one day, because by the following Wednesday receiving has shorted you on something, a clerk has checked out the wrong fitting under the right-looking SKU, and the decay starts all over again. Whether your store carries 40,000 SKUs or 15,000, there is a better way to keep the number honest, and it does not involve closing.

Why the once-a-year count lets you down

The annual full count fails you in five specific ways, and only one of them is the donuts.

The number is right for a day and wrong for the next 364, because every receiving error, mis-pick, and unrecorded torn bag chips away at it from the moment the count ends. Those errors compound, since a unit-of-measure mistake made in February does not get caught until the following January, which means eleven months of purchase orders built on a bad on-hand number. The event itself costs you a closed store or a crew on overtime, and tired people count badly, so hour six of a full count produces some of the least reliable data your business will ever generate.

Worst of all, a full count surfaces thousands of variances at once, and nobody investigates thousands of anything. The variances get written off in a batch, the root causes walk out the door with the clipboards, and the same errors come back next year wearing the same coat. Cycle counting flips that around, with small counts done often and variances small enough to chase while the trail is warm.

ABC cycle counting: count the movers more often

The backbone of most cycle count programs is ABC analysis, where you rank every SKU by annual dollar movement (annual units sold times cost) and count the top of the list far more often than the tail. The common rule of thumb has A items, roughly the top 10 to 20% of SKUs carrying most of your dollars, counted monthly or more; B items, the next 30%, counted quarterly; and C items, the long tail, counted once or twice a year. It is a rule of thumb rather than a law, so tune it to your store.

Here is the arithmetic for an illustrative hardware store with 20,000 active SKUs and 20 counting days a month (weekdays).

Class SKUs Counted Counts per month
A 2,000 Monthly 2,000
B 6,000 Quarterly 2,000
C 12,000 Once a year 1,000
Total 20,000 5,000

The 2,000 A items counted monthly produce 2,000 counts a month. The 6,000 B items counted quarterly produce 24,000 counts a year, which works out to another 2,000 a month, and the 12,000 C items counted once a year add 1,000 a month. That totals 5,000 counts a month, every SKU in the building gets touched at least once a year, and your top 2,000 movers get verified twelve times, which puts the most checking on the SKUs where an error costs the most money and does the most damage to your ordering. If you want to go deeper on what those dollars are doing for you, our piece on inventory turnover and GMROI picks up from there.

Beyond ABC: count by risk and by rhythm

Dollar rank is the skeleton, but a supply store's error patterns follow the rhythms of the business more than they follow the dollar curve, so add these triggers on top of the ABC schedule.

  • Count before you order. The day before the weekly distributor order goes out, count that vendor's section, so the purchase order starts from a true on-hand number. The ordering math itself lives in our article on the reorder point and safety stock formula.
  • Count when truck day goes wrong. When a receiving line does not match the packing slip, count that item on the shelf the same day, while the truck is barely out of the lot.
  • Count negative on-hand items immediately. A negative on-hand quantity is your system raising its hand to tell you it is wrong.
  • Count unit-of-measure items often. Bin fasteners, rope and chain sold by the foot, and anything received by the box and sold by the each are where conversion errors hide, and they hide well.
  • Count seasonal aisles twice. Count once before the season starts, so preseason orders begin from truth, and once at the end, so you know your leftovers before you plan next year: ice melt and snow shovels in the fall, lawn fertilizer and grass seed in late winter. Those end-of-season counts also feed your dead stock decisions.
  • Count bulk bagged goods on the pallet. Feed, mulch, softener salt, and bagged concrete all suffer torn-bag damage that never gets written off, and a ripped 50-pounder kicked under the rack is shrink with a trail of pellets behind it.
  • Count high-shrink items more than their rank suggests. Batteries, small power tools, blades, drill bits, and trimmer line may rank as B items on dollars, but their walk-out-the-door appeal says A.
  • Count the contradictions. Anything showing zero on hand that sold last week, and anything a customer says you are out of while the system insists you have six, goes on today's list.

A counting schedule that fits in an hour

Take the 5,000 counts a month from the worked example and divide by 20 counting days, and you get about 250 counts a day. At a brisk three to four items a minute with a phone scanner, 250 divided by 3.5 comes to roughly 71 minutes of counting a day, which two people can split before opening or during the mid-afternoon lull between the contractor morning rush and the after-work homeowner crowd, without anyone missing a selling day.

Layer the triggers on top: Monday's count includes the vendor section for Tuesday's order, truck day counts ride along with receiving, and the seasonal aisles get their two scheduled passes on the calendar. Shrink never takes January off, and now neither does your counting.

Handling variances: fix the process, not the number

A variance is information, and how you handle it decides whether you learn anything from it. Start with blind counts, where the person counting does not see the expected quantity, because a counter who can see the system number has a funny way of finding the system number. Then set a recount threshold, such as a rule that any variance over $50 at cost, or over 10% of expected units, gets recounted by a second person before it is accepted, while everything under the line gets posted.

When a variance survives the recount, resist adjusting the number and walking away, and ask which of the usual suspects did it, because in a supply store the list is short:

  • a receiving error, where shorted or substituted items were received as ordered
  • a unit-of-measure mismatch, such as a box of 100 received as one each, or the reverse
  • a counter substitution, where a similar item got checked out under the wrong SKU
  • damaged product returned to the shelf and never written off
  • product committed to a will call sitting on a shelf
  • stock in the back room or on an endcap that did not get counted with its home location
  • theft

Track your variances by cause as well as by dollars, because the fix for a receiving problem is a receiving process change rather than a recount, and you will never see the pattern if every variance disappears into one adjustment bucket. Decide two more rules once and apply them consistently, namely how you treat goods on open will calls (committed inventory) and whether you count during open hours, since consistency matters more than which answer you pick.

Counting on phones instead of clipboards

The clipboard workflow counts the same item twice, once on paper in the aisle and again at a keyboard where a tired person types it into the system and occasionally invents a new number in the process. Counting on a phone collapses that into one step: you scan the product with the camera, key the quantity, and the count is in the system before you have moved to the next bin. Cutting out the double entry and the handwriting is what makes an hour-a-day schedule realistic, and it makes blind counts easier to enforce, since the phone can show a counter what to find without showing what the system expects. The best inventory scanner in your store is the one your team already carries.

When you still need a full physical count

Cycle counting does not retire the full physical inventory forever, but it does retire the pointless ones. You still reach for a full count when your accountant or bookkeeper wants one at fiscal year-end to support the inventory value on your books, so ask what they need and whether good cycle count records can shrink the scope. You need one when a lender, a bank covenant, or an insurance claim calls for documented numbers, and when you are buying or selling the business or bringing on a partner, because everyone at that table wants the same verified starting point.

Count into a new POS system when you switch, so the new system starts clean. And if your cycle counts ever show accuracy falling apart across the board, with many departments posting double-digit variances at once, a reset count beats chasing the problem item by item.

How Rundoo handles it

Rundoo supports two inventory count scopes: a Full count for everything, and a Cycle count for a specific set of products. To start one, your team opens Counts in the POS, starts a new inventory count, picks Cycle as the scope, and adds products individually or by rule, such as vendor, category, product line, or product tag, which turns "count the vendor's section before the order goes out" into a quick setup rather than a printout. The count has "to count" and "counted" tabs so progress is visible, staff enter the counted quantity rather than the variance, and completing the count applies the adjustments.

The Staff app runs the count from the phone in your team's pocket, on iOS and Android, scanning products with the phone's camera so shrink shows up in your numbers instead of as a surprise, and counts start blind, so nobody confirms last month's number on autopilot. Inventory in Rundoo is real-time, so stock, cost, and on-order quantities update when you check out a customer or receive a purchase order, across the web app and the Staff app, and your team can look up price and stock across locations and print shelf labels from the same phone. Our hardware store POS page shows how the rest of the system fits a deep aisle.