Cycle counting: keeping the number in Shopify true
Cycle counting is counting a small subset of your SKUs on a repeating schedule, so every item gets physically counted over a year without ever closing the shop for a full stocktake. If you are here because your reorder maths keeps firing late, the count is usually the reason: the number in Shopify drifted away from the number on the shelf.
What cycle counting means
Cycle counting is the practice of physically counting a small, chosen group of SKUs on a repeating schedule, so that across a quarter or a year every SKU has been counted, and no single day requires the whole catalogue to be counted at once.
The point is accuracy, not ceremony. Every reorder decision you make runs off the stock number your store holds. Reorder points, days of inventory remaining, suggested order quantities: all of them take recorded stock as the starting figure. When recorded stock is wrong, the arithmetic is still correct and the answer is still useless.
Records drift for ordinary reasons. A unit gets damaged in the stockroom and written off in your head but not in the admin. A return comes back and goes on the shelf without being received. A picker grabs the 8oz instead of the 4oz. A bundle sells and pulls from the wrong variant. None of these announce themselves. They just sit there, quietly, until the day a customer orders the last three units you do not have.
A worked example
A store carries 600 SKUs. Counting all 600 in one evening is a closed shop and a tired team, so it happens once a year, if at all. Instead, the merchant counts 25 SKUs every Monday morning. That is roughly 1,300 counts a year, enough for every SKU once and the fast movers several times over. Each Monday takes about forty minutes, and the variance from each count gets written down before anything is adjusted, so the merchant can see whether accuracy is improving or a particular shelf is the problem.
How one count runs
- Pick the list before you walk the shelf
Choose the SKUs for this count from a rota, not from a hunch. A hunch counts the things you already suspect and leaves the quiet errors in place.
- Count blind
Count without the expected figure in front of you. If you can see that the system says 14, you will find 14. A blind count is the only kind that tells you something you did not already believe.
- Record the variance, then adjust
Write down counted versus recorded for each SKU before you correct anything. The variance history is the useful output; the corrected number is just the side effect.
- Name the cause where you can
Damage, mispick, unreceived return, bundle logic. A cause written next to a variance is what stops the same variance appearing next quarter.
Counting the SKUs that matter more, more often
| SKU group | A rhythm you might set | Reasoning |
|---|---|---|
| Top sellers and hero SKUs | Monthly | They move fastest, so errors compound fastest, and running out hurts most. |
| Steady mid-range lines | Quarterly | Enough movement to drift, not enough to need constant attention. |
| Slow movers and long tail | Twice a year | Little movement means little drift, so frequent counting buys you almost nothing. |
| Anything that just surprised you | Next count, regardless of group | A SKU that sold out unexpectedly is telling you its record was wrong. |
Related terms
Cycle counting sits next to several terms on this site, and each one leans on the counted figure being right.
- Days of inventory on hand turns current stock into a number of days. An overstated count overstates your runway.
- Reorder point is the level at which you place an order. A count below it is the trigger; a wrong count hides the trigger.
- Committed vs available inventory decides which figure a reorder point should use, and a physical count measures neither of them directly, it measures what is on the shelf.
- Stockout is the outcome a cycle count is meant to prevent.
- SKU sets what you are counting in the first place, and how many lines your rota has.
- ABC analysis is the usual way merchants decide which SKUs earn a monthly count and which earn one twice a year.
A full stocktake and a cycle count are not rivals. A stocktake gives you one clean baseline and a closed shop. Cycle counting gives you a rolling, cheaper correction and a running error rate. Small stores with a few hundred SKUs generally get more from the rolling version, because forty minutes a week is a cost they can actually pay every week.
Questions
How often should I cycle count?
Often enough that a bad record is caught before it costs you an order, and rarely enough that you keep doing it. Weekly counts of a small list work for most small catalogues, with fast movers appearing on the list more often than slow ones.
Is cycle counting the same as a stocktake?
No. A stocktake counts everything at one moment, usually with the shop closed. Cycle counting counts a slice at a time, on a rota, while the shop keeps trading. They answer the same question at different cadences and costs.
What is a blind count?
A count taken without the expected figure visible. It removes the pull towards confirming what the system already says, which is where most false agreement between record and shelf comes from.
What level of variance is acceptable?
That is a judgement for your store rather than a universal figure. The practical test is direction: if variance on your fast movers is shrinking count after count, the process is working. If one shelf or one supplier keeps producing variance, that is where to look.
Does counting change my reorder point?
Not directly. A reorder point comes from how fast a SKU sells and how long your supplier takes. Counting changes whether you can see that you have crossed it, which is the part that decides when the order goes out.
Do I need an app to cycle count?
No. A rota, a clipboard and a note of each variance is a working system. Software earns its place after the count, by watching the corrected numbers every day so you do not have to check them by eye.