Stock sync latency: the price of stale data
From the number of orders, the share of orders on low-stock SKUs and the sync interval calculates how many orders a month sell what is already gone, what it costs and whether more frequent refresh pays back.
At a glance
- What question does it answer?
- What stale stock on a channel costs and whether more frequent refresh pays back.
- What data is needed?
- Orders a day and the share of orders on low-stock SKUs.
- How to read the result?
- On the default data (1,200 orders a day, 6% low-stock, 15% conflicts, sync every 6 hours) there are 2.7 orders a day costing 36,450 a month. With hourly sync it is 6,075: a saving of 30,375 against a cost of 25,000, a net plus of 5,375 a month. Speeding up pays back from 988 orders a day. Example on the default values: Oversold orders a day: 2.7; Cost of stale data a month now: RUB 36,450; The same after speeding up: RUB 6,075; Saving a month: RUB 30,375. The figures are illustrative: enter your own to see your result.
Why and how to reason
Stock between the warehouse, the accounting system and the marketplace is not updated instantly but in batches every few hours. Until the exchange runs, the marketplace sells what is no longer in the warehouse. Such an order has to be canceled: that is lost margin, a marketplace penalty and an unhappy customer. The risk does not apply to all orders, only to those falling on low-stock SKUs, and not every such order becomes a conflict. So the calculation multiplies three shares: orders × low-stock share × conflict share = 1,200 × 6% × 15% = 10.8 conflicts per day of delay. With an exchange every 6 hours the window is four times shorter than a day, so 2.7 orders a day, and at 450 per case 36,450 a month. Speeding up to one hour cuts the loss to 6,075, and the saving of 30,375 a month covers the cost of 25,000.
When to use it
It is useful when you have several sales channels with a shared warehouse and are deciding whether to pay for a more frequent exchange: integration, an event queue, a channel reserve. Do not use it if stock is updated on events (then the delay is seconds) or if almost all SKUs carry large stock. Assumptions: orders are even through the day, the conflict share is measured rather than guessed, the price of one case includes the penalty and the cancellation.
How to read the result
Compare “Saving a month” with “Cost of speeding up”: here 30 thousand against 25 thousand, so the speed-up pays back, but with a small margin of 5 thousand a month. “Pays back from orders a day” (about 990) shows the threshold: at a lower flow the saving is below the cost. If the speed-up does not pay back, do not give up protection entirely: a safety buffer on the channel and hiding the last units are cheaper and work at once. If you enter a one-off cost, look at its payback period: it must be shorter than the life of the solution.
Common mistakes
Typical trap: taking the conflict share by eye, treating all orders as exposed, forgetting that the exchange interval and the real delay are not the same (queue, failures), ignoring the marketplace penalty. Measure the conflict share over a month from the cancellation log before deciding on speeding up.
What data is needed?
- Orders a day and the share of orders on low-stock SKUs.
- The conflict share per day of delay and the cost of one order.
- The interval now and after speeding up, the cost of speeding up.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Orders a day on the channel | pcs | 1,200 | The site or platform the stock is pushed to. The example value is illustrative. |
| Share of orders on low-stock SKUs | % | 6 | SKUs with a few pieces left: take it from your own data. The example value is illustrative. |
| Share of conflicts per day of delay | % | 15 | What part of such orders hits an already sold unit during a day without a refresh. Measure it on your own data. The example value is illustrative. |
| Cost of one oversold order | ₽ | 450 | Cancellation, handling, a platform fine or compensation, excl. VAT. The example value is illustrative. |
| Days in a month | days | 30 | To convert to a month. The example value is illustrative. |
| Current sync interval | h | 6 | How often stock is pushed to the channel. The example value is illustrative. |
| Interval after speeding up | h | 1 | The target frequency: an hour, minutes. The example value is illustrative. |
| Cost of speeding up a month | ₽ | 25,000 | Integration, server, license: the contractor’s offer, excl. VAT. The example value is illustrative. |
| One-off cost of speeding up | ₽ | — | Integration development and launch: the contractor’s offer, excl. VAT. Empty: the payback period is not calculated. The example value is illustrative. |
How it is calculated
| Step | Formula |
|---|---|
| Conflicting orders a day at a one-day delay | orders × low-stock share × conflict share |
| At an interval of T hours | multiply by min(1; T ÷ 24) |
| Cost a month | number of orders × cost of one × days |
| Payback | cost ÷ (saving per order a day) |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Oversold orders a day
- 2.7At the current sync interval.
- Cost of stale data a month now
- 36,450 ₽
- The same after speeding up
- 6,075 ₽
- Saving a month
- 30,375 ₽
- Saving less the cost of speeding up
- 5,375 ₽
- Pays back from orders a day
- 988Orders a day at which the saving equals the cost.
- Payback of the one-off cost
- no one-off cost entered
Data as a table
| Measure | Per month |
|---|---|
| Now | 36.5k |
| After speeding up | 6,075 |
| Saving | 30.4k |
Speeding up pays back at your numbers. Check that the conflict share was measured rather than guessed.
The model is linear in the interval up to a day and ignores lost customer trust. Reconciling stock between ERP, warehouse and platform on a schedule and records accuracy: the “Stock record accuracy” calculation.
Nuances and limits
- Measure the conflict share on your data: it determines everything.
- The model is linear up to a day.
- The model does not count lost customer trust.
- Sometimes a safety buffer on the channel is cheaper than frequent sync.