Guide

Reading the inventory accuracy report

See how far your posted cycle counts landed from what Skuvelo expected, by month, by location and by the person who counted.

Last updated

How accurate are your stock records, and where are they going wrong? Every posted cycle count records what Skuvelo expected on the shelf and what was actually counted. The Inventory accuracy report on Reports adds those up so you can see whether your stock records are getting more or less trustworthy over time.

What the variance % means

For each group the report shows:

  • Lines counted — the count lines that were posted (a recount replaces the line it recounts, so one shelf is never scored twice).
  • Expected units — what Skuvelo's records said was there when it was counted.
  • Variance units — the total of the differences, ignoring direction. Five units over on one product and five short on another is ten units of variance, not zero.
  • Variance % — variance units ÷ expected units. The column header spells the formula out: sum of |counted − expected| ÷ sum of expected.

Lower is better. A group where nothing was expected shows a dash instead of a percentage.

Inventory record accuracy (IRA)

Variance % tells you how far off the counts were. IRA tells you how many lines were close enough: the lines within tolerance ÷ the lines counted, against a 95% target unless you change it. A line is within tolerance when it meets the count's own threshold, if the count set one, or otherwise its ABC class's tolerance. Until you change them, class A allows 0%, class B 2% and class C 5%.

A worked example

Say one posted count has two lines:

  • A class A item: 100 expected, 95 counted — 5 units off.
  • A class C item: 50 expected, 52 counted — 2 units off.

Variance units are 5 + 2 = 7, so variance % is 7 ÷ 150 = 4.67%. For IRA, the class A line is outside its 0% tolerance, and the class C line is 2 ÷ 50 = 4% off, inside its 5%. One of two lines is within tolerance: an IRA of 50.00%, well short of 95%.

The four tables

  • By ABC class — each class with its tolerance, such as A (±0%). A class A IRA well below target is the first thing to fix, because those items carry the most sales value.
  • By month — grouped by the month the count was posted, in your workspace's time zone. Choose 3, 6 or 12 months at the top of the page.
  • By location — which warehouses or stores drift furthest from their records. A location that keeps coming out high is a good candidate for more frequent counts.
  • By counter — grouped by the person who recorded each line, shown by name or email. Lines counted before Skuvelo recorded who counted them show as "Not recorded".

Only posted counts appear. Counts that are still open, in review or cancelled are left out until they post.

People who can see costs also get the variance at cost: a gross figure, where an over and a short never cancel out, and a net figure, which is what the counts moved the books by. The lines valued at an estimated cost, or with no cost at all, are counted beside them.

The decision this helps with

  • Where to count more often. A location or class whose variance % keeps rising needs a shorter cadence.
  • Who needs help. A counter well above everyone else may need training, or may be counting the hardest shelves.
  • Whether a fix worked. Compare months before and after a process change.

Common mistakes

  • Reading net variance as accuracy. Overs and shorts cancel in the net figure. Use variance % or gross variance to judge accuracy.
  • Comparing counters without context. Counters who take the busiest locations will show more variance.
  • Looking for open counts. Nothing appears until a count is posted.

Questions people ask

Which location miscounts? Look down By location for the highest variance %.

Is shrinkage the same as variance? No. Variance counts every difference; net variance at cost shows whether stock went missing overall.

Exporting

Each table has an Export button that downloads the rows as a CSV, with the variance % as a plain number.

Was this guide helpful?

Try the plan on your own stock.

Start a trial with your own sales and stock, or look around the live demo first.