Ask most Shopify stores how accurate their picking is and the honest answer is "we don't hear many complaints." That's not a measurement. It's a guess built on the customers who bothered to write in, and it's about to be tested by the busiest weeks of the year. Here's how to put a real number on it before Black Friday, with a spreadsheet or with an app.
What pick accuracy actually means
Pick accuracy (often called order accuracy) is the share of orders that shipped with exactly the right items in exactly the right quantities:
Order accuracy = orders shipped without an error ÷ total orders shipped
Ship 400 orders in a month, learn that 3 went out wrong, and you're at 397 ÷ 400 = 99.25%. Simple enough. The hard part isn't the formula, it's the "learn that 3 went out wrong" part.
Why complaints undercount your errors
The only errors you can count after the fact are the ones customers tell you about. Plenty don't. Some keep the wrong size and never reorder. Some get a cheaper item than they paid for and shrug. Some leave a one-star review instead of emailing. So the number you get from complaints and returns is a floor: your real error rate is that number or higher, never lower.
That's why one number isn't enough. You need to look at your picking from three sides.
The three numbers to track
1. Errors that shipped
Wrong item, wrong quantity or missing item, as reported by a customer. This is the number your customers feel. Count orders, not complaints: one order with two wrong items is still one bad order.
2. Errors caught before shipping
The near misses: the packer who noticed the navy shirt should have been black, the scan that didn't match, the short item flagged before the box was sealed. These never reach a customer, so most stores never write them down. They should, because they tell you where mistakes start. If the same product keeps getting caught at the packing table, it will eventually slip through.
3. How much of the work was actually checked
If nobody checks anything, you'll catch nothing, and "zero errors caught" will look like perfect picking. Track how many orders (or units) were verified by a scan or a second person, out of everything shipped. It's a process number, not an accuracy number, but it tells you how much to trust the other two.
Read together, they tell a story. Caught errors going up while shipped errors stay flat means your checks are working. Shipped errors going up while caught errors stay at zero means you're not checking. A low check rate means both numbers are mostly guesses.
How to track it with a spreadsheet
No app needed to start. Make a sheet with one row per mistake, caught or shipped:
| Column | Example |
|---|---|
| Date | Nov 4 |
| Order | #1482 |
| What happened | Wrong size: sent M, ordered L |
| Product / SKU | TEE-NAVY-L |
| Where it was found | Packing / Customer |
| Who picked it | (optional, for coaching, not blame) |
Then, once a week:
- Shipped error rate = orders with a "Customer" row ÷ orders fulfilled that week. Shopify's order list gives you the fulfilled count for any date range.
- Caught count = rows found at picking or packing.
- Patterns: sort by SKU, then by date. Look for repeats.
Two habits make the sheet honest. First, when you process a return, use a wrong-item reason whenever that's what happened (Shopify's own return reasons include one), so shipped errors don't hide under "changed mind." Second, don't trust a rate from a tiny sample: with 15 orders a week, one mistake swings you by almost 7 points. Wait until you have at least a few dozen fulfilled orders before reading much into the percentage, and look at the counts instead.
What to do with the numbers
The point isn't the percentage, it's the pattern. Most stores find their errors cluster in a few places:
- Look-alike variants. Sizes and colors that differ only on the label. Fix: scan each unit, or at least separate them on the shelf.
- Shared or unclear bins. Two products in one spot, or a bin label that doesn't match the system. Fix: one product per bin, and a bin-sorted pick list so pickers go straight to it.
- Busy days and new people. Errors spike on sale days and in someone's first week. Fix: more checks on those days, not fewer.
- Quantities. "2 of" read as "1 of." Fix: a pick list that shows totals clearly, and a count at packing.
The five picking and packing mistakes guide goes deeper on each fix.
How Pickadoodl measures it for you
If you'd rather not keep a sheet, Pickadoodl tracks the same three numbers automatically. The Accuracy card on Analytics keeps them apart:
- Caught before shipping: every scan Pickadoodl refused (wrong item, wrong bin, wrong tote, too many) plus exceptions your team raised on the floor.
- Not checked by a scan: units confirmed by typing a quantity instead of scanning, so you can see how much work went unverified.
- Confirmed shipped errors: return lines logged with the reason "Wrong item sent," shown as a rate only once there are at least 20 fulfilled orders in the date range.
Scan compliance (scanned units out of all confirmed units) sits underneath as a process measure. What you won't see is a single big "99.8% accurate" badge. That number would mix things that shouldn't be mixed, and it would always look better than reality. More on the Accuracy card.
Start this week
Whichever way you track it, start before the holiday rush, not after. A few weeks of baseline numbers now means that when volume triples, you'll know whether your picking is holding up, instead of finding out from the reviews in January.
