Safe throughput is the number of shoppers per minute who can reach checkout and complete a purchase without timing out. Measure it in a rehearsal, set the queue to that number, and keep one-in-one-out running so the store never exceeds it.
Throughput is a checkout number, not a traffic number
The mistake is setting throughput from traffic. A store that saw 5,000 visitors in the first minute of its last drop sets the queue to 5,000 a minute, and checkout collapses, because checkout was never the bottleneck's friend. Checkout is the bottleneck. It has a payment gateway, an inventory check, a fraud screen, and a confirmation page, and each one takes time. The safe number is whatever checkout completes per minute without errors, and it is almost always lower than the traffic number.
The way to find it is a rehearsal with real load. Run the queue against a test product, meter 10 users a minute through, and watch the checkout error rate. Raise it in steps and stop when errors appear. The number just below the failure point is the safe throughput. A store that skips the rehearsal is guessing, and the guess is usually wrong in the expensive direction.
Related reading on this site: Home and Waiting Room Throughput Math: How Many Shoppers Per Minute Without Breaking Checkout cover the neighbouring problems in more depth.
Why shoppers abandon in line
A queue that moves too slowly is its own failure. A shopper who watches their position tick down by one every two minutes does the math, decides the wait is forty minutes, and leaves. The abandonment is not impatience. It is arithmetic. The fix is not to hide the position. It is to set a throughput that moves the line visibly, and to show the estimated wait honestly.
The second cause of abandonment is silence. A waiting room with no updates, no stock count, and no message when the product sells out leaves shoppers guessing, and guessing shoppers leave. Show the stock ticking down in real time, show the position, and trigger a sold-out message the moment stock hits zero. A shopper who learns the product is gone leaves informed. A shopper who waits twenty minutes to learn it leaves angry, and angry shoppers do not come back for the next drop.
One-in-one-out, and what to do when stock runs out
One-in-one-out is the rule that keeps the store at safe throughput for the whole drop. When a shopper completes checkout or their session expires, the next person in line is admitted. Without it, the queue admits a burst at the start and then trickles, which is the worst of both worlds: a checkout spike followed by a stalled line.
Set the session timeout short, five minutes in the waiting room, so abandoned sessions free their slot fast. And plan the sold-out moment. When stock hits zero, the waiting room should say so immediately and offer the next best action: join the waitlist, get notified of the restock, or browse the rest of the store. A sold-out drop with a waitlist is a demand signal. A sold-out drop with a blank page is a lost customer.
Frequently asked questions
How do I calculate safe queue throughput?
Rehearse with real load. Meter shoppers through in steps and stop when checkout errors appear. The number just below the failure point is the safe throughput.
What is one-in-one-out?
When a shopper completes checkout or their session expires, the next person in line is admitted. It keeps the store at safe throughput for the whole drop.
Why do shoppers abandon a waiting room?
Usually slow movement or silence. A line that barely moves, or a room with no stock count and no updates, makes shoppers do the math and leave.
How long should a waiting room session last?
About five minutes. A short timeout frees slots fast when shoppers abandon, which keeps the line moving for everyone behind them.