Uptime

ZapBounce publishes no uptime percentage. We have not run long enough to have measured one, and a number invented for a marketing page is worth nothing to you when you are deciding whether to put this in a signup flow.

Every vendor in this category shows four nines somewhere. Ask any of them what window it covers, what counts as down, and who measures it, and the answer is usually not on the page. A figure without those three things is decoration.

So here is the commitment instead. Once there are ninety days of production traffic we will publish a rolling ninety-day figure, measured by an external prober against the live API, with the definition below and the incident log next to it. Until that day this page stays without a number, which is the same reason we publish no accuracy percentage yet either.

You already hold the data for a figure of your own. Log the status and latency of every call you make to POST /verify. At month end, count the 5xx answers and the timeouts, and leave the 429s out, since those reflect your pace and not our health.

Say you made 40,000 calls and 60 failed. That's a 99.85% success rate by request. It isn't the same thing as uptime by the minute, and the difference is worth seeing once. A 30-day month has 43,200 minutes. If all 60 failures fell inside one three-minute window, minute-based uptime was 99.993%. Spread across 60 separate minutes, the same failures give 99.86%.

Both readings are honest. They answer different questions, so write down which one you measured.

Turn any percentage into minutes before you compare vendors. Ninety days is 129,600 minutes. At 99.9%, a service can be down for about 130 of them and still meet its number. Move to 99.99% and the allowance shrinks to about 13 minutes. Two lines on a pricing page that look almost the same are ten times apart.

Then ask what the promise pays when it's missed. A service credit for the downtime is common across software, and it rarely covers what a stalled signup form cost you. It tells you how much the vendor has riding on the number.

Design for the allowance and stop hoping it goes unused. For a signup form, that means accepting the address when we don't answer. With bulk work, decide how long a list can sit in your queue before somebody needs to know. If a campaign goes out at nine, start the batch the day before and an outage at midnight won't touch it.