Quiet Hours for Accessibility Alerts: When Not to Page Anyone

Not every accessibility finding deserves a 2 AM page. How to set quiet hours and severity routing so alerts reach the right person at the right time.

Accessibility monitoring only works if the team reads the alerts. And the team only reads the alerts if the alerts respect their time. A low-contrast footer link discovered at 2 AM is not an emergency, but plenty of monitoring setups treat it like one, paging whoever is on call for findings that could have waited for morning. The result is predictable: the team starts ignoring the channel, and the genuinely urgent finding, a keyboard trap deployed to checkout at noon, gets missed in the noise. Quiet hours and severity routing fix this. They are not about caring less. They are about making sure the important alerts get through.

Severity is a routing decision, not a label

Most teams assign severity and then send everything to the same channel, which makes severity decorative. Instead, map each tier to a destination. Critical findings, things that block purchase or break a core flow, page the on-call engineer immediately, any hour. High findings open a ticket in the next business day's queue. Medium and low findings go into a morning digest the team reviews with coffee. The mapping should be written down and visible in the alert itself, so nobody has to decide at midnight whether a contrast failure on the returns page is worth waking someone. Decide once, in daylight, and let the routing do the work.

Set quiet hours around the release calendar

Quiet hours should follow the store's rhythm, not just the clock. Overnight is the obvious window: few shoppers, no developers awake to act. But the release calendar matters more. During a code freeze, new findings are almost always pre-existing issues surfacing in a fresh scan, not new breakage, so route everything to the digest. In the hour after a deploy, invert the rule: watch closely, because that is when regressions land. Holiday traffic peaks deserve their own policy: page only for checkout blockers, digest everything else. The principle is simple. Alert urgency should match the team's ability to act, and nobody can act well at 3 AM on a finding that will still be there at 9.

The digest that replaces the midnight ping

A digest only works if people read it, and people read digests that are short and actionable. Cap it: five findings max, each with the page, the barrier in one line, and the fix in one line. Link to the full report for the rest. Send it at the same time every morning so it becomes a habit, not an interruption. And keep a simple rule for the digest itself: anything that appears three mornings in a row without being ticketed gets escalated to the weekly review. That prevents the slow fade where medium findings sit in the digest forever, quietly becoming permanent.

Review the routing every quarter

Alert routing drifts. The store adds a new template, the team grows, the definition of critical quietly expands until everything pages again. Once a quarter, pull the numbers: how many pages went out, how many were acted on within the hour, how many digest items got fixed within the month. If pages are being snoozed, the threshold is wrong. If the digest is growing, the team needs capacity or the bar needs raising. Routing is a living configuration, not a one-time setup, and the quarterly review is what keeps it honest.

Frequently asked questions

What actually deserves a middle-of-the-night page?

Anything that stops shoppers from completing a purchase right now: a keyboard trap in checkout, a payment form that cannot be operated without a mouse, a newly deployed page that screen readers cannot parse at all. The test is revenue plus immediacy. If it can wait until morning without costing orders, it waits.

Should quiet hours be different on weekends?

Usually yes, because weekend staffing is thinner and deploys are rarer. Many teams run a stricter page policy on weekends, critical checkout blockers only, and let everything else accumulate for the Monday digest. The exception is peak sales weekends, when the revenue cost of a blocker is highest and the page threshold should be at its most sensitive.

How do you keep low-severity findings from being ignored forever?

Give them a shelf life. Anything in the digest for more than two weeks without a ticket gets bundled into a monthly cleanup sprint. Track the age of the oldest untriaged finding as a team metric. Old findings are usually either trivial, in which case close them, or quietly important, in which case the aging rule surfaces them.

Our client wants every alert the moment it fires. What do we tell them?

Show them the fatigue curve. Teams that get paged for everything start ignoring everything, usually within weeks, and then the critical alert gets missed. Offer the compromise most mature teams land on: instant pages for criticals, a same-day ticket for highs, and a morning digest for the rest, with a weekly summary the client can read. They get full visibility without destroying the signal.