Facebook, Instagram, and WhatsApp went dark again on July 22 — the second time in 72 hours.
Meta's three flagship platforms failed together, in the same afternoon window, for the second time this week. Users worldwide got logged out, saw "account temporarily unavailable," and could not refresh feeds or send messages. Service returned within one to two hours. The cause has not been officially confirmed.
WHAT ACTUALLY HAPPENED
Around 13:00 GMT on July 22, monitoring logged a spike in failures across Facebook, Instagram, and WhatsApp at once, hitting users across Europe, the UK, the US, and beyond. This mirrors an outage just three days prior — same platforms, same pattern, inside one 72-hour window.
Meta has not published a root cause. Analysts have floated a DNS resolution failure, an internal backend configuration error during maintenance, or a BGP routing fault. None has been ruled out, and Meta has not said whether the platforms share the failed component or failed independently.
WHY THE REPEAT MATTERS MORE THAN THE OUTAGE
A single outage is an incident. A second failure of the same three services within 72 hours, with no disclosed cause after the first, signals whatever broke was never fixed — only masked or worked around. Three products users treat as separate apps went down together twice, confirming they share upstream infrastructure deep enough that isolating one from the others' failure was not achieved, or not attempted.
For any organization running multiple customer-facing services, the lesson isn't "outages happen." It's that an unexplained recovery is not a resolved incident, and a repeat failure on a short interval means the postmortem missed the real cause.
WHAT TO DO NOW
DO NOT CLOSE AN INCIDENT ON RECOVERY ALONE — restoration without a confirmed root cause means the fix was luck, not a repair.
MAP WHICH SERVICES SHARE FAILURE DOMAINS — if three "independent" products go down together twice, they aren't architecturally independent; document the dependency now.
TREAT A SECOND OUTAGE WITHIN DAYS AS HIGHER-PRIORITY — recurrence within 72 hours should trigger executive review, not another ticket in the same queue.
PUBLISH WHAT YOU KNOW, EVEN INCOMPLETE — silence between incidents lets customers fill the gap with worse explanations than the truth usually is.
Two outages in three days across three platforms is not bad luck. It's an open incident that never actually closed.
If your last outage recovered clean but you never confirmed the root cause, is it actually resolved — or just quiet?