What happened at Workday
On Friday 13 November 2026 (US time), Workday was checking the weekly maintenance window for its disaster recovery tenants when services at the recovery site unexpectedly connected to production databases. The Disaster Recovery Journal reports that the check was separate from Workday's formal twice-yearly recovery test. Workday's notice, which Florida Atlantic University's IT office republished, said Workday had put all WD1 production tenants into maintenance and that production was unavailable, with no restoration time given.
Workday told customers the fault had left some data "in an inconsistent state". UC Today's timeline says scheduled jobs and integrations could fail, along with tasks that hit missing-transaction errors. Staff at affected organisations couldn't log in, clock in or out, check their pay or request leave. Downdetector logged nearly 3,000 reports on Friday afternoon, Pacific time. A Workday spokesperson said the incident wasn't a cybersecurity event.
Service came back in stages. FAU's final update marked Workday Production as available at 10:30 am Eastern on Saturday 14 November. An earlier Workday update warned that reindexing time varied from tenant to tenant. Some customers came back later than others, and one US college pushed out its benefits enrolment deadline. Workday has more than 11,500 customers but hasn't said how many were affected. Workday also hasn't explained how recovery services were able to reach live data, and the root cause is still preliminary.
Why a US outage matters in New Zealand
Workday lists Beca, Mercury, Heartland Bank, Westpac New Zealand, NZTA and Fire and Emergency New Zealand among its local customers. Workday hasn't said whether any New Zealand tenants sit on WD1. UC Today's reports came mostly from the US and India. A local Workday customer may not have noticed the outage at all. The same failure could still happen on any multi-tenant SaaS platform, whichever region hosts it.
Most resilience planning for cloud ERP and HR assumes the provider's recovery site is the safety net. The Workday incident shows a second risk: the recovery environment can damage production when isolation between the two fails. Workday's recovery tooling connected to live databases during a routine check. Nobody had declared a disaster, and no test was underway. Customers had no warning and no way to step in. Once service returned, the work moved to customers, who had to reconcile timekeeping and payroll records for the hours when staff couldn't record anything.
New Zealand employers face a particular exposure here. The Holidays Act requires accurate records of hours and leave, and payroll errors are expensive to remediate after the event. If a payroll cut-off falls inside an outage window, a firm has to choose between processing pay on stale data and delaying pay. Neither choice is good, and the time to decide is before the outage. Firms on Xero, MYOB Acumatica, NetSuite, Business Central or Dynamics 365 rely on the same provider-run recovery arrangement. They also rarely see the provider's change controls.
What the contract and status page won't tell you
Most SaaS agreements describe availability targets and service credits. Few explain how the vendor separates recovery environments from production, or how quickly the vendor confirms data integrity after an incident. Workday's own notices show the gap. Customers knew within hours that production was down. The question that mattered for payroll, which records were left inconsistent, took longer to answer because reindexing ran tenant by tenant.
Integrations make the problem harder. Many mid-sized organisations feed a SaaS HR or finance platform from timeclocks, rostering tools, banks and a separate payroll engine. When the core platform goes into maintenance, upstream systems keep sending data, and scheduled jobs either fail quietly or queue up. The cleanup usually takes longer than the outage. Firms that log integration runs and keep a list of every interface in one place can find the failed jobs in minutes. Firms without those records end up checking each connector by hand.
Practical next steps
The checks below suit any organisation running payroll, timekeeping or finance on a SaaS platform. Most can be done before Christmas payroll runs.
- Ask your vendor where your tenant runs and how recovery is isolated. Ask which data centre hosts your production tenant, where its recovery site sits, and what technical controls stop recovery services from writing to production. Ask for the formal root cause analysis of any incident affecting your tenant, and put the request in writing.
- Read your contract for data integrity commitments. Service credits for downtime don't pay for staff time spent reconciling records. Check whether the agreement says how and when the vendor confirms data consistency after an incident, and raise the gap at renewal.
- Write a manual fallback for pay and time. Decide in advance how staff will record hours during an outage, who approves manual time sheets, and what happens if a pay run falls inside the window. Keep a recent export of employee pay details somewhere the platform can't touch.
- Map and monitor your integrations. List every interface into and out of your ERP or HR platform, with an owner and an alert for failed runs. After any vendor incident, rerun or check each interface against that list.
- Treat vendor incidents as your own incidents. Log the outage in your service management tool, record its business impact, and review it the way you would review an internal outage. Our ITIL 4 service management work covers incident and supplier management processes that fit mid-sized teams.
- Test reconciliation as well as restoration. At your next continuity exercise, include the step of reconciling a day of missing timekeeping and finance transactions. The test shows how long the cleanup takes and who has to do it.
For organisations reviewing a current ERP or choosing a new one, recovery isolation, integration monitoring and post-incident data checks belong in the selection criteria from the start. Our ERP advisory and implementation page explains how we approach platform selection and integration design for NZ and Australian firms.
Sources
- Five Outages in Four Days: What the FAA, Microsoft 365, Workday, United, and Chase Failures Tell Us (Disaster Recovery Journal)
- Workday's Backup Test Took Payroll and Timekeeping Offline (UC Today)
- Florida Atlantic University OIT notices, 13 November 2026
- Workday for New Zealand
- ERP LEADERS: Workday's backup systems connected to its live databases
Want to know what this means for your organisation? Our team in Auckland can show you your top three opportunities in a free 30 minute assessment.
Get my free assessment How we help