An NZX incident report pins last Thursday's outage on a "human and process control error".
In short, someone restarted the exchange's trading platform by mistake - and the final couple of hours of the trading day were lost as an overseas support call was required to mop up after the glitch.
An "NZX operator was completing a task that involved restarting a non-production version of Nasdaq ME but had logged into production in error," the report said.
Online services typically have a "production" or live version of their operation, plus a "development" or "non-production" version where tweaks are tested before going live.
The exchange is implementing a new "Privileged Access Management" safeguard across all of its systems as a long-term solution.
The report details how the Nasdaq ME Trading System (which the NZX uses under licence) threw up an alert at 3.08pm on Thursday August 6. The first market alert was sent at 3.23pm.
Multiple attempts to restart failed before a Nasdaq support team in Stockholm was contacted at 3.47pm.
A Nasdaq team was able to advise on how to successfully restart the system, but at 4.35pm the NZX took the decision not to reopen trading "due to concerns around all participants being able to connect in time to conduct an orderly market close".
In April NZX beefed up its new technology sub-committee with the appointment of Nasdaq alumnus Peter Jessup as an independent member.
The move is one of a series of steps being taken by the local exchange as it followed the recommendations of a scathing Financial Markets Authority report issued after the cyber attacks it suffered in August and September, plus technical issues earlier in the year.
The market watchdog, which found a pattern of under-resourcing and under-investment, also said the NZX's next steps needed to include hiring a relationships manager, a chief risk officer, a head of network architecture and a head of IT security.