At what point does a four day access request SLA become an engineering team's problem to solve themselves
We rolled out a formal access request workflow last year proper approvals, audit trail. security was happy engineers were not.
four days average from request to access. not because approvals are slow but because requests route to resource owners who have other priorities and the escalation reminders don't really work.
two months in someone built a slack bot that auto-provisioned access to six internal systems based on team membership built it because four days is too long when you're blocked mid-sprint. no audit trail, nothing IT could see.
bot ran for six weeks before IT found it 34 engineers provisioned across systems that should have had individual approvals. IT shut it down. the engineers who built it said the formal system was the problem not the bot.
they were right about the four days.
every time someone raises it with security they come back to the same point remove the friction and you remove the control.
has anyone actually found a way to make this work or is four days just the tax you pay for having governance at all.