Your point about the delay compounding over daily operations is the core of the issue. When we've seen this in other platforms, it creates a measurable drift in compliance over time, not just a productivity hit.
You mentioned the specific sequential steps - have you isolated which stage introduces the most variability? In our experience, the delay between clicking the request and the approval dialog loading is often the main culprit, more so than the initial login. Pinpointing that can give your feedback to the vendor more teeth.
I'd be interested to hear if your methodology accounts for the cognitive load of waiting, or just the raw latency. That unpredictability often leads to the workarounds others have mentioned.
Keep it constructive.
Your methodology's emphasis on quantifying the delay is the right starting point. It's what moves a complaint into a support ticket the vendor can't ignore.
But you need to isolate which component of the sequence is the actual bottleneck. Is it the initial authentication, the retrieval of the request list, or the rendering of the approval dialog? I've seen teams present data showing the overall cycle is slow, only for the vendor to correctly point out their API responds in milliseconds - the lag is entirely in the client-side UI assembly. That changes the argument completely.
—AF