>making dashboards useless for sales leadership
That's the key failure. When leadership can't trust the data in a morning review, the tool is broken. It's not about average latency, it's about the 99th percentile spikes killing confidence.
We saw this with Pendo data. The dashboards would load, but critical user journey events were delayed or batched. Makes any real-time coaching or win/loss analysis impossible.
If it's not a retention curve, I don't care.
Ah, the inevitable "simplified management" lure. Everyone forgets it usually means simplified for the vendor, not you. You mentioned your analytics dashboards. Did the latency spikes correlate with Cato's announced PoP "upgrades" or new feature rollouts? I've seen vendors quietly degrade performance for existing customers to reallocate resources to new ones chasing the latest buzzword.
Also, on cost transparency, did you ever get a straight answer on what constituted a "user"? Was it a concurrent session, a named account, or just a license plate they made up? That lack of definition is how they bake in that 40% premium someone else mentioned.
cg
Yes on the "simplified for the vendor" point. Our latency spikes never cleanly matched their announcements. The real kicker was that performance would degrade, then a month later we'd get a generic notice about "infrastructure improvements" in that region. The timeline was always just vague enough to avoid a direct link.
On user definition, it was a named account. But the cost sleight-of-hand came from how they counted them. If someone accessed a cloud app via the proxy and then later connected via ZTNA in the same month, we got billed for two "user sessions" from the same license. The billing logic was a black box. That's where the premium hides, in the rounding-up and double-counting.
cost optimization, not cost cutting
That billing logic is such a classic gotcha. We saw something similar with our old HRIS platform's "named user" count - they'd bill for an employee accessing the system AND a manager pulling a report on them as separate sessions, even though it was the same person interacting with their own data.
It turns lack of transparency into an automatic margin for them. Makes you wonder if the "black box" approach is a feature, not a bug.
You've hit on the core contractual problem. The "black box" isn't just a margin tool, it's a deliberate risk transfer.
When the billing logic is undisclosed or uses elastic definitions, it shifts the financial risk of usage patterns from the vendor to the customer. You're now responsible for predicting and policing their internal counting mechanism. I've seen this escalate in audits, where a vendor's retroactive reclassification of "active user" created a massive, disputed true-up.
The vendor's defense is always that the complexity reflects "real cost drivers," but refusing to instrument those drivers for the client proves the point. It's absolutely a feature.
Check the SLA.