What a bad support floor costs, in numbers
Customer operations is treated as a cost line and behaves like a revenue line. Here is how we model the real figure, and what staffing to a forecast changes.
Most companies can tell you the cost per contact and almost none can tell you the cost of a missed one. That asymmetry is why support budgets get cut in the same quarter that churn goes up.
Start with the forecast. Contact volume is not random — it follows your release calendar, your campaign calendar and your delivery promises. If you are staffing to an average, you are overstaffed on Tuesday and failing on Monday.
Then measure resolution, not handling time. An agent who closes the issue in nine minutes is cheaper than two agents who each spend four minutes moving it along. Handling time as a primary KPI reliably produces the second outcome.
The third number nobody tracks: how many contacts were caused by a product decision. We tag every contact by root cause and send the top five to the product backlog each month. On one retail account that removed a fifth of the total volume in two quarters.
AI helps, in a specific and limited way. It drafts, it summarises, it deflects the genuinely repetitive question. It does not handle the angry customer with a legitimate complaint, and pretending otherwise is how brands end up on social media.