# Variance _Know when to react_ --- July revenue comes in at £1.1m. Prior month was £1.25m - down 12%. Prior year July was £1.2m - down 8%. Two negative variances in the monthly pack. What's wrong? Actions get assigned. The sales team reviews pipeline. Finance are asked to pull a customer-level analysis. A meeting goes in to review findings. --- Maybe nothing. June was a quarter-end month. Deals were pulled forward to hit targets, which inflated June and hollowed out July. Last year's July included a one-off implementation fee from a contract that didn't repeat. Plot a rolling twelve-month average and revenue is growing. August comes in at £1.26m - nothing's happened, apart from a lot of distracting busywork. --- Every variance is measured against something, and the something is usually chosen by default rather than by design. Month-on-month assumes all months are equal, when they differ by working days, seasonality, and period-end effects. Year-on-year assumes the prior year was representative, when it might have included one-offs, pricing changes, or other quirks that have long since been forgotten. --- A twelve-month trend chart with a band around the average tells you more in five seconds than any snapshot variance table. A new point inside the band gets noted. A point outside, or three months trending in one direction, gets investigated. The conversation should be "is there something unusual happening?" instead of "why is July down on June?". Most months, the answer is no, and the meeting moves on quickly. No root cause. No action items. No one to blame. That's okay.