The revenue trend is the first chart most people look at and the easiest to misread. This guide is for anyone using the board daily. It covers what each line represents, how to compare periods honestly, and the two reading mistakes that lead to the wrong decision most often.
The three lines
Actuals are what the business earned, taken from the store rather than from any advertising platform. This line only moves when an order settles, and once a day has closed it does not change except for refunds applied to orders in it.
Forecast is what the model expects, built from the previous month’s real result plus your assumptions. It extends past today.
Year on year is the same calendar period twelve months earlier. It is the most useful comparison an ecommerce business has, because it holds seasonality constant. Comparing November to October tells you that Christmas is coming. Comparing this November to last November tells you whether you are growing.
The last day is usually a trap
The most recent day on the board is yesterday, and it is complete. Today is deliberately absent, because a part day against a full day is a comparison that always looks like a collapse.
There is a second version of this trap that catches people more often. When one source lands later than the others, the board still shows the latest day revenue exists for, and the metric that has not arrived reads zero for that day. A final data point that drops to the floor is nearly always a sync that has not run yet rather than a business event. Check whether the source is current before you react to the shape of the line.
Compare like for like
Ecommerce is weekly in a way that is easy to forget. Weekends and weekdays behave differently for most brands, so a seven day window compared against another seven day window is honest, and a five day window compared against a nine day window is not.
The comparison selector at the top handles this. Previous period matches the length of the range you are looking at. Same period last year matches the calendar. Use previous period for operational questions such as whether last week’s change held, and same period last year for questions about growth.
Two mistakes worth avoiding
The first is reading a single day. Daily revenue for most brands has enough natural variance that any one day tells you almost nothing. A day that is thirty per cent down is often inside normal variation. Read seven day rolling figures for operational decisions and monthly figures for strategic ones.
The second is treating a forecast divergence as a forecast failure. If actuals run below forecast for a week, the useful question is which input changed. Revenue is the output of traffic, conversion rate and average order value, and the trend line cannot tell you which of the three moved. The summary can.
What to do with it
Look at the trend to decide whether something needs attention, then leave it. It is a detection surface, not a diagnostic one. When the line does something you did not expect, the answer is in the sections underneath it: efficiency if spend changed, the website funnel if conversion changed, retention if the mix of new and returning customers changed.
The trend tells you that something happened. It very rarely tells you what.