Projecting Revenue

Everyone wants to predict future sales, but there isn’t general consensus on the techniques that should be used. Because there’s generally a seasonal curve to the businesses I’ve run, I’ve developed a habit of using the portion of the period that’s already complete, coupled with data from the previous period and how that period ended. Let’s call this the Proportional Method, and it works like this:

By cross-multiplying, we come up with a formula as follows:

So if, at the end of Q1 we want to predict sales for the year, our formula would look like so:

And our cross-multiplication gives us this to solve:

So, if you’re Q1 2016 sales were $1.5M and 2016 Total sales were $9.3M, and Q1 2017 sales were $2.8M, you’d get the following:

Again, by cross-multiplying, we get this problem to solve:

So, in this case, our projection for 2017 would be $17,131,579, or $17.1M.

Of course, the larger the sample period, the more accurate the projection is going to be. In January, with 11 months of the year unknown, small variances will have an outsized effect. Let’s take a look and see how the projection becomes more accurate as the sample size increases. The following are projections based on available data at the time:

As you can see, projections made in the first half of the year are +/- 50%, but projections made in the 2nd half of the year are generally within about 10%.

The question is, is there another variable we can introduce that accurately predicts the deviation of this model? I haven’t found it yet, but I’m definitely on the lookout.