This makes me really mad. Microsoft has done it again. Over and over again they prove to be deceitful, which just shows that they are a company without any sense of moral value. I can’t believe that the world’s largest and most influential company can be so evil. Okay fine - evil might be an overstatement, but I really hate how they conduct business.

I was reading a research paper produced by the Atlas Institute from Microsoft. They are part of Atlas - the advertising section of Microsoft that produces research data to help advertisers [convert into paying-suckers for Microsoft's overvalued products]. The purpose of the research is to show how the Atlas platform is able to more closely predict actual reach with projected reach. They even had the balls to put this up on the first page of the paper:

So what was in the research paper that made me so mad? This…

At first glance, it looks like any digestible graph, however, upon further inspection one can see that the axises (Actual Reach) and (Predicted Reach) do not start with the same distances from the origin. In fact, the bottom axis (Actual Reach) starts slightly to the right of the graph. This would be okay if you were comparing two completely different kinds of values such as time with temperature, but we are talking about the same kinds of numbers here - we’re talking about the number of audiences. To make matters worse, the scale is on a log10 scale! That means the increments in the axis increase logarithmically (with base 10)! A slight shift in the graph would mean HUGE inaccuracies in the graph.

Before I go any further, note that the closer the samples (presented as dots) are to the solid line, the more accurate the prediction reach is to the actual reach. Microsoft has skewed the solid line so that it better matches with the actual numbers!

So, I took the liberty to fix the graph a bit so that the line looks like how it’s suppose to look like:

The blue line that I have drawn is the actual line. The closer the dots are to the line, the more accurate the data is. At the bottom in green, I separated the increment of 10 million into smaller increments of 800,000 to show how off Microsoft was in their mistake.

They were approximately off by 800,000 to 1,500,000 audiences! That’s a huge difference.

Screw you Microsoft. I’m not going to use your stupid-ass Atlas product.

Note: I’m sure that this mistake has gone unnoticed for most executives/managers. Not everyone has the time to look at the graph carefully. And that’s exactly what Microsoft counts on…

- Peter Kao