Let me preface this by saying that I’m a math and statistics geek.
Any of you who have trouble with math, you may as well stop reading right now, because what I’m about to post will make little sense to you. That is not to say I’m trying to insult your intelligence or anything, but if somebody were to blog about comparing operas or ballets, I wouldn’t understand that either.
I was in a meeting earlier in the week, and we were reviewing a report that is used to try to drum up sales for our company. The whole concept of the report is to be able to say, okay, we’ll charge you $10,000 for our services, but we’ll save you $25,000 a year, and then provide tables to back up those figures.
The major table that backs up the figures shows the number of people who we can reach out to, the percentage of people whose behavior we can expect to change, the number of people whose behavior we can expect to change, and the amount of money that will be saved by changing those people’s behaviors.
So here’s the problem, the number of people whose behavior we can expect to change is rounded down to the nearest integer, and then that is multiplied by the savings per person to get the total savings.
So, if we reach out to 100 people, and can expect to change 1.9% of those people’s behavior, the savings shown is based upon 1 person changing their behavior and not 1.9 people (which, in statistics, is a concept known as the “expected value”). I ran the numbers for an entire report, and we’re basically underreporting how much money we can save them by 10%.
I created a nice little Excel workbook showing exactly how the underreporting was happening and how it would hurt our ability to sell our services.
Since then, I’ve been in a pretty long e-mail discussion with a guy who has a PhD in economics, and he just doesn’t seem to get this concept! But, unfortunately, he is a director and I am not, so if I can’t convince him with my last ditch attempt e-mail, I’m going to have to move on.
fixingstupidity 
Saw this after I emailed you. You are right from a mathematically perspective but your Economics PhD Director is right from the perspective of member behavior change. When you are talking about changing behavior it has to happen at the whole individual level not a part of individual level. For e.g. if I was a smoker, then I would have to completely stop smoking (i.e. 1 member stopped smoking) for someone to bank money on my change in behavior. One half, one fifth, one ninth of me can’t stop smoking. From that perspective, you can only bank savings for 1 member for 1.9% rather than 2. If it was 2.5$ then you would bank savings for 2 people. In another instance, Hope this helps.
Oh, and I also suggested that, since the idea of the report was to show economic value, that we should completely omit the number of conversions. This is the Opportunity Assessment report, did you ever work on it?
Maybe, I worked over a lot of reports out there so can’t be sure. For Opportunity Assessment you only have to take the % of members that are not compliant with medical guidelines. That is your pool whose behavior you can change and that’s your Opportunity. 90% for one member is much better than 90% for 2 members (which turns out to be 45% for one member). Member behavior change is black and white, its yes or no. There’s no room for a shade of gray. is your PhD guy by change DN? You have to show past pattern to predict future pattern. If you don’t show conversions, why would anyone want to drink your koolaid at a cost. 🙂
True, Shaema, but the reason I made that point is because, in my opinion, there is basically a 90% chance of converting 2 people and a 10% chance of converting 1.
without conversions, you can show Potential Savings.
One thing that happened in the past, that i disagreed with is that you have to incldue everyone in the pool that you message as your denominator. You can’t exclude people who you had just message so as to increase your percent. Clients didn’t like it that they would not take the entire pool of members who were messaged to calculate their conversion rate.