Granted, the application of statistical analysis used in baseball (specifically, Sabermetrics, invented by Bill James) seems to be a little “out there”. And it is, kind’ve. But using the data available lets us do more than simply report the measurement of observations of our campaigns. You can take these observations and apply the numbers in ways that help you properly judge past efforts, and see what you need to work on to improve future performance.
For example… Runs Created. Bill James came up with a system that could grade a player’s offensive ability regardless of if he is a speedster who walks and steals a lot, a high-average singles hitter, or a low-average slugger. the essential computation of Runs Created is:
RC=(H+W-CS)X(TB+.7SB)/AB+W+CS
Runs Created equals hits plus walks minus times caught stealing multiplied by total bases added to seven-tenths of bases stolen divided by the sum of at bats plus walks plus times caught stealing. Sounds a little goofy (especially having Caught Stealing on both sides of the equasion). But it woks. The deviation from real life is a scant 4% and generally within 3%. There are a couple exceptions, but overall you can take this formula, apply it to an entire team over the past 30 years or so it’s very, very close. Which is to say after 20 or so games into a season you can begin using this to predict how many total runs a team will score for the season and get pretty close. Not Vegas Bet close, but close.
Now, how do we apply this to Affiliate Marketing? Let’s look at the information we have available to us. We have Traffic, ClickThru, Sales, Visits, Pages Viewed, etc. Traditionally we have all been measuring ourselves with CTR, CPC, etc. Which are extremely simplistic measures. What if we apply a little bit of Sabermetric thinking and come up with our own measurement for Runs Created? How about Sales Created?
Let’s lay out a possible scenario:
SC=(Clicks-Bounces)X(Page Views+Uniques)/(Visits+Clicks+Bounces)
This would reward Clicks and Uniques, penalize Bounces, and rate it all against the total number of Pages Viewed. Run some of your stats over this and see what the ratio of the resulting number (SC) is compared with the total number of sales made… I’m seeing some results that are making sense. The first site I compared this too was spooky. I had cleared 74 sales on an offer. This example came out with a score of 72.50. Thus “predicting” the number of sales within 2% of actual.
I compared it with a few other sites, called a couple friends and got some example numbers. So far, it’s doing a reasonable job of computing how many sales based on the numbers and statistics alone. The interesting thing is that one of the examples was high traffic with high bounce rates, another had low traffic with very low bounce rates… And the numbers held up. I don’t believe this is the final form of this. Let’s not kid ourselves, this is counter-intuitive and I spent less than a week on this. But a logical conclusion to this little experiment is that we arn’t recording and analyzing the data we’re already gathering in effective ways.
We Affiliate Marketers need to start really looking at the data, the history, and figuring out better ways to measure what is really happening with our offers, etc. Judging ourselves by the observations made (logs), and discovering ways to make ourselves better.
Start discussion.....