Now on Twitter
28 October, 2009 § 3 Comments
In the pursuit of writing my next blog post, I’ve wanted to look in to some of the research that was mentioned at StackOverflow DevDays in Toronto, but have been unable to find what I’ve been looking for.
So I created an account on Twitter and asked Joel for his citation. Fittingly, this goes right along with what Greg Wilson was preaching, and I hope that it was just a coincidence, but I do give him credit for re-energizing me and everyone else in the room.
My goal is to look at a study that Joel mentioned in his opening keynote. He brought up a study that was conducted in a California grocery store where shoppers were presented with either 6 or 24 samples of jelly.
| # of flavors | % stopped | % purchased |
|---|---|---|
| 6 | 40 | 30 |
| 24 | 60 | 3 |
I would like to use the naive Bayes theorem to turn this situation around and ask the question, “Given that a customer was going to purchase, what effect did the number of flavors presented have?”
If you haven’t noticed, I’m really loving this Data Mining class I’m taking.
JAWS 
I suppose this is an issue of profit. Suppose you are a jelly connoisseur and like the idea of infinite flavors of jelly, the more flavors the better. But if it is more profitable to reduce the number of flavors then you don’t get to try all those flavors and would be unsatisfied. I like the prospect of a business’s goal is NOT how to squeeze the maximum money and profit from the customers.
In this story, the store still stocked all the flavors of jelly, yet only presented differing numbers of jelly to sample.
What they found was people were more satisfied with the number of items to choose from and made more purchases after sampling when there was less to sample.
I think a store like this could still stock many flavors, yet work on how they display all the flavors so as not to overwhelm the customers.
[…] JAWS Fresher than a Prince from Bel-Air « Now on Twitter […]