Spreadsheet Solutions

It doesn't have to be a spreadsheet, it's short-hand for a neat little model or tool that doesn't require a beefy stats package to create.

Most people use Excel on a daily basis, but few use it its full potential. You might not even know that it can figure out whether two bits of data are correlated, how a range of inputs explains an output and even forecast trends into the future.

Do either of the below sound like they might be useful to you?

Problem 1 - Les is a bookie. He's been having a bit of trouble recently with people not paying up though. Not being the sort to get heavy with them (he's a lover not a fighter) this is costing him a lot of money. He's heard that the bigger high street bookies have computer systems that can predict how likely a customer is to not pay up on their bets. He wants one of those!

Solution 1 - A quick bit of digging in Les' customer database shows that the factors that best describe whether or not a client will default are their age, gender, previous betting history and the postal district they live in. Building a regression model on this data a spreadsheet is created into which Les can quickly enter his customers' details. The cell next their name turning red, amber or green will tell Les whether or not they are likely not to cough up. Neat!

Problem 2 - Candice has overhauled the creative she uses for the mailing campaign for her mobile disco. She did the right thing and tested the new (test) creative against the old (control). 50% received the test and 50% the control. She bit her nails and waited for the results to come back. 11% of the people that got the control responded vs. 12.5% who got the test. That doesn't sound like much of an improvement, did it really do better?

Solution 2 - Alas Candice, no, it didn't do any better. Statistically that is. Never fear though, there's every chance that it really did do better in the real world, but statistically we can't be confident in that because you didn't send it to enough people. We'll give you a spreadsheet that will do two things for you. Firstly, enter the response rates you expect and the degree of statistical certainty you want in your result and, hey presto, it will tell you how many people you need to send it to in order to get a result you can trust. Secondly, plug the results of your tests into it and it will tell you whether your test was a winner. Now you can test to your heart's content and be sure that you're making the right call!

Complex problems don't always need a complex solution...

Get in touch with us here to see if we can deliver you a neat solution that will inject some real intelligence into your business. Alternatively, just take our free advice on how these solutions can help. You're free to take it or leave it.