I’m currently looking through some best-practices as we look to expand (again) our web analytics and reporting capabilities. There’s lots of info out there about warehousing great swaths of log data, navigation funnels, user behaviors.
“Warehouse” projects I’d been involved with previously had been owned and driven by Technology groups. When I ran across this article, it reïnforced a simple message that’s true for implementations of almost any size: Before you start building anything, ask the Business about their primary goal. From the piece:
Business L (person) – I need a data warehouse!
Geek – Why?
BL – Because I need to know stuff.
G – What stuff?
BL – You know, likes sales data and stuff.
G – Can you express that as a single, English question?
BL – I want to know how many of which products we sell in a month, what it costs us, and how much we sold it for.
G – So you want to look at quantity, cost, and purchase price?
BL – Yes.
G – And you just want to know what was bought , when it was bought, and where it was bought?
BL – Exactly.
G – No problem.
BL – Wait, I need a seperate report sorted by sales region, one for product line, one for…
G – Easy turbo. Let’s just start with one specific question and see how that works before worry about any more.
BL- Thank you very much. I find your responses constructive and supportive. I look forward to working with you.
Can’t you produce a simple prototype very quickly on that information? Wouldn’t it answer a lot of questions that are all basically variations on a single theme?
Can’t you just extract a sample of data and slap it on your dev box and go back to BL in a day or two and show him something?
My project management philosophy is simple and effective:
- Tell folks what you plan to Do
- Do it
- Tell them what you’ve Done
Before you can effectively let the business know what you plan to do, it’s beneficial to get as solid an idea as possible of what the business thinks needs doing.
On this current analytics project, I’m the primary client / user as well as the one building the solution. Even so, I’ve let my team know what my plans are and why; I’ve got their buy-in as I’ve already produced several sample reports – and related actionable suggestions – to demonstrate the kinds of insights and intelligence I’m working to produce on an ongoing basis.
I can now proceed accordingly, produce regular results and suggestions, and hope to see those suggestions put in motion.
I recommend reading the referenced article and especially the comments beneath it. Don’t think it applies only to datawarehouse projects – the ideas work for any number of projects.