Back in 1990, you'd feed a 5.25 inch floppy disk into your IBM-AT desktop computer. You'd fire up SPSS. At the bottom of the screen ticked the number of records that were being processed .... 20 records ... 40 records ... 60 records ... 80 records.
Today, I loaded a large dataset, wrote six hundred lines of code, and began processing the information. At the bottom of the screen, I could visualize the records as they were being processed ... 200,000 records ... 400,000 records ... 600,000 records ... 800,000 records. Thank you, Acer, AMD and SPSS for providing a fun computing environment.
It's a blast to see new data, information you haven't been exposed to before. The data at Nordstrom seldom changed during my six years there. Sure, occasionally the annual retention rate would vary (maybe 67.2% one year, then 69.3% the next year --- if you are at Macy's or Neiman Marcus or Saks, sorry, those aren't the actual numbers). New customers might vary by fifty thousand verses forecast. There were seldom huge surprises. Such is the case when a business consistently meets or exceeds expectations.
When you get to see new data from a new company, there is a sense of exhilaration. It is like opening up a box of puzzle pieces. You find the corners and the border pieces first. Each piece fits into another piece. Eventually, the pieces provide a path for you to get to the end of the assembly process.
Each line of code produces reporting --- the reporting tells a story. Existing customers are retained, lapsed customers repurchase, new customers feed the future growth of a business. Within minutes, the lifeblood of a company is evident on your computer monitor.
Looking across years, you can visualize the decisions that executives made, decisions that caused increases or decreases in customer counts. These increases or decreases drove subsequent decisions, which drove increases or decreases in customer counts. Within an hour, Multichannel Forensics illustrated what happened, and forecast what is likely to happen in the future. Suddenly, there is a story to tell.
There is a huge difference between Data Mining and Storytelling. Data Mining seeks to explain the data. Storytelling is an art form that translates information in a way Executives can digest, understand, and act upon.
Data Mining has a place without Storytelling. Data Mining coupled with Storytelling yields potential. Data Mining and Storytelling that speaks directly to a current, future or perceived Executive need (as defined by the Executive) causes change.
Helping CEOs Understand How Customers Interact With Advertising, Products, Brands, and Channels
April 24, 2007
Fresh Data And Storytelling
Subscribe to: Post Comments (Atom)
Many of you have forwarded articles to me outlining how your local mall is considering taking empty space and turning it into pickleball cou...
It is time to find a few smart individuals in the world of e-mail analytics and data mining! And honestly, what follows is a dataset that y...
Sometimes you think "people already know this stuff". Sometimes you realize that Google Analytics give smart analysts almost no op...
If you want to understand why clients don't trust vendors and trade journalists, read this little peach from a week ago: Direct Mail is ...
Good point about storytelling vs datamining. Also, what I constantly point out to my bosses is the MISSING story from sales data. I work for a vendor who sells to a wide variety of retailers and catalog companies. One of the mysteries that data doesn't show is when a product line does very well for one retailer but not for another all factors being the same(region, demographics, store type). When we visit the stores we see two main differences===first is presentation---one has our line prominently displayed or at least that department well done, and the poor selling retailer doesn't. The other difference is how well sales staff is trained and present(couldn't find a department associate for several stores in the poor selling one).
This is not something you can determine from just data mining.
And it hurts because the corporate buyer from the poor retailer just sees the product line not selling and says it is a losing line. We aren't privy to the other vendors' data so...impression is ours is the bad line. That leads to increase in "damages", destroy authorizations, and in markdown allowance, etc. as retailer tries to get rid of the inventory that didn't sell. The bad part is that the buyer looks just at that small piece of data and sees only one side without the story.
On another tangent, I am now reading Lean Solutions by Womack and Jones. Have you read this before? I realize book is two years old, but curious what you thought of the book if you read it. Thanks.
I have yet to read that book, sounds intersting.ReplyDelete
It would be interesting to be on the vendor-side of the equation, to see products work well for some retailers, and not for others. Interesting, and frustrating.
Would have guessed you were in the SAS tribe of heavy-lifting software -- perhaps a future post might compare and contrast SAS vs SPSS?ReplyDelete
I can go either way on the SAS vs. SPSS stuff. I used SPSS first, back in college in the mid-80s, so that is why I'm leaning that way now.ReplyDelete
Either tool gets the job done!