quicksand

Building Models on Quicksand

  |  February 28, 2013   |  Comments

How often do you start with a clean slate and rebuild your model because the variables themselves have changed?

My good friend, frequent mentor, and occasional network diagnostician Mark Gibbs sent me the following quote the other day and it rang a bell. Several of them. I knew I would have to explore the resonance here.

The only man I know who behaves sensibly is my tailor; he takes my measurements anew each time he sees me…The rest go on with their old measurements and expect me to fit them."
- George Bernard Shaw

tailor

Three bells went off at the same time when I read that. And they harmonized.

1. Fails expectations. The people asking for numbers know what they want the answers to be. If the numbers are not in line with pre-conceived notions, these people naturally question:

  • The data
  • The data collection method
  • The definition of the data that's been collected
  • The time frame in which the data was collected
  • The analysis
  • The analyst
  • The weather
  • The I Ching

2. Been there, done that. The team leader has solved that problem already and sees no reason to go back to revisit it. Once the problem is solved, it's not only no longer a problem, it's no longer interesting. The world may have changed and time marched on, but the team leader will assure you that we don't need to spend any more time on this issue. Asked and answered. Fixed. Mischief managed. Why are you bothering me with this?

3. A worldview is a terrible thing to waste. This is actually the same as number two, but with one important difference. This time, the old stick-in-the-mud is you. And there's good reason for it.

As humans, we thrive on learning the things we don't need to pay attention to. That noise in the night? The neighbor's dishwasher. That scathing email from a customer? Oh, it's that customer. That error message in your browser? Not your fault.

We make many assumptions in order to function: gravity; time tables; the fury of a woman scorned. And then we make assumptions that are harmful: the person in the other car knows I'm going to turn; using Bcc in emails is OK; thinking that there are some variables we can ignore while engaged in marketing analytics.

We know that this segment of our prospects always responds this way to this sort of offer.

It's a fair assumption. After all, it might be the very definition of that segment. But how often do you dust off the cobwebs and start with a clean slate?

"Most of our assumptions have outlived their uselessness."
- Marshall McLuhan

We have entered into an era of marketing mix modeling, predictive analytics, and chasing the e-commerce genome. Over time, by necessity, we build models on top of models of the world around us in order to predict where our marketing investments will best pay off.

mathematics-formula

Each model is built upon assumptions and validated against real data. But how often do you start with a clean slate and rebuild your model - not because the values of the variables have changed but because the variables themselves have changed?

I propose twice a year. Think of it as spring/autumn cleaning. Take a good, long, hard look at the assumptions that went into the models on which you rely and try to look at them from a new angle. Then bring in some new people to help you.

Make sure those people have a different background from you and a different worldview. Make sure they have difference assumptions. The forecast/prediction/model/rear end they save might be yours.

"Your assumptions are your windows on the world. Scrub them off every once in a while, or the light won't come in."
- Isaac Asimov

Tailor, Mathematics Formula, and Quicksand images via Shutterstock.

Tags:

ClickZ Live Chicago Join the Industry's Leading eCommerce & Direct Marketing Experts in Chicago
ClickZ Live Chicago (Nov 3-6) will deliver over 50 sessions across 4 days and 10 individual tracks, including Data-Driven Marketing, Social, Mobile, Display, Search and Email. Check out the full agenda and register by Friday, August 29 to take advantage of Super Saver Rates!

ABOUT THE AUTHOR

Jim Sterne

Jim Sterne is an international consultant focused on measuring the value of the online marketing for creating and strengthening customer relationships. Sterne has written eight books on using the Internet for marketing, produces the eMetrics Marketing Optimization Summit and is co-founder and current chairman of the Digital Analytics Association.

COMMENTSCommenting policy

comments powered by Disqus

Get the ClickZ Analytics newsletter delivered to you. Subscribe today!

COMMENTS

UPCOMING EVENTS

Featured White Papers

BigDoor: The Marketers Guide to Customer Loyalty

The Marketer's Guide to Customer Loyalty
Customer loyalty is imperative to success, but fostering and maintaining loyalty takes a lot of work. This guide is here to help marketers build, execute, and maintain a successful loyalty initiative.

Marin Software: The Multiplier Effect of Integrating Search & Social Advertising

The Multiplier Effect of Integrating Search & Social Advertising
Latest research reveals 68% higher revenue per conversion for marketers who integrate their search & social advertising. In addition to the research results, this whitepaper also outlines 5 strategies and 15 tactics you can use to better integrate your search and social campaigns.

Jobs

    • Sales Planner
      Sales Planner (Verve ) - New YorkAbout Verve   Verve is the leader in location powered mobile advertising. We manage one of the largest mobile...
    • Systems Analyst/Support
      Systems Analyst/Support (Agora Inc. ) - BaltimoreIRIS (Increased Revenue Intuitive Software ) is proprietary software that helps marketers...
    • Client Services Support Specialist
      Client Services Support Specialist (Agora Inc. ) - Delray Beach OVERVIEW:  This position requires a highly motivated and resourceful individual...