Decision Mechanics

Insight. Applied.

Superconsumers

By

The Economist’s Schumpeter column recently cautioned that

Research shows that firms ignore passionate consumers at their peril

These “superconsumers” represent 10% of all customers, but are responsible for 30-70% of sales, an even greater share of profits and almost all consumer insights. In addition, they influence their social circle with their unbridled enthusiasm.

The article offers advice to companies wishing to engage with their superconsumers. First, identify them through data analytics—they are easy to find. Second, reward them—e.g. through loyalty programs.

I’ve written previously about companies that focus on their best customers, as opposed to trying to appease unhappy ones. Superconsumer research seems to support such a strategy.

Is wrong better than nothing?

By

Last week Italians voted “No” to constitutional reform. The Economist reports that

The 20-point margin of defeat—–60% to 40%—–was almost double what pollsters had foreseen.

Yet again, polling has failed. However, come the French and German elections next year who doubts that we’ll continue to celebrate or denounce each poll result with partisan fervor?

There’s a hunger for this kind of information and consumers will take something over nothing—even if that something isn’t telling them very much.

I recently wrote about how the Idiosyncratic Rater Effect renders a lot of human resource data next to useless—yet organizations continue to collect and act on it.

If we want things to change it seems we can’t just point out that the emperor has no clothes—we have to dress him.

All maps are wrong

By

Map of the world

Maps of the world are a perfect example of how difficult it is to prevent bias creeping into your modeling.

People always have a perspective. And it’s difficult to eliminate or compensate for it.

Vox have a great video that discusses the difficulty of producing an accurate map of the world. It starts with the presenter slicing up an inflatable globe and trying to spread it out across the floor—with limited success.

You can see the impact of different mapping approaches in a gallery on Wikipedia. And, if you want to see how the dominant Mercator projection mangles the size of countries, there’s an interactive tool that illustrates the effect. Try dragging Greenland onto Africa for a dramatic example.

The National Geographic Society have apparently adopted the Winkel tripel projection due to the trade-off it makes between preserving shape and size.

Effective data science is often about choosing the appropriate trade-offs.

Supporting data users

By

stress photo

A critical, but overlooked, area of data science is linking analytics to action. Decision-makers must

  • Understand the information being presented to them
  • Have confidence in the analysis
  • Be able to place information in the context of the decisions they are required to take
  • Understand how to apply the information to make better, more informed, decisions

Too much focus has been placed on the needs of the analysts—and this is stifling the overall effectiveness of data science initiatives.

Brian Williams has produced an article—Creating an analytic culture through data interaction—discussing this topic. He has some interesting things to say.

He defines the problem as follows.

The complexity of information environments is challenging most organizations. Not all decision makers are ready to become more analytic, and as we focus on visualization, data mining, and predictive analytics, decision makers will become even more reliant on analysts and may not perceive they have the abilities to keep up with the quants.

One area of data science I’ve found to be weak is visualization. Many data scientists assume that charts are visualization tools that can be presented to decision-makers. Wrong. Charts are tools for data scientists. Managers may not have the time and/or the skills to draw conclusions from a chart.

The article has this to say on visualization

When analysts design visuals without user interaction in mind, their charts and infographics simply are a static sharing of their own analysis. […] Design needs to be sufficient enough to be a starting point for a decision maker that invites interaction.

Tooling clearly has a major part to play in bringing visuals to life.

Managers who are clued-up about data science—and its strengths and limitations—will be able to direct analysts to produce more actionable insights. There’s a constant cry for this sort of guidance from data scientists who feel they have the tools, but not the questions.

If we are to realize the substantial potential of data-driven decision-making we need to develop suppliers and consumers in lockstep. The current focus on supply-side tooling and training will result in disillusionment and throwing the baby out with the bathwater.

Scaling knowledge

By

The data team at Airbnb have written an interesting article on how to manage data science research as you bring more and more people on board.

They developed an internal process and tool based on five key tenets

  • Reproducibility—There should be no opportunity for code forks. The entire set of queries, transforms, visualizations, and write-up should be contained in each contribution and be up to date with the results.
  • Quality—No piece of research should be shared without being reviewed for correctness and precision.
  • Consumability—The results should be understandable to readers besides the author. Aesthetics should be consistent and on brand across research.
  • Discoverability—Anyone should be able to find, navigate, and stay up to date on the existing set of work on a topic.
  • Learning—In line with reproducibility, other researchers should be able to expand their abilities with tools and techniques from others’ work.

Their tool is built on top of Git, Jupyter notebooks and (R)Markdown.