Musing on challenges of measuring

My hero, Peter Drucker, is often quoted to say (I paraphrase here) “What you cannot measure, cannot be managed” and this idea inspired many analytic initiatives by large companies as well as by budding startups, like this one. There are hundreds of companies that monitor, listen and analyze every aspect of web traffic, impact of media messages, both digital and analog, and just about anything else under the sun. There is surely no shortage of technology and tools, and current interest from businesses and consumers is quite high, but…there is still not enough conclusive evidence that measuring and managing to the specific parameter can produce measurable result. It often is still a challenge to interpret measurements into predictive models, that produce or support specific actions or decisions. Perhaps it is just my personal, limited experience and I look forward to be proven wrong  in your comments, but for now I would like to propose a few potential reasons for these disappointing experiences.

Is it possible that we often measure wrong things? Many people would argue that NPS (Net Promoter Score) is a meaningless thing to measure and the Social Media influence, measured by Klout and others, does not translate into any specific action. We often measure what is easy to measure, listen to what is easy to hear, without a difficult effort of understanding and interpreting into an action that can produce measurable improvement. Many people find it easy to identify metrics that measure the worth of their work:

salespeople have sales targets, production managers track whether inventory is delivered on time and under budget, but for most of us it is very difficult to associate and measure our direct contribution to the desired outcome.

Perhaps the most actionable metrics are derivative – a combination of a signal, statistics, interpretation and analysis. Measurement of atmospheric temperature and pressure, compared with historic observations and combined with predictive algorithms, do produce relatively reliable weather forecasts. Perhaps measuring multiple aspects of customer experience, compare them with competitive alternatives and combining it with  predictive algorithms, can produce more accurate sales forecast.

Is it possible that we have unreasonable expectations? We often expect direct causation while operating in an open system environment. Business environment is not a scientific experiment and unpredictability of market conditions cannot be isolated to prove validity of specific measurement methodologies.  We only can improve odds, but we often expect certainty. Uncertainty is the reason for any important measurement effort. Measurement improves confidence in a quality of the decision is supports, but it cannot guarantee an outcome, after all according to Warren Buffet “It is better to be approximately right than to precisely wrong.”

This entry was posted in Market Research and tagged , , . Bookmark the permalink.