I read all the time about how companies are facing new challenges and opportunities in dealing with so-called big data and how to use it to extract value, particularly with regard to user behavior and other predictive analytics. I am devoting this column to some ideas about how lubricant and additive companies might utilize big data to improve their businesses. This may be increasingly useful as more and more vehicles are equipped with internet connections and online analytics of various kinds.
I would like to start with an interesting study documented in the book The Undoing Project by Michael Lewis, which deals with how humans make decisions based on data versus how computers make decisions using algorithms. The Oregon Research Institute asked a group of radiologists how they deduced whether a person had cancer or not based on a stomach X-ray. The doctors said there were seven major signs that they looked for, such as the size of the ulcer, its shape and so on.
The Oregon researchers took the information from the doctors and created a relatively simple algorithm based on their decision process. They then asked the doctors to judge the probability of cancer in 96 different individual cases, and the conclusions were compared with the results of the algorithm. Surprisingly, the algorithm did extremely well in what might seem a nuanced and complex decision process. Furthermore, the doctors diagnoses were all over the map; the experts didnt agree with each other, and when presented with duplicates of the same ulcer, the doctors were found to contradict their own previous diagnosis!
The conclusion here is not that the machine can replace the human; but rather that the machine can improve upon the foundation the human has built by eliminating random errors in judgment and other situational effects (such as human fatigue, distractions, etc.).
How does this relate to the lubricants and additives business? I am reminded of the huge and expensive data sets generated over many years by additive and lubricant companies as directed by experts in their fields. Surely this is an opportunity for more sophisticated algorithms and improved predictive analysis, built upon the foundations created by researchers, in order to improve lubricant quality, optimize lubricant costs, reduce testing expenses and improve speed to market.
It is possible that companies are already doing this to some extent, and if so that is to be commended. Such tools can potentially be used as a marketing opportunity to underpin a competitive technology advantage, or to provide a service to customers who may want to do some formulating but lack the necessary quantity of data or the predictive tools.
Improved use of data should also yield benefits in the supply chain. Predictive analysis of customer order patterns can allow better planning and result in fewer outages or delays, as well as lower costs and improved utilization of tankage and transportation. I could also envision an integrated risk assessment tool that would allow one to do scenario analysis for sole-source or higher-risk raw materials and incorporate options for back-up products or formulations.
Another potentially rich area is understanding end user information with the aim of designing better products and services. For example, a deeper understanding of end user information could allow a richer ability to segment the market and provide products more tailored or differentiated by mileage, weather conditions, type of driving conditions (urban, suburban or rural), vehicle type, driver demographics, etc. Some of this data may not be easy to obtain, so one would have to determine its availability and utility before proceeding.
Another area of understanding that would benefit from better data analysis and predictive capability is lubricant demand. I am sure most companies have their own tools for forecasting lubricant demand with sufficient granularity to plan their manufacturing capacity needs for the next 10 years or so. There are also industry studies for sale that serve the same purpose. However, predicting lubricant and additive demand by country or geographic region and by type and quality level is a complex job, but one that is becoming increasingly important due to the potentially significant changes coming over the next 20 years with growth in Asia-Pacific, increasing penetration of electric vehicles, fewer diesel cars and upgrades in lubricant quality.
Not only would it be useful to have a demand model, but one that allows analysis of different future scenarios to provide better insights into component, additive and base oil needs on a dynamic basis, resulting in improved capital allocation.
Any of these avenues for improved analytics and predictive capabilities will likely require considerable resources; so how might you decide where to focus your attention? Concentrate your big data efforts in your area of competitive advantage. If you see yourself as a primarily customer-focused company, then improve customer and end user data and drive new customer segmentation strategies and differentiated products and services. If your competitive advantage is in the technical arena, then improve predictive algorithms and tools and harness the full power of the massive data sets you already possess. Consider, too, how you might utilize these tools to help customers improve their products or reduce their costs.
As you embark upon such projects, be sure to do your homework as to how other companies (including those in other industries) have tackled such problems and which consultants may have expertise or advice to offer.
Investing in harnessing data within the field of your competitive advantage should allow you to reinforce or expand the size or longevity of the competitive moat that you already have in place.
Sara Lefcourt of Lefcourt Consulting LLC specializes in helping companies to improve profits, reduce risk and step up their operations. Her experience includes many years in marketing, sales and procurement, first for Exxon and then at Infineum, where she was vice president, supply. Email her at saralefcourt@gmail.com or phone (908) 400-5210.