Interview Series — Massimo Cealti
Hosted by Career in Analytics
Welcome to the next installment of Career in Analytics’ interview series. This forum is designed for decision science professionals — both beginners and veterans — to meet one of our members and engage in a conversation with them. We want our group to be a place for great conversation and debate.
CiA: Welcome, Massimo. Can you tell us a little about yourself?
Massimo: I choose my career because I like numerical representations of reality, started over 25 years ago in an unknown field called at the time “Mathematical Theories of Accounting” by which basically all companies activities needed to be represented numerically. This enabled me to embed the principle that every economic event, corporate decision or activity can and should be measured, because only what can be measured gets managed.
Almost no one understood what that was about, but discovered how to link accounting to Systems Theory, and ended up roaming into Epistemology. My passion was fueled by Multivariate statistics, and I choose a job that has never left me stranded without analyses to dig deeper into the phenomena I was studying. I must acknowledge that living in France has enabled to unleash my “statistical creativity”. Having met and worked with some famous Statisticians freed me of the methodological rigidity that is sometimes dominant in the academic mindset.
Maybe 15 years ago it was too early, people would stare at you if you tried to explain the difference between predictive and prescriptive analytics, and I have met hundreds of top managers who did not understand the difference between simulation and optimization. This is why I am excited that today these topic have risen in relevance. Not that the same people would now understand them, but they would just pretend to and no longer stare… it is a more comfortable position to be in.
Over time I had the chance to apply analytical principles to a great variety of domains, especially to FMCGs, but also to durables, media, OTC, and lately applied my knowledge to fragrances, such a delicate and almost “immaterial” topic which requires the most sophisticated algorithms to simply bypass the obvious.
CiA: What does analytics mean to you/your company?
Massimo: Analytics are an enabler, they tremendously help structure business decision making and thus contribute to building competitive advantage. They can be both forward looking or based on historical information, depending on the type of decisions they support. In my personal experience they are particularly helpful in the following 5 areas:
Strategy and Strategic Planning
While Vision and Strategic direction are naturally and rightly based on “qualitative” information (passion and emotions can fit in there as many CEOs have stated) in my career I have seen the most successful business decisions taken after carefully planning, collecting and analyzing information that enable to choose, plan and implement strategy.
Working in teams whose task was to make strategic decisions such as whether to enter a new market or product category, in 2 of the companies I worked for we used to spent up to 5–6 months each year collecting and analyzing information for Long Range Planning, and run simulations and scenarios both at Macro (Market), Microeconomic (Category) and Company (product of brand) levels to select and prioritize the best growth opportunities. After translating Strategy into tactical activities the rest of the year was spent in execution and in adjusting the forecasts with fresh data.
Sales Forecasting and Market Modeling
This is the day-to-day operational battlefield which is the bread and butter of analytics. I am still puzzled at the knowledge gap between what can be done and what actually gets done in sales forecasting. Data scientists are well positioned to know that the uncertainty and churn are only excuses for not tackling the real issues. In this context the Predictive Analytics used in the digital world will certainly help “democratize” these techniques and create additional opportunities for successful application in the “brick and mortar” world.
Segmentation and Consumer Behavior Modeling
This is another big area where analytics have historically created and will continue to impact. The wide array of Classification techniques available today is amazing, and using machine learning, both supervised or unsupervised, data scientists can really dig and understand how markets are structured or which behaviors can best be leveraged upon to tap into new revenue growth streams. As new approaches keep surfacing, from the quantitative inclusion of Behavioral Economics to the application of Quantum Probability, Data Science has moved out of (Linear) Regression Models. In the past often we were trapped in the linear world.
My enthusiasm is also motivated by the fact that when I started, many of the existing algorithms were not included in software applications, and you had to code them, but often once you had coded, the processing power was insufficient and you ran out of memory! Luckily this is now a less frequent issue.
Marketing KPIS
the biggest future opportunities for Analytics may derive from this area. Marketers no longer struggle to acknowledge that growth is not just the result of advertising, promotional, CRM or POP activities. However plenty still needs to be done and here again the application of Data Science can help clarify and put in perspective the role of each lever in the assessment of Marketing ROI.
Innovation
in this context Analytics and Artificial Intelligence are already blooming. Every other week we witness the emergence of new “creations” where the role of analytics is not just a facilitator. Ideas of this type of applications can be found in the pharmaceutical, the automotive and in the fragrance industry where Advanced Analytics can play a relevant role in modeling the chemical structure of a fragrance to achieve desired features. I should also mention the blooming applications of biofeedback and brain-computer interfaces, most of those algorithms have been developed in the last few years and they will help understand the role of emotions in the perception of stimuli.
CiA: You already covered a number of challenges over you career. Is there anything most recent that you would like to add or emphasize?
Massimo: Currently my interest is focused on very specific applications in two areas: Sensory — researching how to replace expert evaluations by machine analyses, and how to replace consumer declarative hedonics, which can be fallacious even with goodwill with direct measurement — and Brand Equity Assessment.
CiA: What are the biggest analytics mistake you’ve seen people making?
Massimo: Here things can be both funny and tragic. Other than conclusions based upon spurious correlations which can lead to funny situations, mistakes are often due to lack of experience. However the biggest and most frequent one is Analytics denial! A management reaction to a feeling of being overwhelmed by:
- the complexity of the tasks and the process articulation (Data Science is still not for everybody…)
- the political implications of putting together data owned by different stakeholders and the need to break silos and heroically “set the data free”
- and the obvious fear of the unknown.
The solution to this impasse, as one of the best leaders I know taught me by saying “Rome was not built in a day” is to start small, pilot the activities and once the benefits start to surface deploy them on a large scale. It seems obvious, but I have noticed in recent years that project management capabilities have been losing momentum, quite likely due to management short term focus, everything is “right here, right now” and there is no time for doing things right.
CiA: Do you have any career advice for aspiring data scientists?
Massimo: My piece of advice comes from direct experience, both recent and from a forlorn past. I will remember all my life a presentation I delivered to a packed-full audience of Managers of a global household appliances manufacturer. I had run a brand positioning study and was highlighting the strength of their brand vs their direct competitors. The results came after a Hierarchical Classification run after a simple Factorization. However for reasons I have forgotten I had chosen to use a modified version of the BCG matrix model and in my charts the axes were not orthogonal to show the correlation. While I was explaining these to the auditorium I could not help but notice an audience of about 70 people with heads slanted sideward…
To the future Data Scientists, please think that the end of all your analyses is but only the start of the compelling story you need to build for your clients or colleagues. If you fail in building a storyline that keeps them engaged you will lose them after the first 3 minutes. Based on my experience, just like for insights, the communication of Analytical findings succeeds best when it is performed by multidisciplinary teams, so once you have got your thinking right find someone who is a great storyteller to help you out.
Remember you start with a handicap, the majority will still think your complex analyses are little more than “gibberish” and if you end up there your sweat and tears will have been shed for nothing. Instead if data scientists can explain their findings in simple terms, then management will grasp the depth, breadth and relevance and the critical contribution these can bring to business decision making.
Career in Analytics is a forum dedicated to connecting beginning analysts with experienced and veteran mentors. Our topics cover a variety of interests in the area of analytics and professional career development.
We would also like to thank — Corsair’s Publishing for their help in bringing this content to you!
