Machine learning takes customer centricity one level up

Can machine learning help pharmaceutical companies make better business decisions?  And can machine learning support the move to more customer centric organisations in the industry?

The answer is “yes, it can”, if you manage to build enough expertise and trust within your organisation to accept machine learning and the disruptive way of thinking that comes with it.  It will help you to better understand the drivers of your customers’ behaviour and it will make personalised strategies for individual customers become a reality.  This is what we learned from a presentation by our colleague Björn Van Loy, Global Head of Analytics at Trilations, delivered at Eyeforpharma Conference recently.

• Machine learning is already part of our lives

Already today we seem to accept without too much thought the help of tools that are based on machine learning algorithms.  We trust Siri on our iPhones to come up with tailored answers to our needs (find me a restaurant nearby with food that I like!).  Or we trust smart mobility applications to automatically remember where we parked the car or tell us how to avoid traffic jams.

Also in our pharmaceutical business life, machine learning is becoming more and more embedded in our work processes.  In pharmaceutical research the value of machine learning processes in diagnosing and disease identification and reading genome sequencing data has been accepted.  Anonymised clinical big data projects are emerging everywhere. Physicians are contributing to ever larger volumes of digitalised health information. Health apps for monitoring and steering food and exercise habits of patients are on the rise as well as patient support programs monitoring bodily functions online or helping them to improve their treatment adherence.

The expertise of Trilations, however, is how to use big data for developing customer centricity strategies for optimising the use of sales and marketing resources, based on advanced analytics.  Machine learning is now an additional tool for customer data analysis.

• Not everything is big data, not all machine learning is the same

Using the metaphor of “data lakes” for big data about customers from pharmaceutical companies, Björn Van Loy explains that any company venturing into machine learning should first work on understanding how deep its lake is and how pure its water.  What is the volume of the data, the frequency with which it is updated, what about velocity and volatility?  How many sources are there and how can they be combined?  And how good are the data: are there any quality standards governing data capture by sales reps, for instance?  Are they complete?  Are they standardised?

To efficiently use machine learning in customer centricity strategies such understanding of the data should be complemented with some organisational investments: companies should accept that it is a long-term investment yielding improved results over time as the algorithms are fine tuned.   Teams working on customer strategies need to build trust in the power of “blind” use of data through machine learning and acquire new competences through education or adding different data analyst profiles to the teams.

At the basis for the need of this organisational change is the transition from human-mind-learning by the traditional data-analysts, on the one hand, to learning from the data directly steered by more unsupervised algorithms, built by data scientists, on the other hand.  It is still easy for us to understand how data handling can be used for automated classification or to use past behaviours and opinions to predict future customer behaviour.  But making a leap of faith to accept that computing will come up with actionable results without detailed understanding how this was derived from the data requires some organisational change management.

• Machine learning at the service of customer centric strategies

Bjorn Van Loy explains how at Trilations we complement our existing advanced data analytics with the new machine learning knowledge for better customer strategies and more efficient use of sales and marketing resources in pharmaceutical companies.

When helping customers setting up or improving customer centric strategies, Trilations combines data from its own systematic research with the “data lakes” owned by the customers themselves: CRM activity data, sales data, information on prescription behaviour, real life patient data, … These combined data provide insights on what drives the prescription behaviour of customers: this includes arguments about the medical qualities of the products obviously, but also market conditions (such as regulations, reimbursement, restrictions for use, …), company reputation issues (can a bad company deliver good products?) and even patients response to the products.

These “machine learned” insights can help pharmaceutical companies to better organise their customer facing efforts and focus their resources where they are most needed.  This can mean shifting resources between therapeutic areas or disease areas, but also between different types of activities.

Trilations’ data knowledge systems deliver valuable information on the optimisation of the multichannel mix: what is the best way to interact with the customer for different products and in different markets?  It looks at the channels (sales force, congress, mail, post, remote online communication tools, …) and helps to select the most efficient mix in line with available budgets and company targets.

• Machine Learning’s main new deliverable: personalised customer strategies

The latest algorithms, however, take customer centricity strategies to the next level.  It effectively allows to have personalised strategies for individual physicians or HCPs.  It understands for every individual what the preferred nature and level of interaction should be and even knows which content and information will be most relevant and appropriate.  Modern internet and mobile application users know how efficient and effective companies such as Google can be in delivering content and information to us that is genuinely of interest to us.  And they deliver it in ways we like at times that are convenient to us.  And that is what machine learning did for them.  Why should we not use it to even better address the unmet needs of our customers in pharma and other industries?