Data-Driven Culture & the Reptilian Brain

Posted on Posted in News

http://bit.ly/2lLt53L
June 28, 2018 at 11:13AM via http://feeds.feedburner.com/FeaturedBlogPosts-DataScienceCentral

Artificial Intelligence (A.I.) is a wide umbrella of emerging technologies which have the potential to completely transform business and society.

In its immense complexity, artificial intelligence is like human intelligence. Effective analytics teams in business interact in the same way as our nervous system does to enable human intelligence:

– Analytics Organization: data professionals are like neurons in our brain, which organise cognitive and analytical functions akin to an analytics organizational chart. Leaders need to understand analytics roles, hiring and retention techniques, organizational models and the role of the Chief Data Officer

– Data-driven culture: culture is like the reptilian brain, the oldest part of our brain, whose primary role is reflex responses, not rational thinking. Relevant topics include building data-centric culture, managing cultural clash, managing people and robots, as well as data ethics.

– Analytics Strategy is like the neo-cortex, the part of our brain which allows us to develop complex thoughts. Leaders need to define a use case-driven analytics and A.I. roadmap and get buy-in for it.

– Analytics Execution is like an agile human nervous system which reaches all parts of the body. Relevant topics are managing an experimental project portfolio, agile methodology, metrics and governance.

Successfully managing a team through an analytics and artificial intelligence transformation requires skills and philosophies different from managing other groups of talented professionals. It is wise to be aware of potential trade-offs when it comes to defining the right organization, culture, strategy and execution. 

This is the second article of a series of four and focuses on data-driven culture: the reptilian brain.

Building a data-centric culture

Organizational culture is determined by history, technology, strategy, national culture, and most importantly, the CEO’s management style – it becomes innate to the organization and new members must learn it in order to be accepted.

Culture determines the first reactions of employees –for example, whether their first instinct when faced with a problem is to look into the data, to identify and manage responsible parties, etc. Reptilian instincts require little reasoning – a crocodile does not think or plan much before it jumps on its dinner.

Crocodile hunting a nu in the Mara River, Kenya

A data-driven culture does not happen overnight. Very deliberate & tangible steps need to be taken to foster a data-oriented culture in the following order:

1.   A single source of truth: Data sources in large organizations are often siloed in independent systems. Staff can pull the same metric from different systems and get different numbers. A single source of truth can be a large investment, but the drag on competitiveness from inconsistent data will only worsen as the industry moves towards greater precision. While this investment takes place, the data team can add great value by acting as data connoisseurs, knowing what’s available and what to recommend for each problem.

2.   Standard data dictionary: Data scientists and business managers need to agree on a data dictionary with clear, unambiguous and agreed definitions to help team members with different areas of expertise get on the same page.

3.   Broad access to data: The entire organization, not just data scientists, needs access to data. Without it, you cannot achieve collective business expertise in analysing data. This requires a simple self-service reporting system with the right governance and access levels based on the needs of customer service, and product & marketing specialists.

4.   Data Literacy: With good access to data must come good understanding of data. Provide compulsory training to reinforce basic data literacy across the whole employee base. There are 3 basic subjects which every employee should be comfortable with:

– Descriptive statistics: Basic ways of summarizing data (e.g. mean, percentiles, range or standard deviation) and knowing when each is appropriate.

– Data visualization: Clear communication of insights or concepts to speed up comprehension & collective problem-solving within the team.

– Inferential statistical tests:  To help detect, for instance, if a difference in sales between weeks is significant or if it is just random variation.

5.   Decision making: Many teams are governed by HiPPOs (highest paid person’s opinions). This can be especially bad when HiPPOs decide based on instinct or experience, when the choices could have been tested and supported with statistical proof. One way to counteract this is to cultivate experimentation with A/B testing. For example, in order to decide which website design or marketing messaging is most effective, managers can determine success metrics and sample sizes, and let the A/B tests run and let the data speak for itself.

6.   Once the five basics above are partially in place, an internal hackathon does wonders to energize the organization and take it to a whole new level:

–    Start with a customer problem (Problem Derived Innovation Analytics) and solve it through various design and analytical stages.

–    Be deeply cross-functional, by involving not only data scientists but IT, front-liners, brand leaders, product owners, graphic designers, customer service representatives, and coders.

–    “Everything can and should be challenged” should be the guiding belief for all participants.

–    Create minimum viable products, and not just PowerPoint slides –ideas should be materialized.

–    Keep the momentum going after the hackathon, with reward mechanisms which encourage employees to report on progress and to adopt new behaviours.

7.   After implementing these steps successfully, senior leadership can start evangelizing an automation and intelligence first imperative: Each problem-solving session must begin with A.I. (Similarly, start-ups in emerging markets have a mobile-first imperative).

Managing cultural clash in analytics

Traumatic experiences can sometimes lead to people experiencing split personalities. Formally known as Dissociative identity disorder, deeply involves our reptilian brains, and our limbic brain, by altering chemical balances, causing to weakened connections between various brain sections, and inappropriate activations of the fight-or-flight response, which resides in our reptilian brain.

Weaker connections in brain with dissociative identity disorder

Sudden & mismanaged transitions to a data-driven culture can cause the same trauma, and hence similar symptoms at an organizational level. There is often profound disagreement on where data responsibility lies, or what it looks like. Lingering cultural clash can become a major obstacle if the C-suite do not fully support the analytics strategy.

Medically, this disorder is cured by gradually re-growing the connections between brain areas. And so it is for teams, where managers must take the lead on bridging cultural gaps with intermediate approaches which suit both cultures.

Clash between waterfall and agile

The traditional project management methodology is called waterfall, which means you do not start the next step unless you have finished the current one. Analytics, on the other hand, needs a methodology called agile, where teams iterate versions of a model to make it better over time.

IT employees are often unfamiliar with agile principles in data management, and may prefer to build traditional data warehouses on the waterfall method (Software Development Life Cycle or SDLC methodology), which takes years to build before companies can enjoy the benefits.

This clash happens not only with IT, but with any business or support teams typically run on the waterfall method: procurement, legal and HR. Agile-based teams will feel that things are not moving fast enough under all the rigid procedures, while waterfall-based teams will feel things perpetually being done ad-hoc and in a rush.

Changing to agile is difficult. The rule of thumb is that as much as one third of the team might choose to leave, one third remains but is not able to adapt, and the remaining third succeeds to change. The solution is often to find a compromise by reengineering and simplifying support process so that they are faster and can keep both analytics and support function at ease.

Clash between practicality and scientific correctness

As with any project, analytics projects are constrained by a trade-off between technical quality on the one hand, and investments of time and money on the other. However, analytics managers cannot fully appreciate this trade-off (as they might in non-analytics project) if they are not data-science literate. The misunderstanding is mutual: many new data scientists ignore the rich business experience that line managers can offer and miss out on essential insights that would produce better results faster.

Not every project requires highly nuanced outputs or exceedingly robust performance; for some, a directionally correct rough-and-ready solution is sufficient. Sometimes managers suspect that a data scientist’s caveats about quality are simply academic concerns without long-term business consequences. This friction can strain relationships between data science teams and other business units. There are two ways of bridging this gap:

– Involve analytics translators: Business consultants can help to translate while data scientists and domain experts attempt to work together to define the right solution.

– Use explainable A.I.: Advanced models, like deep learning, are black boxes that cannot be interrogated due to their massive complexity: A.I. can tell you which customer will churn, but not why. Simpler models are easier to interpret – regressions, ensemble methods, tree algorithms – and can improve relations with line managers: if they cannot understand the model, they will not approve.

Empathetic team management

Empathy is the capacity to feel what another person feels from within their frame of reference. Studies show that observing another person’s emotional state activates parts of our brain responsible for that same emotional state. Any leader needs to understand his team, but it is even more vital in analytics, due to three main sources of frustration faced by data professionals:

Empathy: first person pain processing in yellow and empathy for pain in blue, the overlapping areas in both in green

– Reality not lining up with expectations: Analytics professionals want to solve complex problems with impressive machine learning algorithms with significant business impact. In reality, they need to look at data infrastructure, clean out unreliable data, and create basic reports and charts for tomorrow’s ad-hoc meeting. Data professionals will understandably be frustrated with the underutilisation of their expertise. Leaders need to make deliberate measures to incorporate their expertise for daily needs, defining success as the day when their expertise can find its way into the manager’s ad-hoc requests for basic reports and charts.

– Cannot add value independently of other teams: A lack of understanding of how hierarchies and organizational dynamics work might lead to further stress. Providing the right mentoring for data scientists to navigate the organization and develop their business sense is vital.

– Challenge of learning data science and business: Developing a business sense, learning how to completely manage a project end-to-end, and to problem-solve effectively, is challenging. Without the right support, it can even be demoralising. Data professionals need opportunities to learn, so it is essential that their career paths require business exposure via secondments to consulting teams or revenue-generating subsidiaries.

Managing humans and bots together

AI could become disruptive to workforces. McKinsey estimates that about 64-78% of all time spent on data collection, processing and predictable physical work can be automated. Although concerns about imminent job losses often overestimate the actual situation, introducing A.I. will create emotional stress among employees, as our fight-or-flight responses light up in the face of possible redundancy. 

Robot Sophia interviewed by a TV journalist

A.I. will drastically change our life at work. Paradoxically, it will actually make us more human because it will push us to focus on inherently human activities which cannot be automated, such as:

– Judgement: using our knowledge of organization and culture, as ethics and business sense.

– Creativity: bringing together diverse ideas and designing integrated, workable solutions

– Empathy: social relations such as networking, coaching, and collaborating

Think about a call centre agent seamlessly taking over an interaction from a chatbot. The bot will handle simple requests on its own and the agent will be refocused on higher-value queries that are often more rewarding.

New operating and governance models will become necessary to let machines and humans interact in an agile environment of constant learning, where tasks fluidly pass back and forth between them. Additionally, workplace communication, education, and training will need to be part of the design from the initial A.I automation pilots onward.

Adhering to data ethics

Neuroethics is a field of study dedicated to understanding the brain processes in Ethics. Ethics is a reflex about what we feel is right or wrong and is also partially related to our primitive brain layers.

Big data analytics raises a number of ethical issues, especially as companies begin monetizing their data externally for purposes different from those for which the data was initially collected. The scale and ease with which analytics can be conducted today completely changes the ethical framework. We can now do things that were impossible a few years ago, and existing ethical and legal frameworks cannot prescribe what we should do. While there is still no black or white, experts agree on a few principles:

–      Private customer data and identity should remain private: Privacy does not mean secrecy, as private data might need to be audited based on legal requirements, but that private data obtained from a person with their consent should not be exposed for use by other businesses or individuals with any traces to their identity.

–      Shared private information should be treated confidentially: Third party companies share sensitive data – medical, financial or locational – and need to have restrictions on whether and how that information can be shared further.

–      Customers should have a transparent view of how our data is being used or sold, and the ability to manage the flow of their private information across massive, third-party analytical systems.

–      Big Data should not interfere with human will: Big data analytics can moderate and even determine who we are before we make up our own minds. Companies need to begin to think about the kind of predictions and inferences that should be allowed and the ones that should not.

–      Big data should not institutionalize unfair biases like racism or sexism. Machine learning algorithms can absorb unconscious biases in a population and amplify them via training samples.

There are certainly more principles we need to develop as more powerful technology become available. Data scientists, data engineers, database administrators and anyone involved in handling big data should have a voice in the ethical discussion about the way data is used. Companies should openly discuss about these dilemmas in formal and informal forums. When people do not see ethics playing in their organization, people in the long run go away.

About the author

Pedro URIA RECIO is thought-leader in artificial intelligence, data analytics and digital marketing. His career has encompassed building, leading and mentoring diverse high-performing teams, the development of marketing and analytics strategy, commercial leadership with P&L ownership, leadership of transformational programs and management consulting.

linked-in