The Democratization OF JUDGMENT
In an age of Big Data and Artificial Intelligence, the exercise of good judgment by employees throughout an organization has never been more important.
the announcement of a promisHARDLY A DAY GOES BY WITHOUT ing new frontier for Artificial Intelligence (AI). From fintech to edtech, what was once fantastically improbable is becoming a commercial reality. At the same time, corporate investments in Big Data and the dividends they yield in terms of consumer insights are trumpeted on a daily basis.
Oddly, we don’t hear much about the demand created by this rising ‘supply’: In a world of Big Data and AI, the demand for sound and distributed judgment is increasing. ‘Qualitative judgment’ — the ability to make a decision based on a personal interpretation of the context and available facts — has never been more important. In this article I will exlain why judgment has become so important, and how to go about enabling it throughout your organization.
The Rising Demand for Judgment
A basic decision process can be deconstructed into four logical steps: collect and organize available data; analyze it for patterns and insights; predict the best possible courses of action; and use judgment to make a final decision. This last step is more important than ever, and there are three main reasons for this.
1. Qualitative judgment is the last preserve of humanity in making decisions.
There is no question that Big Data and AI offer important advances in the realm of management. Already, they are helping organizations analyze their markets and consumers more effectively and make informed predictions. But certain types of decisions — particularly those around innovation and those relating to consumers — will always entail a component of qualitative judgment.
For example, in healthcare, AI is having a huge impact. But, even if AI can support a doctor in making a diagnosis and suggesting medical treatments for a particular cancer patient, only
the doctor herself is able to factor in the overall condition (physical and mental) of the patient and his emotional context (and that of his family) to decide whether to proceed with surgery or chemotherapy. It is not possible for a machine to factor in the emotional and political context of any situation; yet few would argue that both contexts are critical for most decisions in business — and elsewhere.
Some of the best management decisions in business history have been made based on qualitative judgment rather than data alone. Consider the story of Nespresso by Nestlè, which has become the leading global brand of premium-portioned coffee. Nespresso machines brew espresso from aluminium capsules — preapportioned single-use containers of various high-quality coffees and flavourings. Those familiar with the Nespresso story know that the brand only took off when it stopped targeting workplaces and started marketing itself to households.
Quantitative evidence had suggested that individual consumers’ intentions to purchase did not meet the threshold requirements set by Nestlè’s product-launch procedure. However, Jean-paul Gaillard, the young marketing head of Nespresso, believed strongly in the idea, and thanks to his skillful interpretation of the data and his willingness to go against Nestlè’s previous innovation ‘rules’, he convinced the company to take the risk. If he had only listened to the data, the concept would never have gotten off the ground.
Business history is full of similar stories, where people have willfully complemented data with their qualitative judgment and reaped great rewards. Creativity, emotional understanding and pure imagination are things that humans excel at, and the availability of a huge amount of additional data will not change this fact of life.
2. As the cost of prediction goes down, the demand for judgment will increase.
In their November 2016 article for Harvard Business Review, “The Simple Economics of Machine Intelligence”, [Rotman School of Management Professors] Ajay Agrawal, Joshua Gans and Avi Goldfarb framed the trade-offs between Artificial Intelligence and judgment. I would like to elaborate on this brilliant article, stressing the authors’ analogy to Production Theory — i.e., the economic process of converting inputs into outputs.
As Prof. Agrawal et al. indicate in the article, technological revolutions impact the cost and value of important input factors. In our case, thanks to the advent of Big Data, the cost of finding and organizing data and running analyses has become much cheaper. As the authors indicate, AI is a prediction technology, so the cost of prediction will also become cheaper over time.
When the cost of any input factor falls, certain microeconomic rules can be applied — and not only to production, but also to the decision making process. First, we will substitute other input factors (human skills) with the low cost (and better) technology to collect data and develop predictions; and second, the value and demand of complementary factors will rise.
For example, when data and prediction are cheap, companies can generate more frequent customer insights, which creates the need for more-frequent decisions regarding customer support, promotions, product customization and new product development. This, in turn, will lead to greater demand for the application of judgment and emotional understanding — provided by humans — to make decisions. This is exactly what happened at Unilever, after it introduced a number of data-driven systems accessible to all of its global marketers: The availability of real-time, frequent, data-driven consumer insights generated ever-greater demand for judgment and decisions by the company’s marketers.
3. As data-prediction technologies are distributed more widely, so must judgment be.
Big Data and AI will provide managers and employees at all levels with accurate data and predictions at their fingertips. Using distributed IT architectures, these tools can allow employees throughout an organization to make the right decision for a particular context in a timely manner. As a result, the smartest companies will ensure the distribution of judgment-based decision powers.
Recognizing the power of data-based distributed decision making, Affinity, the Minnesota-based credit union, issued a framework to guide its employees in making decisions regarding loans. Its ‘MOE’ system (Member, Organization, Employee) operates like a ‘constitution’ to free up the judgment powers of
It is not possible for a machine to factor in the emotional and political context of any situation.
employees and provide a ‘North star’ to guide them when applying these powers. Employees have full latitude with respect to rates and can override the bank’s policies based on their judgment of ‘what is right for the customer in that context’, supported by customer analytics. The MOE Constitution states:
“No employee will ever get in trouble for doing what is right for the customer. There is only one operating policy or guideline you ever need: Trust your feelings. If it feels right and makes sense, do it on behalf of the customer. Do not consider the system capability, policy, or procedure; err on the side of doing whatever is necessary for the customer and allow your manager or supervisor to take care of the rest. Finally, be prepared to defend your decision! If your intention is to do what is right for the customer, you will have the support of management and your co-workers.”
Every Affinity employee can now decide, on the spot, whether to provide or not provide a loan to a particular customer, and if so, at which rate, by using a blend of customer analytics and personal judgment. When an employee deviates from the bank’s policies, she is required to justify her decision and post the rationale in Affinity’s Touche system, which stores all data and electronic records of members/clients for all to see, as well as a full history of employee explanations for lending. The result: When Affinity employees started to make judgement-based decisions in large numbers, charge-off rates for higher-risk clients dropped by almost 50 per cent — from 1.9 to 1 per cent.
Implications for Organizations
The three factors discussed above indicate that now, and in the future, companies will require more rather than less human judgment for their innovation- and customer-related decisions.
To get there, judgment must be democratized across the organization. Most companies cannot rely on a lone individual like Jean Paul Gaillard to override the existing culture and procedures, and that is why every organization needs to create its own Judgment Protocol. Much like Affinity’s MOE, this is a system that legitimizes the exercising of judgment within your organization across all levels — and one that will change the century-old
When data and prediction are cheap, companies can generate more customer insights, which creates the need for more-frequent decisions.
‘command and control’ philosophy that many companies still use to make decisions.
Following are four guiding principles for leaders who are eager to embrace this new imperative.
1. Democratize Judgment Power
Companies tend to believe that innovation and market-related decisions are the responsibility of a few, highly-positioned people. There is a widespread autocratic view, which conceives that only the ‘elected ones’ are entitled to make decisions that affect customers. By way of contrast, consider the credo that Toyota embraced in its Toyota Production System (TPS). In TPS, everybody is responsible for the search and implementation of ideas to improve operational performance. Responsibility is pushed down to the very lowest level in the organization. In the TPS, two worlds — manufacturing and market innovation, which appear so remote from each other — share the same philosophy for success.
2. Foster Qualitative Judgment Skills
As soon as we push down the responsibility to identify issues and make decisions, we will want to increase the probability that our employees will chose the right course of action and execute on it properly. The second core principle of the TPS is to train everyone in the workforce in quality, lean/six sigma tools and techniques. Widespread training on standardized tools increases the probability that people will come up with the right insight, decision and execution to impact performance.
Other organizations should apply this same principle and standardize tools, methods and techniques to improve their employees’ skills in generating insights — and applying judgment. Doing so will require a shift in perspective, to a mindset that views judgment as a key organizational capability worthy of investment.
For example, Unilever encourages every one of its employees to engage with consumers to gain insights about their needs, providing allotted time during the workday for this activity on a regular basis. To raise the effectiveness of the time and freedom provided, Unilever trains its employees in both consumer observation and probing methods, as well as on how to use some of their newly developed Big Data marketing tools like the People Data Centre, which combines social media and business analytics capturing conversations in 40 languages.
As an example of this approach in practice, consider the Knorr brand’s ‘Love at First Taste’ campaign. Data suggested that ‘people are attracted to others who like the same flavours as they do’; Knorr marketers decided to act on this finding by setting-up people with the same taste on blind tests and videotaping the results. The video reached 100 million views in a few weeks: Data plus insight and judgment spawned a marketing hit.
3. Provide Data Access to All
Data access will raise the effectiveness of employees in using their judgment. Of course, some companies are better than others at transforming data into actionable insights. Prior research has tended to emphasize the role of data scientists who have the skills to analyze data. This implies that companies with more data scientists have better chances of generating value. My own experience as a consultant, supported by academic research, indicates a different view: Firms that hire an army of data scientists do not always generate better value. Rather, it is the process of data management — and particularly, the democratization of access and use of data among managers and employees — that creates tangible value.
Consider internet platform companies, where data is at the core of the business model. Airbnb has taken a step ahead in the democratization of data: Its entire workforce, including human resources, has access to its data science tools to make timely decisions related to requests from both users and providers of homes, as well as act swiftly on innovation opportunities.
However, Airbnb also understands that fully-inclusive data access is not enough: Its employees are also trained on how to use data tools and extract insights to make informed decisions. Data University is Airbnb’s attempt to make its entire workforce — not just its engineers — more data literate. It has designed 101-level copurses on data-informed decision making which are available to all employees. The result: Since launching the program in late 2016, Airbnb has seen the weekly active users of its internal data tools rise from 30 to 45 per cent.
4. Loosen the Reins of Control
Organizations tend to be uncomfortable at the prospect of decision-making authority being pushed down the hierarchy. For many, the loss of control is synonomous with risk, and this has been a major barrier to the true empowerment of the workforce. The solution lies in shifting from a traditional ‘Prevention-control Model’ to a ‘Post-detection Model’.
For example, in a bank, if an exception to a loan policy is being requested, the Prevention-control Model would require authorization signatures several levels up. Even when a loan applicant has a a perfect credit score and fits with the bank’s policy, most likely the loan will need to be signed by the employee and her supervisor before being approved. Prevention Control Models are the greatest barrier to true empowerment.
Let’s return to Affinity as an example of a Post-detection Model: When an employee decides to offer a loan to a customer because it ‘feels right’ (per the MOE Constitution) but is an exception to the bank’s policy, the employee must write up a rationale for the decision taken and post it on the client data system. As such, the rationale is transparent to colleagues and supervisors, generating a social control that reacts only in extreme instances.
In closing
The time has come to walk the talk with respect to democratizing decision-making authority. Low cost data-prediction technologies, coupled with an official, company-specific judgment protocol can help to free employees from the shackles of hierarchy and create truly agile and customer-centric organizations that are able to adapt quickly to market signals. And, if the companies who have embraced this approach are any indication, profitable growth is sure to follow.
Data access can raise the effffectiveness of employee judgment.