Ethics as Conversation: A Process for Progress

MIT Sloan Management Review

Most organizations can agree on what questions to consider before making a decision about marketing, finance, or operations. But many stumble when the issue at hand has ethical consequences.

One former CEO of a large financial services company put it this way in a recent conversation: “By the time the issue gets to me, it’s been analyzed, scrubbed, and PowerPointed to where I don’t have much idea what really might have happened. But I still have to make a decision that has ethical consequences.” He said this meant he had to be confident that people in his organization had posed roughly the same questions he would ask. He wasn’t so sure that was the case.

We began to use this insight in our conversations with executives and students. We ask them to define what we call “your ethics framework.” Practically, this means defining what set of questions you want to be sure you ask when confronted with a decision or issue that has ethical implications.

The point of asking these questions is partly to anticipate how others might evaluate and interpret your choices and therefore to take those criteria into account as you devise a plan. The questions also help leaders formulate the problem or opportunity in a more nuanced way, which leads to more effective action. You are less likely to be blindsided by negative reactions if you have fully considered a problem.

The exact questions to pose may differ by company, depending on its purpose, its business model, or its more fundamental values. Nonetheless, we suggest seven basic queries that leaders should use to make better decisions on tough issues.

How does this make me feel? One of the hallmarks of tough ethical issues is that they have an emotional basis as well as a factual one. Your stakeholders will not necessarily evaluate your choices rationally, particularly if they are affected negatively. Attending to your own emotions is one way to anticipate stakeholder responses. Of course, this question presupposes a fairly careful analysis of the facts and circumstances, but if we do not try to get a handle on the emotional part of an ethics problem, we run the risk of just intellectualizing it. We need to get better at listening to our feelings as well as doing a hard-nosed analysis of facts, constraints, and contexts.

Who is affected and how? Every person is enmeshed in a complex array of relationships with others who can be harmed or benefited by their choices. Likewise, organizations have a set of stakeholders with whom they are trying to create value. Knowing how a particular ethical issue affects these relationships is crucial for good decision-making. In addition, some of these stakeholders — both personal and organizational — may well have legitimate claims that must be respected. Tough analysis of these “who” and “how” questions are prerequisites for effective action.

Who can I talk to about this issue? We learn ethical values and principles through our conversations with family, friends, and others. When we find ourselves enmeshed in a situation with ethical consequences, it is a good idea to reach out to confidants to talk about what we should do. Sometimes, reasons of confidentiality (itself an ethical concept) preclude sharing the specifics of a problem, but in most cases, we can still get useful advice from those who won’t simply tell us what we want to hear.

Are there alternative framings of the issue that I should explore? Once a decision is framed to be about a certain issue, making the decision itself is often fairly easy. As American philosopher John Dewey noted, “…a problem well put is half-solved.” However, framing the issue is usually the most difficult part of the process. If an executive frames an issue only in terms of “what is best for shareholders,” the answer she reaches may be fundamentally different than if she had framed it as “what is best for stakeholders.”

What are my best options? Thinking about framing leads to considering new ways to approach an issue, especially the timing of different alternatives and their associated actions. Considering different choices is critical for effective action because each has its own set of strengths and weaknesses. When decision makers rely on only a single idea, they are more likely to be blind to its weaknesses. Having multiple possibilities for addressing an issue increases learning as you compare and contrast their consequences. Managers may also feel pressure to make trade-offs among stakeholders when they don’t see routes to satisfy multiple groups. In dealing with complex stakeholder issues, options also need to be tested and refined with stakeholders’ input.

What would happen if my thinking and decision became public? Ethics is deeply personal, but it is also deeply social. Many people respond to criticism of their ethics by saying, “I have to look at myself in the mirror. I have to live with myself.” That is, of course, correct. But your organization also has to live with you and your choices. Ethical decision-making must pass some kind of publicity test. What would happen to your organization if your decisions were on the front page of the newspaper? Would you still be proud to tell your children or other loved ones what you did?

What would cause me to change my mind and my decision? We often think that once we make a decision, the issue is closed. But a key feature of ethical choices is that this is rarely the case. We need to continuously monitor circumstances to be able to decide whether our decision needs to be modified. This is especially true with the deployment of new technologies: As the use of a particular technology emerges and we learn more about its consequences, we need to be ready to modify our course. The attitude of humility that this question fosters is essential for continued learning and adaptation.

Unethical behavior is harder to hide thanks to the explosion of social media. Leaders who want to embed ethics into how their businesses operate need to articulate and practice using questions that foster continuous conversation about what their organizations do and how they can do it better.

Whatever your ethics framework and whatever set of questions you bring to addressing ethical decisions, we encourage you to continuously think through them. That will help you hone and improve them so that you and your organization can make even more effective choices.

It’s one of the hardest decisions a manager will ever have to make.
The Business of Automation and AI-Powered Retail Predictions

Artificial intelligence and machine learning are increasingly at play in just about everything we do, both personally and professionally. It is because of these technologies, for example, that an online store can automatically determine what products should be featured on the homepage. These main pages and lists of product recommendations are oftentimes custom-tailored to the individual customer, boosting the likelihood of sales and growing the bottom line.

As you might imagine, similar advancements in technology can be seen clear across the world of retail business, spanning nearly all industries, verticals and niches. You have to think about your market, your customers, and their actions as continuously moving targets. It's not enough for you to know where they've been or even where they are right now; you need to accurately, quickly, and reliably predict where they will be tomorrow.

Be Ready for Tomorrow Today

You need to anticipate their movement such that you'll already be where the opportunity will arise tomorrow. Because if you're already there, you'll be ahead of the competition and you'll be poised to capitalize. That's the power of automation and artificial intelligence, because too many manual analysis and manual tasks take up far too much time and way too many resources.

You need tomorrow's information today. And it needs to be fast and accurate.

Perhaps one of the most incredible examples of this in action today is Endor, a protocol that allows for automated predictions on encrypted data. In the past, the only way that you could gain access to this level of business intelligence is by hiring a team of data scientists who could then sift through the mounds of available data. With Endor, the process is automated with artificial intelligence.

This is accomplished by leveraging proprietary social physics technology and massive machine power, making accurate predictive analytics available at a scale to just about anyone. Whereas the old prediction model required a team of data scientists, utilized limited data, and took upwards of two months per prediction, the AI-powered predictions of Endor are vastly more affordable and can be produced in under a day.

The price per prediction is dramatically reduced, making these sorts of AI predictions accessible for even smaller retail businesses. This empowers them to stand up against behemoth corporations with their huge data science departments. That's huge!

Accurate Predictions Now

So, say for example that your company is developing a new product. You're in the final stages and while you're certain you'll make the product available in English, you'd also like to approach the international market. Because you have limited resources, it would be impossible to offer your product in dozens in languages, so you'd like to zero in on about three. But how can you decide?

It's not just a matter of what languages are most common around the world; it's a matter of where your product has the best shot at success based on a myriad of factors. When you upload the customer data that you have available, the AI protocol can work its way through it and provide you with its best possible prediction. And you'll get that prediction within a day.

We have to realize that these types of forward-thinking predictions, getting ahead of the curve before it even gets there, is really how you can take your business to the next level. By gaining the ability to be proactive with your decisions, rather than simply reacting to crises, you can improve engagement, increase sales, and bolster customer loyalty.

Getting Ahead of the Curve

As possibly the world's largest online store, Amazon is understandably oftentimes the elephant in the room. This is particularly true in conversations revolving around the modern retail environment, even if Amazon is primarily online and not brick-and-mortar in the traditional sense. You'll find that Amazon utilizes automation and artificial intelligence extensively.

One example of this really drives this point home: With the promise of two-day delivery with Prime and even same day delivery in some situations, Amazon cannot afford to simply react to customer orders as they come in. An incredible level of automation is necessary if they want to avoid sending parcels express clear across the country as the majority of their shipments.

The prediction engine at Amazon anticipates not only what products are going to be popular within the next while, but also WHERE those customers are located. The inventory can then be allocated among the distribution centers proactively. That way, when an order comes in, chances are that the item will be in stock at a fulfillment center closest to the customer. At least, that's the ideal.

These predictions are based on an incredible amount of customer data collected over the years, looking not only at categories of products, but specific items, down to color and size. When you consider just how many products are available on Amazon at any given time, this is no small task and it is certainly a task that is unsuitable for a human being. Artificial intelligence is necessary to work quickly and accurately enough at this kind of scale.

More AI Prediction Applications

A gargantuan corporation like Amazon can take advantage of machine learning and effective artificial intelligence for effective inventory management across its vast network of fulfillment centers. But how can predictions and AI work for retail for businesses of all sizes? Predictive questions, based vast organizational data, can help you to determine:

Who is most likely to upgrade to a premium product? What subset of customer will respond best to a new promotion? Which inactive customers are most likely to re-activate? What product color will lead to the greatest conversions? Who is likely to spend the most money on products in the next 3 months? When should the email newsletter be sent for maximum click-through rate?

The resulting predictions lend themselves to actionable insights that you can then integrate into your workflow and timeline. In addition to Endor, some other companies that offer AI data management and predictions include IBM Business Intelligence, Pega, Deloitte Omnia AI, and Qlik.

A Brand New Paradigm

Conventionally, data scientists would need to construct a new data model for every prediction they'd like to form. The circumstances are different and thus they call for a different approach to arrive at an accurate prediction. But this is time-consuming and prohibitively expensive. Even with the advent of machine learning, building a new data model for every new prediction is far too resource-intensive.

To be quick and nimble, a new framework needed to be developed to handle the growing demands of today's retail environment. And it needs to do so rapidly and accurately. Machine learning and AI predictions represent the future face of modern business. If you want to get ahead, you'll need to get there even before your customer realizes where they're headed.

Marketing Attribution Platforms: Is It Smarter to Buy or Build In-House?

Companies all over the globe are constantly collecting data that has the potential to provide some valuable insights into consumers — that is, of course, if they can quantify the results.

This is why marketers turn to marketing attribution, where value can be assigned to the various touchpoints that led to a purchase. Whatever role a certain media exposure played in driving a customer through the sales funnel gets some sort of credit, allowing marketers to better plan future marketing spend.

Some companies initially tapped outside vendors and consultants to handle this process — vendors that had already built the software necessary for attribution models. However, because of frustrations with the early deployments of first-generation attribution providers, some of these marketers brought this process in-house. 

Marketers who desire to own their company’s analytics internally do not want to wait around to act on data-driven decisions. In their minds, building their own attribution capabilities lends to greater agility and ensures the risk of exposing high-value data to the public remains low, which can quell any nerves about a potential security breach.

There’s also a (false) perception that if they build in-house, costs would go down and be cheaper in the long run. With a growing number of data scientists in the market, it just reinforces the perception that a martech vendor or consulting agency is no longer necessary.

However, many companies that have taken attribution in-house are turning back to analytics providers after struggling to run the process themselves. It can be difficult to build and manage a platform, even with competency in data, analytics, and measurement.

The decision to build or to buy isn’t as clear-cut as you might think — consider the following before making your final decision:

1. Determine whether you have the right expertise for the task.

Someone in your organization will need to oversee the attribution process, especially when building it in-house. Consider whether you have the manpower to manage the program. Do you have the resources and the right people for getting the program off the ground? You’ll need data science experts with experience building predictive models and an understanding of the nuances of machine learning approaches and the advantages and disadvantages of each. You’ll need data architects who have experience with extraction-transformation-loading (ETL) methodologies and approaches, tech solutions architects to understand what tools need to be integrated and how, data visualizers to translate data science language into marketing language so marketers can act on the insights, and perhaps a project manager to keep everything on track.

If you don’t already have the right people in your organization to run an in-house program, you’ll need to decide whether it’s more beneficial to hire those people or buy a platform from a third-party vendor. If it’s important to you that your platform is in-house, you’ll need to recruit, hire, train, and retain all these people. If you can’t afford that (financially or time-wise), it might be wiser to buy rather than build.

2. Consider how much you can reasonably afford.

Software can be expensive, and if you build in-house, you’ll need to consider the costs of building software for data collection/ETL, cloud hosting or physical hardware, integrated machine learning and artificial intelligence frameworks, data operations, quality assurance for deployment, and ongoing care for the platform. All of this will also require significant manpower — you might need at least 15 full-time employees to develop and manage the platform — and their salaries alone can cost hundreds of thousands of dollars per year.

It’s also not just employee and infrastructure costs that you’ll need to consider when building an in-house marketing attribution practice. There’s also the cost of getting all the external pieces integrated — for example, you’ll need to connect with ad servers, data management platforms, customer data platforms, and various other data sources, each unique in its own right. These costs can add up quickly, and if you’re set on building in-house, you’ll need to be prepared to make a significant investment.

3. Think about how much time you have.

If you build in-house, it can take years before the platform is up and running and ready to go. The data experts need time to put together all the pieces and build a repeatable process that marketers can benefit from — and it’s a long process. Even lightweight models that don’t require as much effort to build can still take several months, so if you’re planning to go the in-house route, make sure you understand the timeline before taking the leap. If your company is in a rapidly changing competitive landscape, think about whether you can wait a year or two to start taking advantage of these marketing insights.

Marketers have more data than ever at their fingertips, which can be a double-edged sword. While you have access to a wealth of information about your customers and how they interact with your brand, it can be challenging to organize all that data to glean valuable insights. Marketing attribution is critical if you want to make the most of your media buys. For any attribution model to be successful (in-house or provider), you need a team in place to analyze the results and make decisions using these results that improve business performance.

But the first step is deciding whether to buy a platform from a vendor or build the capability yourself in-house. Making the right decision will not only improve your marketing efforts, but it can also save you money in the long run and increase your company’s bottom line.

No comments