I recently went one-on-one with Jon Francis, Chief Data and Analytics Officer of General Motors.
Adam: What do leaders need to understand about data analytics?
Jon: That is a big question. I think, about analytics, and especially in light of the hype cycle around artificial intelligence right now, that maybe it is often overlooked that the value of more traditional analytics and how you can drive business strategy and decision making. And one of the challenges that I see with AI and a lot of companies are struggling with, and I know a lot of CDAOs from other companies such as GM who see the same challenge, is that I have seen a quote from a BCG report somewhere that 85 percent of companies are not yet yielding the value from AI that they want to be, or should be, or could be, and some of that comes back to just scaling, and it is expensive, and some of the capabilities are not yet ready for prime time. Separating all that out, there is still a ton of value for organizations in the use of data and analytics and data science capabilities more broadly, and where it can play a role. One is on strategic orientation, and just in terms of how insights around what data can tell you help inform the strategy that you can be driving for the company. And there is gold in those hills in the sense that there are very counterintuitive insights sometimes that the data tells you that you may have a different perspective around your consumer base or the constituents that you serve that could lead you in a very different direction. So that is one. And then, of course, the role that data science and analytics can play more broadly, as you then define those strategies and where you want to focus to drive growth through your organization, is how you can bring data science to enable those capabilities. And again, speaking in terms of whether it is predictive analytics or machine learning more broadly, the role that they can play in terms of how you can better target customers or how you can better personalize experiences and create differentiated experiences for customers, those are all core capabilities that have long existed before generative AI that can still provide and drive a lot of value for different companies.
Adam: You shared a great framework for all leaders leading data-driven businesses, which today is just about every business. Number one, think about data in terms of how it can help you make successful decisions. And number two, think about data in terms of how it can help you drive growth. One hundred percent.
Jon: Leading with a little bit of humility as it relates to data and how it can inform, for instance, a different strategy. And I have been doing this for a long time now, 30 years. One observation is the weaponization of data, and usually it happens where data or analytics are being used to support an a priori hypothesis that a business leader or an organization may have around what their strategy should be or what their decision should be, and then tending to focus on the data that supports that hypothesis. And on the flip side, of course, the dismissiveness around data and analytics when it does not support that previous hypothesis, and often then there is in trying to explain away an insight that might be counterintuitive or counterproductive to an existing strategy. So this is the part that I have observed, and I have worked in nine of the best brands of the world over the course of my career over the last 30 years, and the ones that are good in terms of really getting at how you embrace data versus the ones that are not, it is really leading with that humility and that willingness to accept what the data is telling you and how you course correct based on those insights or the directive from the data and really embracing it. And again, back to the humility of, well, okay, we were wrong about this. We tried something. It did not work. The insights or the data are suggesting a different path or course of action. How do we then take those learnings to inform how we can further optimize what we are doing? You hear people talk about fail-fast organizations. That is really core to the success of those companies. I would give Amazon as a place that I spent five years. They really embraced that, and they had that intellectual curiosity and that humility around data to help inform the decisions that they made based on what their data said and were quick to move on and innovate in a different direction if it was counterintuitive to what they thought, versus other organizations that I think of as heritage brands that maybe over the course of many decades have built incredible brands and influence almost in the absence of data and insights. It is in those environments that it is really harder to break through, because there are long-held beliefs that are tougher to change, and there is a cultural dimension to that as well.
Adam: What you are really speaking to is the power that data has and the role that data can play in not only decision-making, but in how you sell your decision. And as a leader, if you view data as a tool to help sell your decision, that is great, but if you view it as a tool to help sell your decision, but you have already made your decision, and you are only using data to help justify the decision that you made, you are completely missing the point. You are going to fail as a leader. But if, to your point, you lead with humility, you lead with an open mind, you lead with objectivity, and you do not come in with any preconceptions, and you let the data tell you what is actually going on, and you look at all the data, you do not look at the data that you only want to look at, and then you make your decision, that is how you get to success.
Jon: One thousand percent. And I think a lot about my role as I am in the business of signal detection and data hoarding, maybe is another expression, which is that I want to take as many signals as I possibly can when it comes to looking at a particular opportunity or a challenge for a business, and whether it is around consumer application or an opportunity, I am looking at not only behavioral data, but what might be said in social platforms about my brand, or even through call centers or reviews of some type. Those are all important signals that work together and should work together to paint a more holistic picture of what is really happening. And when I started my career, there was not the computers nor the storage capabilities really available to be in the business of hoarding data in the way that you can easily now, and the first order of business is gathering the data and getting it organized in a way to make it useful. But then, especially now as it relates to generative capabilities, we can start to synthesize what those learnings are across different sources of information. And how can you coalesce that to make an informed decision? And you are spot on to really stare at that and almost check your ego and your preconceived notions at the door and really reflect, because especially now it is an embarrassment of riches that lets you look at these problems holistically. If you want to be a data-driven leader, reflect on what those signals are telling you. And it does not mean your idea is bad. In some cases, it can mean you have a bad strategy or it is a bad course of action, but a lot of times it can just help. How do you make that idea better? How do you inform a decision or a direction in a way that may not have been intuitive? If you are leveraging those other signals, that can help you push for an optimization that you might not have otherwise had. So I do think there is a lot of power. And again, if I compare where I am now in my career and the capabilities that are available to me when I started, it is a completely different ball game, and leaders should be open to it.
Adam: If you are a leader with that mindset, with the right approach, with the open-mindedness, how can you ultimately translate the data insights that you have into business strategy?
Jon: Yeah, this is a big challenge, and while a lot has happened in terms of AI and analytic capabilities and cloud-based architecture that make it easier to extract value from data and insights and build capabilities, the thing that has not really changed is the ability for skilled practitioners to be translators. Maybe that divide is narrowing with generative capabilities to help do that, but there is still a core skill around data storytelling. How can you make sure that the analytics and data that you are looking at are not academic, and how do you then translate those insights and learnings from what the data is telling you in a way that would be resonant for an organization, and translate it through the lexicon that they find important, how they think strategically about the pillars of the organization and what drives growth. It is really important that you have individuals who, almost from a decision science or translation perspective, can take those insights and translate them into meaningful language that helps a business understand. So that is really an important skill set, and you may call it data storytelling. It is really an important skill, especially in organizations that may have storytelling core to their DNA. I often reflect back on my time at Nike or Starbucks. If you take Nike as an example, it was not uncommon that I would get coaching that I should not put more than one number on a slide, and to make sure the slide was super impactful from a visual perspective, high-resolution imagery around sports, a clear story, and do not show more than one number. I am giving that as an extreme, but there is something about the level of complexity that usually comes with analytics and insights, and it is really important to have someone who is quite skilled at translating that analytics into a meaningful story that can elicit the right level of advocacy and belief in what the data is telling you to drive that strategic direction. So I think data storytelling is an absolutely critical skill, and I do not think it is one yet that generative AIs will replace. Again, they can help augment the data storytelling, but that is something that humans are still quite adept at doing, and it is a very important skill.
Adam: As you are sharing your advice around translating data insights into business strategy and the keys to storytelling with data, a couple of tips you are sharing. Number one, the importance of simplification. Keep it simple, stupid. We have heard that a lot, and it is no more relevant anywhere than it is here. Understand your audience. You are usually not presenting to an audience of fellow data scientists. You are presenting to a totally different audience. It could be an audience of business executives. It could be an audience of marketers. It could sometimes even be an audience of customers. Understanding who your audience is can really help you dictate how you are presenting your data in a way that is going to be as effective as possible.
Jon: Yeah, that is a powerful point, and in some cases, you may have to curate different versions of findings based on exactly who that audience is. Whether you are talking to someone who is skilled in Data Science and Analytics, you would tell a different narrative or story that may be a bit more detail oriented, versus a CEO, versus a board member, versus someone in the communications organization, where you think there may be a very powerful story that aligns with your brand’s purpose and mission that, if told in the right way, can have an emotional impact that really resonates for a broader audience around the brand positioning for a company. So it is really spot on. And oftentimes, I will think a lot about knowing your audience before I even build that first slide. And as I said, in some cases having to build multiple points of view, because in many cases, you may have to share with more than one audience. So I think that is a really astute observation.
Adam: Do you have any other best practices to follow, or other pitfalls to avoid, when it comes to data storytelling?
Jon: Yeah. For the majority of your audience, try to stay out of the weeds, and it is very easy for data scientists to get into the sausage-making of the how. It is pretty easy to lose an audience if there is a focus on that. You need to give enough of the how in terms of establishing credibility and confidence in the work, but that is not the majority or near the focus of what you are trying to do. The focus, and I have learned this through trial and error over many years, really needs to be about impact. So take everything away in terms of the approach, the methodology, the data. What is it that you are trying to do? What is it that you are trying to impact? And in a profit-driven organization, it is usually about creating efficiencies, about driving revenue, about eliminating costs. Make the story about that, the impact that you can have in service of those KPIs. Start there and make it very clear, ultimately, that what you are trying to propose, the data you are trying to position, and the analytics that you have are really going to be in service of how you make it clear that there is a path to impact. Once you start following the dollars, as one of my previous managers told me, it becomes quite easy to create advocacy, because now you are first and foremost thinking like a business person for the organization. I think about it from a cultural perspective, and one is just normalizing the role of data in a company. One thing I have done in the past, as an example of how you can start to normalize it within a culture, is every quarter running an all-people meeting. The focus was on sharing insights about the business and working with the communications team, and really on a quarterly basis, whether it is around customers, business trends, or macroeconomic trends, calling the company together and openly sharing what we have learned in an impactful storytelling way around what is happening with the business. I can tell you, once you start to stand up that capability, and knowledge sessions as well around some of how we are using the data to drive action and strategy for the organization, that creates a level playing field for the entire company in the sense that you really democratize data and insights so that everyone thinks about it through the lens of the work that they do to support the company. That is one mechanism that I have seen work really well, and that is more bottoms-up. From a top-down perspective, leaders should start to role model and demonstrate behavior around asking questions. Anytime they see a strategic proposal or a decision, a recommendation around direction, ask questions of, it may sound silly, but can you bring the data that you have to support the decision, or what experiments were run that led you to pick this path over a different one? Why did we not run an experiment? We should go back to the drawing board on that. I was briefly in banking in my career. I remember being in an executive meeting where one of the most senior executives in the organization was asking questions around why we picked the top decile from this model, only for targeting for a specific product, versus why we did not look further into subsequent deciles. So yes, there is the bottoms-up around how we can create this culture and democratize data. But there is a big part from a senior leadership team in normalizing the expectations around data and asking those questions. It does not mean that they have to be experts in data science, but it does mean that they need to be well-versed enough in things like experimentation and test and learn more broadly, causality, correlation, the types of concepts that any business leader should understand and really expect as decisions are being made within the lower ranks of the company. That mindset and those activities and actions from an analytics perspective should be taken to drive those decisions.
Adam: Clearly, a key theme, if not the key theme, of this conversation is the power of data in decision-making and how, as an organization, data has to be at the center of how you make decisions. How can leaders best incorporate generated data into their decision-making process?
Jon: Always bring the receipts. If I am working with a partner in my organization, I make sure that when we are driving decision-making, we have the receipts through the use of the data. Behavioral insights, it could be ethnography, to really support the decision or recommendation that is being made, and to have it grounded in pretty clear insights that effectively support the hypothesis that you are trying to drive and the decision that you are trying to make. In particular, one thing that is very valuable is the role of experimentation to support decision-making. If you have the ability through pilots and POCs to truly do test and learn, to say, we think we have an idea for a new widget or new capability to support the direction of the business, and if you have a proof point where you have run an experiment and you actually have the causal results to show the impact of that, then that makes the decision that much easier. You can say that you have actually proven that this capability can show or produce incremental lift relative to not doing it, or it can help reduce costs in some way. So that, for me, is really an important dimension. And not to harp too much on experimentation, it is also what made Amazon really special as they built the website, and this was core to my role at the time back in the early 2000s, which was scaling experimentation across the entire website. At any point in time, we had hundreds of experiments running concurrently to best understand how we could optimize the site and what changes we could make to whether it was product detail pages or other features to drive better engagement and better sales in service of our customers. It is also important to know, and I want to be careful to say this, that you cannot replace intuition. I will give you an example from my time at Amazon where this was true. Jeff made it really clear that everything that we launched on the website needed to be AB tested. It had to be experimented on. I remember in particular when we were, and I am dating myself here because this was early 2000s, ready to scale into new product lines beyond just books and music, and DVDs, but into home improvement and other categories. Jeff wanted to understand through experimentation, as we started to add those features, what was happening to the core overall business that we had been running for years. It turns out that when we ran experiments, adding these other categories that had higher price points created more engagement and browsing, but less purchasing, and in some cases, distracted from purchasing in the core business. We actually saw decreases in overall sales. Jeff ultimately was okay with it because he had a clear intuition and vision about what Amazon ultimately would be, the most customer-centric online store for consumers, and he was willing to take those challenges in the short term because he had a clear vision and intuition on where to push the business. That experience also told me that, and coming from a data leader this may be surprising, data and analytics should not rule every decision. I think they should be an important augmentation to decision-making, and sometimes your intuition for what you are trying to do long-term can play an important role that may be counterintuitive based on what data is telling you. But to not have the data in decision-making and strategic direction would be a mistake. At least you ultimately know what you are getting yourself into, and then it can also help you optimize decisions going forward.
Adam: Do you have any advice as to when to go with the data and when to go with your intuition?
Jon: Yeah, that is a really great question. For me, it comes back to a leader who is setting a clear vision and strategic direction. There will be important pillars that you outline in terms of what that vision and direction are, and to execute and bring that vision to life, I would hope and believe that data will play an important role. That will serve as a guidepost to what those decisions are and the execution path that you outline. But I do think that there will be times when intuition, if there is something that you ultimately need and you believe in, probably more from a long term horizon perspective, where from a predictability or confidence level the data can only tell you so far, will only direct you usually over a shorter term horizon, and you may need to trust your gut or intuition because you may not have the right signals. If it is something that also aligns back to your core vision, and you do not have those longer-term signals, you go with your intuition and gut, but then also test and learn and validate those decisions that you have made from an intuition perspective along the way to make sure that you have guideposts going forward to check and adjust. I also think about someone like Steve Jobs, who I would argue was a very critical intuition-led leader. I am sure data played some role in a lot of the decision making he did, but when you start to look at the product expansion for Apple over the years, whether it was the iPhone or iPad, decisions were made along the way that were probably very much aligned with his vision for Apple being a product organization, and maybe not customer data suggesting that they needed to get into those businesses. He created demand perhaps where there was not any. So I do think it is an important role to navigate and to check and adjust on the role of data versus your intuition, especially as you think about your strategic orientation and direction.
Adam: A lot of what you shared, and a lot of what we are talking about, ties in the importance of decision-making with storytelling and data literacy. When you are making your decision, whether you are ultimately making the call based on your gut or making the call based on what the data is telling you, and oftentimes they are one and the same, you need to have access to good data. You need to be able to understand the data that you have in front of you. You could have numbers, but what are those numbers actually telling you? How accurate are those numbers? How reliable are they? A lot of that is on the data scientist and the data science team presenting those numbers to you and being able to effectively communicate their data in a way that is easily understood and acted on successfully.
Jon: There is another dimension, too, and again, maybe from the school of hard knocks, trying to deploy analytics. It comes back to humility, but where I have built capabilities over the years, and I would say they fell on the cutting room floor. We may have had great storytelling, very clear recommendations, and belief in the data quality, as you highlighted and acknowledged, but yet nothing really happened with it. I bucket that as, well, we felt good as a data science team. We may have had a good meeting with a business organization that we were trying to influence to drive change. The problem is it did not happen. The other dimension that is really important that I have learned is around change management, and probably one of the best examples of this from a previous role I had was where I built out a customer lifetime value model. It was very clear, the inputs and the outputs, and everyone understood the role of the model and the power of it. It was productionized in the database, running every month. We could score any customer in terms of their remaining lifetime value. We had clear recommendations on how you could use it to help inform decisions around product and marketing. Yet it was never touched. I thought of it as a little bit of a field of dreams. If you build it, they will come. As I said, we had all the great storytelling around it. Maybe the naivete on my part was the change management of a particular organization, like a marketing function that would really benefit from using a lifetime value model. The problem was that they had a way of doing business and running their workflows, and they were always fully at capacity in terms of the work that they were doing and how they were deploying media and marketing dollars and getting campaigns out the door. For us to come along with something that was completely different, that would, in my mind and in the mind of the CMO as well, be a productive way to help optimize and drive better sales more efficiently through the use of LTV, we needed to approach it differently. Since then, I have spent a lot of time thinking about the change management side of analytics. How are you making sure that you are walking a mile in someone else’s shoes to say, time out. You are bringing something new to the table. Even if they like what they are seeing, you have to be disciplined and orient yourself to help them on the journey of how you actually bring it to life, and find specific use cases where you can start to collaborate on the front two, if you will, of how we used to do things to how we can do things in light of this data or in light of these analytic capabilities to change and optimize how we do our work. That is a pretty critical role for an analytics team to help with the change management, because there is a real risk, and I have seen it. I just gave you one example. I have hundreds of examples where if you are not intentional about the change management, and if you want your recommendation or result to be successful, you need to meet the other organization or the receiver of those analytics and help them think through and navigate how those analytics can help inform how they would do things differently, and test those analytics, because in many cases there are new things being deployed that would need to be tested and learned to prove that they truly bring the value that you are advocating for. I think the change management topic is one that is often overlooked in data and analytics.
Adam: What you are speaking to brings up a really important point that we did not talk much about, but is critical to leading data-driven organizations, which is that it ultimately starts at the top. Leaders need to prioritize creating a culture in which data is incredibly valued. If you are a data scientist in an organization where leaders do not really care all that much about data, you are going to be doing your job. You are going to be showing up doing what you are getting paid to do, but your work is not really going to be moving the needle all that much. But if you are a data scientist in a truly data-driven organization, what you are doing is going to be front and center to what actually moves the needle every single day, and that is on the leader. Leaders have to show up every day thinking about data, thinking about setting a culture of data, putting data first. It starts with the leader, and in a lot of ways, it ends with the leader.
Jon: I really love that observation, Adam. In many ways, when I started this journey in leadership and analytics, I thought my role was more around how I help make the analytics better and how I make sure we have the right data to do our work. Actually, my job has nothing to do with that. I am still technical. I can still lean in and provide support and perspective. But my job is almost the CMO for analytics, because one of the greatest responsibilities for any CDAO, especially in an organization that might not have fully embraced a data-driven journey, is cultural transformation. How can I, as a leader, play a role to help the company be more data-driven, to earn more trust in data, to show the benefits when we do use data or analytics, and the impact it can have on the business in a measurable way. That is where partnering with the finance organization to have the proof points and put those proof points on the board to say that when we lean in with analytics, look what it can do for the business, and to be able to story tell and amplify and democratize. There is a lot in my job that is more around cultural transformation and trying to build advocacy with other leaders for what the data scientists can do. It is not about navigating technical work. It is really about cultural transformation, especially in heritage brands. My role has really evolved in the sense that I feel like I am a chief transformation officer more than a Chief Data and Analytics Officer. That is a really important role. And by the way, what I just described is probably true in most companies, that most companies have not fully embraced being data-driven. You have a lot of tech companies who grew up with it, that it is more wired into the DNA. That is not true if you look at most of the Fortune 500. There is a required transformation in terms of how you really embrace data literacy and data decision-making in a way that is probably not organic to the company naturally.
Adam: Do you have any other advice for how anyone in a leadership role can help drive that cultural transformation?
Jon: For me, it always comes back to humility and being intellectually curious and starting first with a question. A lot of leaders make the mistake, especially with AI right now, of saying, I have this cool whiz bang tool, what can I do with it. The joke that I make internally is that it would be like saying, I need an Excel strategy. No one ever says that. Excel is an important tool and capability. AI is an important tool and capability. I do not think you necessarily need an AI strategy, just like you do not need an Excel strategy. You need a strategy for the organization. Start first with what ultimately are the levers for growth for your organization, or how you can create more efficiency. Then have the intellectual curiosity to say, okay, here is what we have done to date. Here is how we do our business. Let me spend some time with our analytics or data leader to see how we might take these strategic orientations and think about them differently to drive incremental and step-change improvements in what we do. Then bring to bear different analytic methodologies and different data to go solve that problem. The one thing I encourage leaders to do, especially now in light of AI, where it is a real tail wagging the dog problem in my mind, is to remember that it does not excuse the need to start with a really clear understanding of what it is you are trying to do. Forget technology. Forget enablers. Start first with that orientation. What is the problem we are trying to solve? Then figure out what capabilities to bring to bear to solve those problems, and have the humility to think differently, to be open to thinking differently about how you would approach that.
