Data is the lifeblood of many businesses today. It drives decision-making, illuminates opportunities, and spotlights problems.
At least, it’s supposed to do these things. If your business is behind when it comes to data, then you’ll find it almost impossible to keep up with the rapidly changing competitive environment. This is the landscape no matter what industry you’re in.
But too many businesses are behind on data, for one of two main reasons.
1. The data is in silos
2. Data analytics are lagging behind
Businesses obviously need to optimize the way they use data by organizing and storing it well in data warehouses, and by creating real-time analysis tools that empower business leaders to grow their companies.
So it’s full speed ahead toward data-driven decisions, right?
Not so fast.
Why should we hit the brakes? The simple truth is that data is not a completely impartial arbiter. Artificial intelligence and machine learning can actually grow in bias, as this article from Fortune shows. In other words, if your business is only driven by data, you are susceptible to all sorts of problems that will hurt your business instead of help it.
We (your business, and ours, and just about everyone else) have stumbled into uncharted territories when it comes to the ethical use of data. Far too often, companies have ignored ethical norms and ended up hurting people (as well as their corporate brand).
One of the most devastating examples came late in 2018, when Washington Post employee Gillian Brockell wrote a heartbreaking post titled “Dear tech companies, I don’t want to see pregnancy ads after my child was stillborn.” This post showed the dark-side of algorithmic ads based on prior social media posts and online searches. Her experience resonated deeply and went viral.
This story is a warning signal to any company that uses data to make decisions. When we forget that data is really about people, then we run the risk of doing terrible things with data. We may not intend to be bad actors, but the results are the same. A lack of empathy leads to data-driven solutions that harm, not help.
So your business doesn’t just need an awesome data warehouse and cutting-edge analytics. It also needs the right approach to analytics–one driven by ethics, governance, and business outcomes.
To get there, let’s look at the history of three generations of analytics. Then, we’ll use that history to inform the three keys to successful usage of data analytics.
The History of Data and Analytics
As we explore the value of data, it will be helpful to look at the history of data and analytics. We’re not going to go back to the abacus days, but the rapid evolution of data means that we have seen three distinct generations of analytics over the past decade or so.
Generation 1: Data-Driven Decisions
The first milestone in data usage came as businesses moved away from gut-feel operations and instead started focused on using statistical analysis to make data-driven decisions.
Using the data they collected and curated, businesses produced reports and dashboards used to inform business decisions. These decisions were largely reflective, indicating what had happened more than what might happen next.
The data companies used was almost exclusively on internal data, with the occasional exception of something like USPS moves. This meant that data-driven decisions could not really account for competition or market factors in any meaningful way.
The data was managed by a data scientist who was knowledgeable in both data structures and the business’s operations, and the data scientist painstakingly curated the data warehouse.
In these days, data allowed companies to take a leap forward from the “go with your gut” days of yore. But before long, simply having highly curated internal data in a data warehouse wasn’t enough.
Generation 2: Big Data
The next generation of data came as companies began to get access to more data sources. The advent of big data was not as much about the amount of data but about the number of sources providing data. These sources–commercial data sets, public data sets, and data exhaust from user actions and choices, for example–made it possible to get new insights from data.
With these new sources of data, the way data was stored changed. Instead of a highly curated data warehouse, companies moved to a data lake that was less organized but deeper. This also changed the role of data scientists–while some acted as data guardians like before, others became explorers and/or citizen data scientists who dove into the data lake to find trends. While the results of this data exploration were still largely reflective, they sometimes exposed trends that companies could seize upon. The earliest stages of machine learning helped to empower this trend spotting in some cases.
Generation 3: Augmented Analytics
These days, data is all about finding predictive insights and driving them back into the business. The authors of Deloitte’s 2019 Tech Trends report summed this generation of analytics up well on page 7: “Today companies need the ability to predict (I have a good idea what will happen next) and prescribe (I can recommend a response).” It’s a move from assisted intelligence and augmented intelligence toward autonomous intelligence.
To do this, companies need to harvest information exhaust from customers and prospects, and data scientists need to apply machine learning to discover relationships and groups in the data to highlight trends. In fact, this kind of machine learning power is quickly becoming something companies need to view not as different and interesting but as essential, non-negotiable functionality.
Companies also need to smash the silos between data sources and ensure that data is formatted in a common way to provide data quality. This has led to the rise of API integrations that are more reliable and customizable across a tech stack.
As data analytics have grown in capability, governance has become even more critical. This governance has to do with data quality, which we just discussed, but also focuses on data security and privacy. The seemingly relentless stories of data breaches show what happens when security is lacking; the onset of GDPR in Europe and copycat laws in the U.S. highlight how vital privacy concerns will be.
As you can see from this breakdown, the idea of what data can do has increased exponentially in a short period of time. But these capabilities come at a cost. Whether it’s liability for failing to comply with a regulation about privacy, or the damage a brand sustains when it fails to effectively safeguard user data, data doesn’t just do more; it also costs more when things go wrong.
The Capability of Analytics
What can analytics do now? You can take data from Google Perfect Audience or Facebook Custom Audience to drill down on your potential customers with two or three factors, and then automate a marketing campaign to promote products and encourage sales. By tracking clicks and abandon rates, you can fine-tune your parameters until you have nearly a 1:1 overlap of interested customers with mass personalization.
We all know this from our experiences with Amazon, but this trend is moving into the B2B space. The trend of using personal data, combined with engagement with resources, is called social selling, and in 2017 a LinkedIn survey reported that social selling increased close rates by 20% and deal sizes by 30%. The Harvard Business Review shared this example: SAP used social selling for high volume yet highly targeted sales of cloud-based solutions, and saw a 40% increase in pipeline and several won deals worth $300,000 or more.
Of course, it’s important to remember that personalization that happens in social selling will never be a 1:1 overlap with your real customers. Gillian Brockell’s heartbreak reminds us that a customer’s wants and needs can change on a dime when there is a tragic event like a death or an accident. These kinds of unpredictable changes mean that relying completely on data analytics and algorithms that translate them using machine learning can actually do massive damage to your brand–even if it helps to generate sales.=
And the dirty truth is that most tech companies aren’t going to help you avoid these kinds of landmines. Listen to how Fortune recounted a 2018 interview with Microsoft managing director of research and artificial intelligence, Eric Horvitz:
“I don’t think anybody’s rushing to ship things that aren’t ready to be used,” he says. If anything, he adds, he’s more concerned about “the ethical implications of not doing something.” He invokes the possibility that A.I. could reduce preventable medical error in hospitals. “You’re telling me you’d be worried that my system [showed] a little bit of bias once in a while?” Horvitz asks. “What are the ethics of not doing X when you could’ve solved a problem with X and saved many, many lives?”
Tech’s bias is on advancing and pushing the boundaries, without paying a lot of attention to the long-term consequences If you don’t believe this, think about all the controversies surrounding Facebook in the last year or so. Developers are usually going to focus on what can be done with data, not what should be done.
But your company can be different by adopting the right approach to data usage.
The Right Approach to Data
So what overarching philosophies should inform the ways your company collects, stores, analyzes, and acts on data? Let’s think through three categories.
Your company should insist on the following three things when it comes to way you collect and store data.
Quality measurements: Store your data in the highest quality manner possible. Use APIs to maximize quality and create metrics to assess what improvements you need to make next.
Control access: Your data undoubtedly contains private information about people, whether it’s customers or employees. So you need to make sure that the only people who have the ability to see data are those you trust and those who need to use it. This informs how you protect against external data breaches, but it is just as important to control access by your internal teams to limit liability and protect privacy.
Protect intellectual capital: Your data is one of your business’s most important assets. No other company knows how you have succeeded at finding customers, cultivating relationships, and closing sales. The data around your customer relationships is intellectual capital with just as much value as patents and intellectual property; you should protect it with the same vigor.
How should your C-suite look at data? These three principles should serve as guiding lights for your company’s perspective.
Break down silos: Data should no longer be confined to certain divisions. The power of API integrations allows you to merge data, improve its quality, and make it a powerful tool for all divisions. So smash data silos and eliminate company politics so that your data works for everyone in the company.
Data-driven in all aspects: Your business should take a data-driven approach to decisions. Create strategies based on evidence. When you launch a new product or service, start by testing a series of low-fidelity, quickly made prototypes to gather actionable data. Gut decisions aren’t necessary anymore because of the predictive power data now has.
Measure outcomes: Data should inform decisions, but that’s not enough. Evaluate the outcomes of your decisions using analytics. Look at outcomes as a new source of data that will drive your company forward even faster.
Privacy: The sum of your customer data is your company’s intellectual property, but any customer’s individual data belongs to the customer, not to you. This is the driving force behind laws like GDPR, but it is also a common sense approach to treating your customers and employees well. Protect each customer’s data as if it were your own, and don’t share data on a customer that you wouldn’t want to share about yourself.
Transparency: It should be clear to people inside your business what you are doing with data, and it should be clear to customers as well. This transparent approach to ethics in IT lets customers know what to expect, and a commitment to transparency will help keep your company on the right track. The old saying that sunlight is the best disinfectant applies–letting people know how you use data will help you use data ethically.
Training Bias: Machine learning is a powerful technology, but it’s not a perfect one. As we mentioned at the start of this post, there are multiple examples of machine learning accelerating bias against groups of people. Your ethical use of data should recognize this shortcoming and find solutions for it, for the good of your company as well as the people who encounter it based on predictive data analytics.
Data offers your business nearly infinite possibilities these days. But not every possibility is one your business should embrace.
A commitment to governance and ethics, and a business-driven approach to using data, will ensure that your company uses data well. And this ultimately is the straightest path to business success.