Trust the Data: Don’t Run Your Business on Alternative Facts
Today, it seems like you can’t glance at the news without wondering if what you see is truth, or a deliberately misleading lie. This isn’t a new problem, we just have new names for it, “fake news” or my personal favorite, “alternative facts.” If you run a business, information is your power, your most valuable asset, the key to winning deals, increasing revenue and cutting costs. But you can’t expect to beat your competitors if you run your business on alternative facts. You have to base your decisions on real data, and to do that, you have to know that you can trust the data.
Last week, I attended the Gartner Data and Analytics Summit. There were some extraordinary keynotes and focused sessions that really made me think about data and information in a new way. One data quality presentation included this horrifying perception: C-Level executives believe that 33% of their data is inaccurate. It may be true, or it may not be true, but either way, that’s a lot of mistrust.
I think that to a large extent, the C-suite has to shoulder the blame for their own problem. If you can’t trust your own data, then something needs to change in the way your business runs. As executives, as data management professionals and as business people, what can we do differently to fix this mistrust? Using what I learned at the conference, I think there are three areas that can make a difference.
Trust the Data
While there are certainly other factors at work here, there is always the possibility that the data itself is genuinely inaccurate, incomplete or not suited to the need. Data governance initiatives are designed to address that possibility. Setting rules for acceptable data quality standards, then automating processes to make sure the data that makes it to the C suite always adheres to those standards has to be the starting point. There are some excellent tools on the market, such as the quadrant leader Trillium Software suite, to make data quality improvement and governance initiatives easier.
“Data Governance isn’t a technology, it’s a thing that people agree to do.” – Merv Adrian
— Paige Roberts (@RobertsPaige) March 6, 2017
But don’t get fooled into thinking that buying a tool is going to solve the problem. Agreement among people is the important part of a governance plan. The business needs to buy in to the data governance process so that everyone can trust the data. It can’t just be an IT initiative. Governance rules need to be based on the needs of the business user.
If I need metrics on sales by region to decide whether or not to open a new store in a particular area, then I’ve got to be able to trust the data that I see in my dashboards is for that region only, is only for sales, and includes all sales that are relevant to my decision. The only way to know that, is if the data has been carefully governed, if the agreements between all the people who have touched that data match with my expectations of what that data represents.
Trust Your People
Data governance isn’t a technology. It depends on people. In order to trust the data, you have to trust your people. Even if they’re not like you. Especially if they’re not like you.
In the middle of the conference, International Women’s Day happened. I was delighted to see that just from an eyeball count of all the women who were asked to stand up at the morning keynote, and from the fact that the Women’s Breakfast was sold out a month beforehand, women made up around a third of the attendees. At least that number seemed to hold among the presenters as well. One of the most interesting keynotes was from Margaret Heffernan.
She talked about the value of teams, and what research tells us makes the best, most effective teams. Three factors seem to contribute most strongly to the problem-solving ability of a team:
- High levels of empathy and understanding
- Equal contributions of all members
- More women.
It was a good set of studies to discuss on that day in particular. But there’s more to it than just including women more in decision-making positions.
Another excellent keynote speaker, Tim Harford, talked about the fact that creativity is increased by getting out of your comfort zone, and the best teams were not a team of old friends, but teams that included a stranger. The best teams are diverse teams.
But only if you listen to the stranger in your midst. Trust that their view is as valid as yours, allow everyone on your team to contribute equally, and you will find new insights that would not have been possible if you had listened only to people who share your world view.
We as humans have a strong tendency to deny the validity of data that doesn’t fit our mental models of how the world works, and to pay attention only to data that confirms our models. This can completely sabotage the ability of data to affect positive change. Surrounding yourself with people who think just like you do, see the world just like you do, or simply overriding dissenting opinions that are not like yours, will make certain that there is no chance for the data to have any real effect on decisions.
“Without social capitol, data becomes noise, or it just sits there and does nothing.” – Margaret Heffernan
Think about that the next time you’re tempted to shut down discussion on a subject and just make the decision yourself. Think about how much more creative and effective teams are when each person on the team has a chance to contribute. Consider that there may be a better alternative that you hadn’t even thought of. Give your people a chance to surprise you. Trust that they have something useful to bring to the table.
Trust in a Safe Space to Share
Margaret Heffernan also brought up examples of how data’s greatest value lies in being shared, discussed, and argued over. Data that just sits all clean and pristine and accurate and perfect in a data repository is useless. It will change nothing. Data has no real value until it is shared. Socializing the data, sharing it among teams and discussing its implications are what make it a valuable catalyst for change.
Data Lake initiatives are all about sharing data, making the data available to more people across the business. A data lake can help a business combine data in new ways, and share across departments.
But, both social and technical attempts to share data can be sabotaged by one thing: fear.
— Paige Roberts (@RobertsPaige) March 7, 2017
The creator, writer and director of Mr. Robot, Sam Esmail, was also a keynote speaker at this conference. His description of hacker culture and our society’s overall vulnerability to cyber-attack was frightening to say the least. On the technical side, security measures can perhaps counter-intuitively allow data to be shared more safely.
There’s an old country saying, “Good fences make good neighbors.” Data security can help provide the safe spaces for data to be shared among the people who really need that data. Data that isn’t shared isn’t used. Security measures make sure that the data that is needed gets shared and used only by the people who have a right to that data. The job of a security admin is not to keep people who need data from getting it, but to make sure those people can use their data without worrying about it falling into the wrong hands.
As tricky as a good cybersecurity strategy is, a social strategy that makes sharing of data, opinions, and ideas safe, is even harder to implement. And even more key to business success. As I mentioned in the previous section, diverse viewpoints and the free exchange of those views make for better decisions.
That has to come from the top in any organization. You have to encourage participation. You have to make your conference rooms and hallways a safe space for folks to argue, to spitball, to look at data, consider it, throw out opinions as to what it means, and experiment with ideas to act on it. If you’re going to get the best value from your data, you have to give people a safe space to consider its implications. Your organization has to be in a safe space to be wrong, to be different, and to say something that might sound crazy at first. That’s where some of the best ideas come from.
It isn’t easy to trust in an age of mistrust. But it’s essential. If you or your C-suite don’t trust the data, some fundamental things need to change.
“If you want to get somewhere new, start somewhere new.” – Tim Harford
Check out these Common Data Quality Use Cases to learn more about how to build trust in data by ensuring the most accurate, verified and complete information.