Data-driven decision-making is the practice of using insights derived from data to make better decisions.
By relying on data rather than gut feelings to guide the actions of your company, you are able to move forward with fewer mistakes, and better meet the needs of your customers.
Some data-driven decision-making best practices include building for purpose and ensuring data access to everyone who needs it.
Different models can be used to help companies more successfully make decisions based on data.
Improved and easier-to-use technology has democratized the data-driven decision-making process, allowing more companies and individuals to use data to inform their business strategy.
Jeff Allen has spent his career working in technology marketing, product management, and sales roles. He is currently senior director of product marketing for Adobe Analytics and a member of the board of directors of the Digital Analytics Association. Prior to joining Adobe, he was vice president of marketing for AtTask (now Workfront), a hyper-growth SaaS project management company. Before that, as vice president of marketing for another hot startup, Venafi, he helped establish the enterprise key and certificate management market and successfully positioned Venafi as the leading pure-play vendor in the space.
Q: Why is data-driven decision-making important?
A: Being data driven allows you to move faster with fewer mistakes, which inevitably leads to higher profitability.
Culturally, more people are learning to factor data into how they think. As industries have gone more digital, people are becoming more aware that individual intuition can be wrong. When you have data, you can measure when your content does well, and understand why it resonates with customers. You can then use that information to successfully predict how similar content will perform in the future.
Q: How does a company start incorporating data-driven decision-making?
A: On a strategic level, a company has to decide to start making business decisions based on data. It becomes a matter of shifting the entire company culture from intuition to data-based decisions. Too many organizations in this world run on intuition and opinions. People assume that because they have experience, they understand their customers and know what their actions will be in response to marketing efforts. You tend to see more of that in artistic pursuits where opinion or taste is subjective, but when an organization determines that they can do things in an objective way, that's a cultural decision.
Once the company chooses to make the cultural shift to data-driven decision-making, they need the right people and tools for the job. It takes operators who know how to put the correct systems together to gather the data, process it, perform data analysis, and then present the information in a way that will help guide each member of the organization to make the right decisions for their role.
Q: How do companies develop a data-driven strategy?
A: To build a data-driven strategy, start with something you're trying to accomplish, and then seek out ways to measure whether you're accomplishing it or not. Next, inspect those measurements to determine that the data is clean and the signal is accurate. Once you know that those conditions are met, then you can begin analysis and start drawing insights.
When creating a data-driven strategy, it’s important to ensure that everyone who needs the data to make decisions has access to it. But sometimes, silos can exist within an organization that prevent information from being easily disseminated. Data is often either locked away in a system or owned by a team that won’t provide universal access, and that limits how well the organization as a whole is able to make decisions and move forward.
In the modern world, hoarding data is considered bad behavior. But many people associate having data with having power, and they believe that being viewed as the source of the data helps them further their goals or will advance their careers. And it turns out, that's no longer true. If you are seen as one who hoards data, most organizations will quickly correct that. But it happens at the individual, team, department, and company levels. Data creates a competitive advantage, and where there’s competition, there’s risk that data won’t flow freely.
Q: How does data-driven decision-making benefit the customer?
A: Data-driven decision-making is the ultimate way to show a customer that you're paying attention to them and respecting their time and interests. Remembering past interactions, good and bad, helps you get the experience right the next time. Customers benefit from more personalized and tailored experiences. Those experiences can lead them to be more loyal to brands that provide those experiences and, ultimately, spend more money with those brands.
Q: What are data-driven decision-making best practices?
A: When placing an emphasis on data-driven decision making, it’s important to start with an understanding of how you’ll use the data and then build your data systems with those use cases in mind. Build for purpose, as opposed to collecting all the data you can and deciding what to do with it later.
Q: What is the difference between being data driven and being data informed?
A: Some data-driven systems are hardwired to the data, so as conditions change in the data, actions will automatically happen. If the process pauses and asks someone to make a call, evaluate the data, and make a decision, then that's more data informed. A data-driven system is automated based on data inputs, but a data-informed system requires human action.
Q: What is a data-driven model?
A: Companies use data-driven models to get the specific data inputs necessary to determine how to proceed with their business.
There are many types of data-driven models, and which model an organization chooses to use depends on their goals and what information they need to discover. An A/B test, for example, is a data-driven model that offers two different options. A company will run both options against their audience to see which performs better, and then use the data collected from the test to inform future content or campaign strategies.
Most product development, marketing, sales execution, and sales operations programs have one or more data-driven models that they follow as part of their methodology. When you're selecting a model, it's valuable to evaluate it against your use case to determine where it may have blind spots or may have an unacceptable level of false positives. A false positive occurs when a data point indicates that things are in a desired state when they aren't. For instance, in a home alarm system, a false positive would be a motion sensor detecting motion, when it wasn't the kind of motion you intended the alarm to warn you of. If a plant is next to a fan, the moving leaves may set off the alarm. The alarm did its job of detecting motion, but it caught a plant, not a burglar.
In digital marketing, false positives can be very costly because they skew the data in your model, which is why it's key to evaluate any model against the use case you're looking at. It’s important to determine if there's going to be noise in the data that will cause you to make the wrong conclusions, and therefore the wrong decisions.
A vital aspect to data-driven decision-making is inspecting the data. You need to understand not just the actual information the data is telling you, but what its capabilities are. There are often limitations to what information can be gleaned from a specific data source. When evaluating a customer’s journey, for example, you might be trying to find which touchpoints had the largest impact on their decision to convert. With SEO and SEM, however, you can over-allocate credit to those interactions, instead of some other point that was really the ultimate conversion event.
Q: What are data-driven applications?
A: A data-driven application is a system that autonomously executes a task. There are two main kinds: applications driven on static data inputs, and applications driven on dynamic data inputs.
A regular thermostat is an example of an application driven on static data inputs. It receives input to tell it what the temperature of the room should be, determines if the room is at that temperature, and then runs the HVAC system if necessary. It only takes two data points: what is the temperature now, and what does the temperature need to be? It then takes an action, either to turn on the HVAC system or leave it off.
A smart thermostat takes more into account, and it’s an example of using dynamic data. It uses algorithms to learn about the environment and make increasingly complex decisions about how to operate. It will look at the current temperature and ask if it’s a suitable temperature for the season and the time of day. If it’s not the right temperature, the thermostat will decide not only what action needs to be taken to achieve the suitable temperature, but how long the process will take — and the exact steps required to reach its goal.
The data for data-driven applications comes from multiple different sources. Some data sources can be automated, while others are manual. Generally, for most systems, any kind of goal setting is either entered or refined by a user, but it can also be pulled from a standard database. For example, the thermostat could look on the internet and see the current outdoor temperature and understand that 70 degrees in the summer is very different from 70 degrees in the dead of winter.
If the application can get that data from the internet or other machine-based sources, it can pull that data automatically. Otherwise, it's going to require a human to input its goals. But the user might have different goals for different conditions, like different goals for winter and summer, for day and night, and for being on vacation and being at home. If a user tells their thermostat they’re away, then the user will have a much different tolerance for how warm the house should get in the summer and how cool it can get in the winter. In other words, humans are optimized for goal setting, while machines are for pulling normalized data.
Q: What are the disadvantages of a data-driven approach?
A: Data-driven approaches are not easy, and they’re not free. Data-driven decision-making requires a significant investment, but the return on that investment is exponential. The more you invest in being data-driven, the higher the yield is.
There’s also a delicate balance between personalization and invading customer privacy. You have to be very thoughtful about how you implement your practices so the customer doesn't feel like the privacy of their data is being violated. In addition to a customer’s own privacy preferences, there are many laws that govern what kind of data you can collect and keep about a person, and all of these should be kept in mind, which can be time-consuming and expensive.
Q: How has data-driven decision-making changed over time?
A: Everything has gotten better. The level of detail that’s possible to get in your data is continuously increasing. The cost of data storage is dropping. The difficulty of data collection is decreasing. The number of areas you can collect data about is increasing. The ability to stitch data sets together is increasing and the difficulty of doing it is decreasing. The ease of use of all of these tools are increasing. The number of trained data scientists and data analysts is increasing. The sophistication of both is increasing. The number of vendors and tools in this space is increasing, so the number of your options for doing data-driven decision-making are increasing as well.
Competition is leading to better tools and lower costs to store and use data. And digital natives are more comfortable with these tools. They're not just comfortable with the concept of using data to make decisions, but they're more familiar with the tools used to obtain and analyze the data. Ten years ago, you were an expert if you even knew what a pivot table was. Today, very few new college graduates aren’t familiar with spreadsheet programs, and there are no MBAs who don't know how to use pivot tables. In general, the workforce is increasingly data literate, and comfortable manipulating data.
Q: How will data-driven decision-making continue to evolve in the future?
A: New technologies like artificial intelligence (AI) are heavily influencing the practice of data-driven operation. Data-driven operating models are no longer optional. Every CEO is pursuing them or using them and enhancing them. And data-driven operating models are the new competitive advantage. Because of the impact that AI, machine learning, and ease of use have on democratizing access to and the use of data, those will be the trends that drive data-driven strategies in the future.