Is “data-driven” just slowing down your decisions?

We're kicking off a guest series with exclusive content from Ben Rogojan, the data consultant behind Seattle Data Guy. Stay tuned for more.

For decades, companies have been chasing the idea of being data-driven. I recall when I first started the term was plastered everywhere. 

Every blog you read and conference you went to, someone referenced that term along with the idea of data being the new oil. 

Everyone wanted to be driven by data. After all, companies like Facebook and Google were doing so well. And not just them, there were plenty of statistics that showed that companies that used data were more profitable than their competitors.

Here are a few:

• Brainforge: Data-driven strategies make companies 85% more likely to achieve significant revenue growth.

• McKinsey: Data-driven organizations are 23x more likely to acquire customers and 19x more likely to be profitable.

• PwC: Data-driven companies outperform competitors by 6% in profitability and 5% in productivity.

• Forrester Consulting: Companies using data tools for decision-making are 58% more likely to hit revenue goals and 162% more likely to exceed them. 

Who doesn’t want to be more profitable?

The problem is, for many companies, the idea of being data-driven became a distraction. Companies were so focused on building dashboards and data warehouses that many forgot to stop, and just ask why. 

Why this dashboard, why are we building a data warehouse and what are the actual business goals? In turn, they found themselves getting in their own way. In this article, I want to dig into how being data-driven can actually slow your company down and how to use data for actual impact.

How data-driven has slowed down teams

For many companies, the idea of being data-driven has become a distraction and has slowed them down from making a decision. 

What that looks like can vary. 

Here are a few common ways I’ve seen data-driven slow down data teams.

• Analysis paralysis - This is something you’ve likely heard about in your business class and if not, it’s pretty self intuitive. As the name suggests, this involves being so overwhelmed by data and analysis that you freeze. You make no decisions at all. Instead, you’re waiting for the perfect answer. You keep finding new questions to ask.

• Infrastructure for infrastructure’s sake - In some cases, teams don’t even get to the analytics portion of “data and analytics.” Instead, teams spend months on picture-perfect pipelines and new tools instead of delivering outcomes. I’ve seen data teams test out multiple data solutions to try to find the perfect data ingestion tool or BI tool. All the while, the business tapped their foot waiting and frustrated.

• Leadership gridlock - Not all the issues that keep a business from actually using data are because of the data team. Executives also hesitate and wait for “complete” data. Maybe they just don’t like the data so they refuse to make the decision they know is right. In this case, there can be a lot of crossover with analysis paralysis.

• Perfection trap - In particular, the perfection traps I’ve seen are often due to a ML model not being perfect. So the company keeps pushing for new data or features to be added to the model, even when no one knows if the model will be successful or not.

The truth is, the term data-driven came to almost mean “data will give you the exact right answer,” which isn’t how it works. Likely, this is why many people push for the idea of “data-informed” because that really is all the data can do. It can give you insights and possible trends, but it can’t make the decision. 

How much data do you really need to decide?

There are many reasons companies fail to use data to drive decisions. But waiting to have enough data and metrics is one I see over and over again. 

That’s why I think Jeff Bezos has the 70% rule that states:

"Most decisions should probably be made with somewhere around 70 percent of the information you wish you had…if you wait for 90 percent, in most cases, you’re probably being slow." – Jeff Bezos

People are constantly waiting for the perfect answer from data, and the truth is, there likely isn’t one. Even once you’ve made a decision, you’ll find a whole new set of problems. Maybe you build a model and deploy it, and find that even though you calculated that it should return a massive ROI, it doesn't. At least not right away.

So, you go through a few more iterations in the wild, and eventually it clicks. I’ve seen this happen several times because often there are factors that you didn’t account for, but you’d never learn about until you deploy said model.

Changing the mindset

Sometimes an industry has been so focused in one direction for so long it becomes necessary to shift into the opposite mode of thinking. For example, data-driven became a popular term for vendors and consultants to throw around. 

They’d point to big tech and say, don’t you want to be like them? But it’s important to understand that those businesses — data didn’t drive the business, it was critical to the business itself.

Data wasn’t an afterthought; it was the lifeblood. 

If they weren’t using data well, they would have failed. In many ways, they were less data-driven; they were business-driven. 

The business just demanded data, and the usage of data in order to compete. Facebook used it to better serve ads than newspapers, Google used it to automate the cataloging of the web, and calculate intent to sell ads.

What is the actual business-driven reason your team needs to use data?

Building your company’s “decision-driven” muscle

Becoming decision-driven doesn’t just happen overnight. You can’t just start building dashboards and assume that’s all you’ll need to use data to drive decisions. And that’s a key statement: you’re looking to use data to help inform and drive decisions. Or a better way perhaps to word it is, is to drive certain outcomes. 

That’s how I’ve started to describe it for companies. They need to become outcome-obsessed instead of data-driven or obsessed. This is where you’ll first need to start.

• Redefine the goal - Shift from data-driven to decision-driven or outcome obsessed. Data should be used to inform, but you can’t always wait for the perfect answer to make a decision. Sometimes you need to test out an idea, be wrong, adjust, try again, and keep going.

• Right-sized data for the decision - Does every business question need a multi-day analysis? Probably not. In some cases, smaller decisions that don’t drive as large of an impact might allow for you to do faster analysis. On the flip side, sometimes you’re treating a multi-million dollar decision as if you need to make the choice tomorrow. As you are defining the outcome you want, partner it with the amount of rigor or analysis appropriate for the impact.

• Fast feedback culture - Around all of this, a key component is increasing the speed at which you can test out new ideas and iterate. If it’s taking you a month to go through a single cycle, it’s likely too long. Figure out how you can go from iteration one, then two faster.

What can you implement today?

Now, besides resetting your current data analytics process, here are some smaller steps you can put into play today.

• Set a decision threshold (and timeline) - For your next analysis, be clear with when you’d be comfortable making a decision. Should there be a time boundary, or just some level of confidence to when you should go forward? Otherwise, what I often see happen is teams loop over the same decision every week on a stand-up call. No one wants to be the one to make the call. Instead, people come up with more questions to keep asking, and based on my conversations with analysts, they don’t tend to enjoy being stuck answering yet another set of follow-ups.

• Avoid vanity metrics and data overload - A major issue that some companies have is too many metrics, as well as metrics that really don’t drive much. They just sit there on dashboards, and when things are good, leadership praises themselves, and when they are bad, they ignore them. In turn, this leads to data overload. What metrics and levers can your team actually impact? 

• Invest in data literacy and tools, to a point - Data teams and business teams are constantly speaking different languages. So once you’ve even put together a report or analysis, there still might be some gap in terms of the business taking action. A great way to fill this gap is to create a basic data literacy program. Now I am someone who will often suggest that the data team needs a business literacy program, but that doesn’t mean I don’t think the business needs to close the gap as well. In some places, tools can help either by highlighting data points that should be analyzed deeper or providing some basic decisions.

• Encourage “bias for action - Finally, create a space where your business and data teams feel safe to take action and make a decision. You, as a data leader, might need to be the one that demonstrates what this looks like at first. Perhaps you’re trying to figure out where to invest marketing dollars, help demonstrate what a good analysis looks like, and how to recommend and take a stance on where you think action should occur.

Final thoughts

Data is a powerful guide. It can help you detect insights and anomalies that would be missed. But it doesn’t always show the whole picture. It is, after all, a derivative of a far more complex real-life set of actions. In order to ensure your team doesn’t find itself paralyzed by data, you should create a process that helps you go from business needs to clear outcomes, and in between that is your data layer, where there is space to fail and iterate quickly.

At the end of the day, saying you’re data-driven and actually using data to inform decisions that you act on are very different things. You’ve got to do more than just deliver dashboards. You’ve also got to figure out how to cross the gap between the business and the data. You’ve got to be able to keep the project and initiatives rolling even when people and political issues get in the way.

Much of this is a cultural shift, not just shifting for the data team, but the business team. Both are working together to drive impact.