[Short on time? Read the TLDR version at the end.]
Most organizational leaders accept the power of culture in the workplace, and many work deliberately to foster the culture they believe best serves the organization’s goals.
But what does it really take to change a culture that is already established?
The thing is that culture is much more than shared understanding and adoption of organizational values that somehow lead people to align how they behave.
Culture is a system that consists of everything about the way an organization works.
That includes the ways people communicate and collaborate and all the incentives (and disincentives) that keep them doing it, but also all the procedures and tools that they use. And while the system in an organization may not have been deliberately designed to be the way it is, it has evolved over time so that every part supports every other part, and any disruptions get … taken care of.
So, if you want to change the culture of an organization, you have to change everything.
That goes for building a data-informed organization too. Whether you want to implement “evidence-based” programs, introduce AI, or just better manage performance using data, you are dealing with a (probably significant) change in culture. So you too have to change everything.
Ack!
Fortunately, you don’t have to change everything everywhere all at once. But you do have to approach it systematically: you need a system to counter the existing system.
The key is to balance the three categories of effort that I introduced in the first article of this series: leadership and accountability, capacity-building and support, and data management and governance.1
Think of them as the legs of a stool standing on uneven and regularly shifting ground. You should assume you’ll need all three legs, lest you topple over, but the required lengths you’ll need for each will be different, depending on what and where the organization is at any given time. A large organization differs from a small one and a department from a whole organization. And of course the needs when you’re just starting are different than later on.

In this and the next issue I will explore each of the three legs of the framework. I’ll briefly cover what each encompasses, but I’ll also assume you’re just getting started and call out a good candidate for what to do first. We obviously won’t exhaust any of them but, over time, we’ll return to all of them again and again.
The First Leg: Leadership and Accountability
This is the first leg because change generally doesn’t happen without someone leading the way and without some sort of incentives. But what does that look like for data-informed culture specifically?
This one’s actually simple. It just means requiring people to take on the three commitments:
Clearly define the specific results or outcomes you want to achieve.
Find objective ways to tell if your efforts are working.
Take action and change course if they are not.
So what do you actually need to do?
First, of course, is to articulate the vision. Tell people not just what to do, but why and, most importantly, why it makes a difference for them. There are many good things about a data-informed organization: better results, the ability to be more proactive, more efficient use of resources, etc., but those are necessarily generic. What are the compelling reasons for your organization, specifically?
Second, make sure to provide the needed support and resources. Data-informed approaches reduce work over the long term, but also tend to front-load some of the effort. For a normally reactive organization, supporting this up-front investment is a new and unnatural act and requires real leadership support. As I’ll discuss next week, you’ll also need to invest in training, support, and tooling.
Finally, model the behaviors you expect and hold people accountable. Regardless where you are on the journey, people will attempt to get around the (significant) work that the commitments require and it is critical that leaders block that. That goes especially for the leaders themselves. Nothing is more destructive of an effort to change culture than when leadership gives themselves or chosen subordinates a pass.
If you are just getting started, focus on the first commitment and ignore the other two for now. Most organizations tend to commit to activities and just assume that good results will ensue.
If that’s your organization, I guarantee that getting people to articulate exactly what should result for those directly impacted by a service will be transformative. As a bonus, it will also naturally get people thinking about how they’ll be able to tell (commitment 2) and, if the results are otherwise, what they can do differently (commitment 3).
Next week I’ll take on the other two legs of the stool.
Further Reading
The article above is part of a larger series. Here it is so far:
Links & Thoughts
It’s the dull stuff, dummy! Jennifer Pahlka recently highlighted the fact that so many of the critical issues we care about operate within a "force field of tedium." This is a superpower for introverts at cocktail parties, but problematic if you need people to support the long slog of reform. It’s easy to say that people should care about issues because they’re important, but they don’t and won’t. Wishing they would is just an excuse for us to avoid the hard work of creative communication.
“Beatings will continue until morale improves” department. Gallup’s latest State of the Global Workplace report offers the shocking insight that good management, highly engaged employees, and high organizational performance are highly correlated. In particular, while the percentage of most dissatisfied workers depends on country and macroeconomic factors, the percentage of highly inspired workers does not. That depends ... again ... on good management. I would add that good management and the practices of a data-informed organization are not uncorrelated.
tldr
Culture is a system that consists of everything about the way an organization works. So changing the culture of an organization means changing everything. That goes for building a data-informed culture as well.
That requires a systematic approach. For data culture, there is a framework that helps think strategically about what practices to introduce and how to connect them.
The first leg of the framework is leadership and accountability, which consists in ensuring that the organization follows the three commitments of a data-informed organization: clearly defining outcomes, finding objective ways to tell if they’re being achieved, and changing course if they are not.
Implementing this leg requires three key actions:
Articulate the vision,
Provide necessary support and resources,
Model the behaviors you expect and hold people accountable.
The best place to start is to focus on the first commitment since it naturally leads people to begin thinking about the other two.
I help governments and nonprofits think about how to use data to improve results for their communities. To learn more about what I do and how we can work together visit DeepWeave.com.
If you want to share thoughts on anything I’ve said here or have ideas about further questions or topics you’d like me to explore, please feel free to reply to the newsletter email or contact me here.
A Note On The Framework. You might wonder where this framework comes from. I will confess that I’ve been known to invent new frameworks on the spur of the moment, so it’s a legitimate concern.
This one is quite well grounded, however. It derives from the What Works Cities (WWC) certification program, which is designed to guide cities in becoming more data-informed. You won’t find these exact 3 categories there. Certification is based on 43 criteria across 8 practice areas which, honestly, even as an expert I find overwhelming.
So over the last few years I’ve evolved a simplified framework, ending up with these three categories. It doesn’t replace the WWC-identified practices. Rather, it groups them in a way that helps leaders to think strategically and systematically about which practices should be introduced and how they interconnect at any given point.