Behavior is at the heart of nearly every challenge in the workplace, from leadership and fair decisions to high performance and AI adoption. But how should organizations go about influencing behavior? Most HR and leadership teams follow an intuitive model—they inform and inspire with education and communication campaigns, and they build capability with training and development programs. On the face of it, that seems like a watertight strategy. After all, if people are informed that a change is good for them and are given the capabilities to make that change, then surely change will happen.

Unfortunately, that’s not how things typically pan out in practice, and the result is a profound waste of resources and missed opportunity for impact. In a working paper from 2012, researchers reported that U.S. corporations spend $164.2 billion on training and education every year, but noted that most are unable to convert this substantial investment into individual behavior change or improved organizational performance.

For the past five years, we’ve been designing and testing an alternative model in a collaboration between MoreThanNow, a behavioral science practice founded by James, and a collective of academics, including Siri and Edward, from the Harvard Business School, the Harvard Kennedy School, Exeter University, and the University of Chicago Booth School of Business. This has involved partnerships with organizations such as AstraZeneca and Nationwide Building Society, who we share studies from below; published papers in journals such as Science and the Academy of Management Journal; and demonstrable impact for employees and organizations in the real world. Below we share the components of that approach, which we call the “4T model,” and discuss some experiments that have brought it to life.

The 4T Model

Our scientific model of behavior change entails four steps. First, you choose a specific behavior, decision, or outcome to target; second, you develop a theory of change; third, you design a timely intervention; and fourth, you robustly test your effect.

1) Target a specific behavior, decision, or outcome.

Behavioral targeting is both the easiest and hardest part of our model. Easy because there will be dozens, maybe even hundreds, of behaviors connected to a cultural aspiration such as innovation, inclusion, or high performance for you to choose from. Hard because you must ruthlessly prioritize the one(s) that will make the biggest impact, ideally using your data infrastructure to mine for insight and develop hypotheses. This work is often best done through collaborations among behavioral scientists and people analytics teams, who have often featured heavily in our partnerships with HR teams.

2) Develop a theory of change.

Once you’ve identified a specific behavior that you want to change, develop a theory of how you’re going to change that behavior. Often, this is an iterative process of talking to stakeholders and employees and identifying what barriers exist that keep them from changing, and looking at the scientific literature on behavior change for ideas or inspiration. There are few shortcuts here: Like a doctor, you must make your own diagnosis.

3) Design a timely intervention.

Timeliness is both an obvious and dramatically underused principle in our model. Once you’ve identified your target behavior and theory of change, you need to design and deliver your intervention when it matters—that is, when people have the opportunity to act. This might mean motivating people to use AI when there is a tool in front of them that they can try out; changing the way managers think about performance when they are considering their assessment ratings; or developing simple guidelines that managers will actually use to inform their daily conversations with direct reports.

4) Put your work to the test.

At this point, you’ve targeted a priority behavior, you have a theory of how to change that behavior, and you’ve identified a time and a place to do that. Now is the time to test if your intervention—or which intervention among multiple alternatives—will actually lead to improvement. We recommend using randomized controlled trials, where possible, to accurately estimate the causal impact of your intervention on the outcome(s) you care about.. . .

In recent years, we’ve increasingly heard leaders question the traditional approach to behavior change. There’s a burgeoning recognition that more training and more engagement campaigns aren’t leading to meaningful, measurable progress. We present our 4T model as a credible alternative to workplace behavior change, and one that we’ve tested successfully in many major organizations, including the partners mentioned here.

Although the steps of our model are straightforward, they require intelligent and committed application. When you target a specific behavior, you give up on the illusion that you can change everything at once—this model works best through portfolios of small changes that add up over time. The theory of change behind your ideas will not come cheap. Instead, it will emerge from deep diagnostic work with your stakeholders, your organizational data, and the existing research literature on the topic. When you design timely interventions, you must integrate your work into day-to-day experiences and decision-making processes rather than arranging events or campaigns at your own convenience.

Finally, of course, you must put your impact to the test. Proving that your intervention worked as intended will enable you to scale the 4T model to more workplace challenges or to course correct early. You will then be in a position to influence meaningful change among thousands of employees and teams—and prove it to your stakeholders. This last step will make it all count