A neat set of frameworks for creating the organizational behavior change you seek
So much of the organizational work we do aims to change (or “sustain” as a special use case) certain behaviors in our teams. What’s the point of articulating a strategy if it doesn’t cause our teams to change their default behaviors and take a different set of actions to pursue it? What’s the point of a training or a workshop if the behavior before and after it is exactly the same.
The team at Affective Advisory developed a pretty neat framework for driving strategic behavior change:
The D.R.I.V.E Prism for identifying and evaluating effective nudges in practice
The framework builds on the core premise that human behavior is context-dependent and therefore, behavioral interventions have to be tailored to an individual context. Therefore, there are no universal interventions that are effective independent of context. Theoretical concepts need to be adapted into practice using a model-based, evidence-led approach and then tested and validated.
The overarching framework for applying behavior interventions is outlined under the acronym D.R.I.V.E:
- D.efine strategy as a set of preferred target behaviors.
- R.esearch actual (current) behaviors and review related contexts relevant to the strategic challenge.
- I.dentify, evaluate, and adjust suitable science-based solutions.
- V.alidate the selected and tailored interventions across a representative sample.
- E.xecute behavioral interventions realizing behavior change at scale.
Other than the neat acronym, there’s nothing earth-shattering here: define the change you need to make → understand the current situation → select an intervention → do a small pilot to ensure it drives the desired outcome → scale.
The non-trivial piece comes in the middle of the process. Given the premise, how do we identify and adapt the right interventions that are most likely to drive the desired outcome in this particular context?
This is where the prism comes in:
The prism is a three-dimensional taxonomy for categorizing different interventions and selecting the ones most likely to be effective in the specific context.
Dimension I: Intervention levers
The underlying thesis here is that behavior can be changed by a combination of four different levers:
- Contextual triggers
- (De)Motivators (automatic <-> reflective)
- Individual capabilities (psychological & physical)
Different interventions use a different mix of these levers.
This construct can also be mapped to a similar behavior change model that I covered here providing further support to the taxonomy: contextual triggers → reinforcements mechanisms, motivators → understanding and conviction, individual capabilities → confidence and skill-building, feedback → role-modeling.
Dimension II: The cognitive level
The levers outlined in dimension I can be designed to influence behavior through two cognitive levels:
- System 1 — unconscious, automatic, affective, effortless.
- System 2 — conscious, deliberate, controlled, effortfull.
Interventions working through system 1 aim to either leverage or mitigate some of its unique attributes, for example: auto-saving documents based on a time trigger.
Interventions working through system 2 aim to intentionally activate it to correct a system 1 driven behavior, for example: opening a dialogue box reminding users to save their file when they try to close it.
Dimension III: The intervention level
Here, the taxonomy distinguishes between:
- Adding new enablers.
- Removing existing blockers.
For example, a sign showing your current driving speed adds an enabler to drive the desired behavior. While default options and opt-out remove blockers by eliminating the need for a decision to reach the desired behavior.
As it stands today D.R.I.V.E and specifically, prism, are useful tools for honing in on potentially effective interventions in a particular context. Especially if a specific intervention has been tried and didn’t work — exploring a different assumption around one of the prism dimensions may help identify a strong candidate intervention quicker. While contexts are inherently different from one another, I wonder if there are contextual patterns that make a certain type of intervention more likely to succeed than others. If that’s the case, a similar contextual taxonomy can be developed and more resilient mapping between contextual patterns and effective interventions can be drawn.