How to design a behaviour-change program in an afternoon

An afternoon is enough to design a program that will work, as long as you make the decisions in the right order. Most program design goes slowly because it starts with the sessions. Sessions are the last decision, not the first.

Here is the order we use, and the order the program designer in Paldar walks you through.

1. Name the capabilities you want to move

Start with what should be different in three months. Not the topic, and not the title of the program, but the observable capabilities. Giving difficult feedback. Running a meeting that ends with decisions. Coaching a direct report instead of solving the problem for them.

Write down two to four. More than that and the program becomes a survey of a subject rather than a change in behaviour. Goal-setting research is consistent on this: specific, somewhat difficult goals produce more change than vague or easy ones, and a small number of goals is easier to hold than a long list.

In Paldar, a program points at capabilities from the first screen. Everything you add afterwards is linked back to them, which is what makes the measurement step possible later.

2. Decide how you will know

Before you design a single session, decide what you will measure and when. A baseline survey in week one and the same survey in week twelve is the simplest honest design. For capabilities that other people see more clearly than the person themselves, a short 360 with a handful of raters works better than self-report alone.

The point is not the instrument. The point is that you have committed to a comparison before the program begins, so the result cannot be quietly redefined afterwards.

Paldar treats surveys and 360s as part of the program rather than a separate tool, and each question maps back to a capability. When the second survey comes in, the movement on each capability is already there.

3. Lay out the interventions across time

Now the sessions, but think in weeks rather than in a single event. A useful default for a twelve-week program looks something like this:

• Week one: baseline survey, then a workshop that introduces the capabilities and sets a personal focus for each participant.
• Weeks two to five: one short lesson a week, a practice prompt, and a coaching conversation with the AI coach or a human coach.
• Week six: a second workshop, smaller, built around what people have actually tried.
• Weeks seven to eleven: continued practice, a mid-point check-in, and a one-to-one session for anyone who has stalled.
• Week twelve: the follow-up survey or 360, and a closing conversation.

Spacing matters. The evidence on learning techniques strongly favours spaced practice over a single concentrated block, and behaviour change is a learning problem before it is anything else. A program that puts all its effort into the first day is choosing the least effective shape available.

In the program designer you add each of these as a row: workshops, coaching sessions, surveys, lessons, nudges. Each has a date or an offset from the start, a duration, and the capabilities it serves. The whole program is visible on one screen, which makes gaps obvious.

4. Decide what happens between sessions

This is the part most programs skip, and it is where most of the change happens. Between sessions, each participant needs three things: a reminder of what they said they would try, a short piece of material that helps them try it, and somewhere to reflect on how it went.

In Paldar, the reminders are scheduled from the program. The material comes from your content library, where lessons can be drafted with AI help and edited before anyone sees them. The reflection happens in conversation with the AI coach, which knows the program, knows the participant's focus, and stays within the boundaries you set.

What you have at the end of the afternoon

A program with named capabilities, a measurement plan, a schedule of interventions, and a between-sessions structure. Participants can be invited from it. Quotes can be built from it, because the durations and sessions are already there. And when it finishes, it can tell you whether it worked.

Paldar is pre-launch, so if you would like to try this with a real program, the waitlist is the way in for now.