A behaviour change workshop is not a day. It is a 90-day programme that includes a day. The day is where most of the visible activity happens, but the work that determines whether the day moves observed behaviour sits in the three weeks before it and the eight weeks after it. Most workshops fail at one or both of these. This post lays out the 90-day programme plan that consistently moves behaviour, week by week, for HR Directors and Heads of L&D who are running the design themselves or commissioning a provider.
Weeks 1 to 3: Diagnostic and Design
Week 1: Frame the behaviour
The first job is to convert the brief from a topic into a behaviour. Briefs arrive as "psychological safety", "leadership presence", "having difficult conversations", "DEI". None of these are behaviours. The week-one work is to ask the senior sponsor and the L&D lead a small set of forcing questions until specific behaviours appear.
What does a meeting look like in twelve weeks that does not look like it today? What does a manager say after a near-miss that they do not say now? Which conversation is currently not happening, by whom, in what room? The answer to these questions is the design brief. Without them, design defaults to a template.
Michie, van Stralen and West's COM-B model (2011) is a useful discipline in this week. For each named behaviour, check whether the gap is Capability (people do not know how), Opportunity (the context does not let them) or Motivation (they could but choose not to). Each requires a different intervention. A workshop that treats the three the same will half-solve one of them.
Week 2: Talk to real people
Three to five 45-minute conversations with people from the target population, on the record but not for attribution. The brief at this point is to test whether the behaviour gap the sponsor named is the behaviour gap the population actually experiences. Usually it is close. Sometimes it is wildly different. The cost of skipping this week is a workshop that solves a problem the population does not have.
Sample two to three of the line managers above the target cohort. The view from above is the second data layer. A divergence between sponsor view, manager view and participant view is itself a useful design input.
Week 3: Design the scenarios and the embedding plan
Two outputs by end of week three. First, two to four scenarios that mirror the actual situations in which the named behaviour fails. Scenarios are written like one-page screenplays: setting, actors, opening line, behavioural target, three plausible counter-moves to keep rehearsal alive. Second, the 60-day embedding plan, named and scheduled in calendars, before the workshop runs. Embedding is not a follow-up. It is half the programme.
Week 4: Delivery Day
The day itself has a recognisable shape if it has been designed properly.
Morning, hour one. Framing, psychological safety contract, behavioural target named explicitly. No corporate platitudes. The hour is structural: it sets whether the room is safe enough for participants to behave as they actually do.
Morning, hours two and three. First scenario. Run, debrief, name the moment where the default behaviour appeared. Rehearse the alternative against a different counter-move. Run again. Most participants meet their own default for the first time in this block. The debrief is where the design earns its keep.
Afternoon, hours one and two. Second scenario, designed to surface the same behavioural target in a different context. Cross-scenario transfer is the test. A participant who can run the behaviour cleanly in scenario one and falls back to default in scenario two has learned the script, not the skill. Two scenarios is the minimum for the deliberate practice that Ericsson's Peak (2016) describes.
Afternoon, hours three and four. Optional third scenario for complex behaviours, otherwise structured reflection and embedding plan walk-through. The embedding plan is handed out, paired buddies are matched, the first week's micro-practice is named, the day-30 group reflection is in everyone's calendar before they leave the room. This is the difference between a workshop that decays and a workshop that lands.
Weeks 5 to 8: The Decay Window
The transfer-of-training literature is consistent: most workshop learning decays within four to six weeks without structured embedding. The first four weeks after delivery are the highest-risk window. The 90-day plan treats them as the most important phase of the programme.
Week 5: Micro-practice in real work
Each participant runs the new behaviour at least twice in real work this week, against a named situation. The paired buddy observes once (in person or remotely) and gives a single specific piece of feedback within 24 hours. The L&D team checks in by email to confirm the practice happened, not what was said.
Weeks 6 and 7: Repetition and reinforcement
Continued micro-practice. The buddy pair runs a 20-minute reflection at the end of each week, on what happened and what was hard. The principle here is Roediger and Karpicke's: behaviour rehearsed in the conditions it has to hold in is roughly 50% more durable than behaviour rehearsed only in the classroom.
Week 8: The 30-day reflection
A single 90-minute group session, all participants back together, in-person if possible. The reflection has a clean structure: what behaviour was hardest to actually attempt, where did the default come back, what shifted in the team meeting that did not used to shift. This is also the diagnostic moment for the second half of the embedding phase. Anything that has not moved needs different support.
Weeks 9 to 12: Consolidation and Measurement
Weeks 9 to 11: Sustained practice
Continued micro-practice, lighter cadence than the first month. The behaviour should now be appearing in real work without being prompted. The paired buddy structure continues but with monthly rather than weekly check-ins. Optional one-to-one coaching is available for participants who have not yet stabilised the behaviour.
Week 12: Behavioural measurement
The measurement step is what separates the programmes that prove ROI from the programmes that hope for it. Three layers, drawn from the Kirkpatrick model (Kirkpatrick & Kirkpatrick, 2016).
Layer one: self-report. A short survey on what each participant is doing differently. Useful as a hygiene check, not as evidence on its own.
Layer two: 360-style observation. Direct reports, peers and line managers report on observed behaviour in real work. Two specific behavioural questions, not a general satisfaction sweep. This is the Kirkpatrick Level 3 evidence.
Layer three: downstream metric. Where the behavioural target is connected to a business metric (speak-up rate, first-time-right on cross-functional handoffs, retention in the target population, complaint resolution time), the metric is sampled at day 90 against a baseline taken at week zero. This is Kirkpatrick Level 4.
The reason measurement matters is not only ROI. It is the diagnostic step for the next cohort. Anything that did not move tells you something about the design that has to change.
What Sponsors Need To Do
The 90-day plan works because it is sponsored, not delegated. The senior sponsor (CHRO or operating equivalent) has four obligations in the cycle.
First, name the specific behaviour and own the brief at week one. Without the sponsor's specificity, design defaults to template.
Second, model the behaviour in their own meetings during weeks 5 to 12. Programmes where the senior sponsor visibly does the new behaviour land at materially higher rates than programmes where the sponsor delegates and disappears.
Third, protect the embedding plan from being deprioritised. The single most common failure mode in the 90-day cycle is the sponsor letting urgent operational work eat the week-eight reflection and the week-twelve measurement. Both are non-negotiable.
Fourth, read the measurement at day 90 and act on it. A programme that produces evidence the sponsor does not look at is a programme that does not produce evidence.
What Sidestream Adds
Sidestream runs this 90-day shape consistently, with the rigour of organisational psychology (UCL, Cambridge, Bocconi) and the craft of immersive theatre. Two worlds that almost never meet. We design the scenarios with professional actors, we structure deliberate practice across rehearsal waves, we name the embedding plan before delivery, and we measure observed behaviour.
We work with the Metropolitan Police, UCL, the University of Cambridge, Bocconi University, Goldsmiths and TCS. The 90-day plan above is the plan we run, simplified for a single cohort. For larger populations the design layers cohorts across a quarter, with the same shape inside each.
If you are scoping a 90-day behaviour change programme for your organisation, the cleanest next step is a 30-minute working conversation about week one of your specific cycle. Bring the brief. We will start the diagnostic.
Book a free 30-min consultation. Or read the companion piece on what makes a behaviour change workshop work and the piece on behaviour change versus awareness training.
We are Sidestream.