When a change stalls, the usual move is to push harder: more meetings, more pressure. Kurt Lewin noticed that often works badly. Removing what's pushing back works better.
What it is
Kurt Lewin, one of the founders of social psychology and organizational development, introduced force field analysis in the 1940s. The picture is simple:
- The current situation is a balance point.
- Driving forces push toward the change.
- Restraining forces push against it.
- The situation stays put while they balance.
To move it, you can add driving force, or reduce restraining force. Lewin's insight was that adding pressure tends to raise resistance. Reducing a restraining force often moves things with less friction.
Each force gets a rough strength, often 1 to 5, and the diagram shows arrows of that size pushing on each side.
Why it matters for bots
Rolling out bots is a change. The technical side is often the easy part. The restraining forces are things like fear for jobs, distrust of artificial intelligence, extra review work, and "we tried a tool before and it failed."
Why people skipped it
It needs honest input about what's holding things back, and it's tempting to skip straight to the rollout plan.
How a bot does it
- Give the bot the plan, plus anything that shows how people feel: survey comments, meeting notes, emails, the stakeholder analysis.
- Have it list driving and restraining forces, each with evidence.
- Have it rate each force, marked as a guess where evidence is thin.
- Have it propose ways to reduce the biggest restraining forces.
- People correct the strengths and choose actions.
The bot brief
Draft a force field analysis for [change].
Inputs: plan [file], stakeholder analysis [file], comments/notes/emails [files].
List driving forces and restraining forces. For each: one-line description,
evidence (quote or cite), strength 1-5 (mark GUESS if evidence is thin).
Total each side.
For the 3 strongest restraining forces, propose one way to reduce each
(not a way to push harder). Note who would need to act.
Keep it about forces and roles, not individuals.Starting point only. Test it on your own data and model, and adjust as needed before you rely on it.
An example
Change: office bots draft all routine correspondence.
| Driving | Strength | Restraining | Strength |
|---|---|---|---|
| Staff overloaded with email | 4 | Fear of errors going out under their name | 5 |
| Owner wants faster replies | 4 | Reviewing drafts feels like extra work | 4 |
| Bot tests went well | 3 | Past software rollout failed | 3 |
| Unclear who's responsible for a bad send | 3 | ||
| Total | 11 | Total | 15 |
Reducing the biggest restraint: nothing sends without the person's yes, and their name stays on only what they approve. See Nothing ships without a yes. That one rule may do more than any amount of pushing.
Where bots differ
A bot can only see the resistance that shows up in writing. The strongest restraining forces are often unspoken. So the bot's list is a starting point. Show it to a few trusted people and ask: "What's missing on the right side?"
What this means for you
- Before rolling out bots, list what's pushing back.
- Have a bot draft both sides from what's written.
- Fill in what isn't written.
- Work on reducing resistance before adding pressure.