Work process experiments

A work process experiment is a limited change that answers a practical question before you alter everyone’s routine. Choose one recurring difficulty, define the result to improve, set a trial boundary, and decide in advance what would make you keep, revise or stop the change. Judge the result by what happened to the work and to the people doing it.

Start with a question the trial can answer

Use a difficulty visible in actual work: requests wait for an owner, a handover needs repeated clarification, or a recurring meeting leaves a decision unresolved. Specify the work type and the point of difficulty. “Will one named intake coordinator reduce unowned requests without slowing urgent responses?” is more useful than “Can we improve communication?”

Speak with the people who send, handle and receive the work. Check whether the apparent problem is excess demand, a missing decision or unclear instructions. Write the proposed change in one sentence.

Identify the service coverage, approvals and quality checks that must continue. If the change affects duties or workload, consult the workers involved and allow for any support or training they need.

Set a boundary and a comparison

Choose a work type, team or period small enough to supervise but broad enough to include ordinary cases. Record who owns the trial, when it starts and ends, and how the previous route can be restored. Tell affected colleagues which route applies.

Before the change, inspect comparable recent work using the definitions you will use during the trial. Record only what helps answer the question: for example, when a request arrived, when it gained an owner, when its recipient could use the result and whether it returned for correction. Note unusual demand, absences or changes in scope.

A before-and-after difference may have other explanations. Observing a comparable group using the existing route over the same period can help, if service and fairness allow it, but differences between groups still matter. Treat the comparison as evidence for a local decision, not proof that the process alone caused a change.

Before vs. after: measuring impact in a trial

Watch the whole workflow

Choose one or two signs of the intended benefit and check for harm. A faster first response is of little value if final resolution takes longer. Fewer meetings may release time while moving decisions into a long message chain. Ask both workers and recipients where effort has moved.

Use existing work records where possible. Give people a way to report a missed handover, unreasonable demand or other concern before the scheduled review.

Separate outcomes from causal claims

Include unintended outcomes as well as the result the trial was designed to improve. Outcomes may be short-term or longer-term, so be clear about which ones the trial review can actually observe.

Decide what the evidence warrants

Set a review date and stopping conditions before the trial begins. Pause promptly if a live service, necessary control or worker’s wellbeing may be at risk. Establish a dependable route for current work and examine the concern with affected people.

At review, ask whether the change was followed and whether the intended result improved. Keep it within the tested scope, revise and try again, return to the previous route, or gather more evidence. Record the reason and tell those affected. Before wider use, check whether other teams have different demand, skills, authority or service cover.