A trial needs a decision at the end

The fictional support team at Linden Desk repeatedly receives internal requests with missing context. A manager proposes a new intake form. The team could roll it out everywhere, declare the project finished, and discover a month later that colleagues now send requests through chat instead.

PDCA stands for Plan, Do, Check, Act. ASQ describes it as a repeating approach to testing changes and using what is learned to guide improvement. Linden’s case and proposed measures are hypothetical; no trial results are claimed here.

Plan around the problem rather than the form

Linden defines the problem as requests that require clarification before useful work can begin. The proposed change is a short form asking for the affected service, the requested outcome, and relevant context.

The team predicts that better initial context will reduce clarification, but the form could also discourage requests or shift them to another channel. A useful plan includes both the desired effect and those possible side effects.

Plan → do → check → act → next plan

Define the change

Plan

Describe the problem, proposed change, expected effect, and measures.

Try it on a limited scale

Do

Run the planned trial and record deviations.

Compare and learn

Check

Examine results against expectations, including unwanted effects.

Decide what follows

Act

Adopt, adapt, or stop based on learning, then begin the next cycle.

PDCA is a repeating improvement cycle. The action after checking depends on the findings; wider rollout is not automatic.

The team proposes a two-week trial with one internal group. That window is a practical choice for this illustration, not a PDCA requirement. Real timing should reflect request volume, variation, and the consequences of making a wrong decision.

Define measures before starting

Linden records the proportion of eligible requests needing clarification and the median time until work can begin. It also checks form abandonment, requests arriving through chat, and the staff time needed to process submissions.

A completion rate alone would be misleading. A form can be completed perfectly while asking the wrong questions. Faster intake can also move the bottleneck to a later stage without improving the requester’s experience.

Question Proposed measure Interpretation to watch
Is context improving? Eligible requests needing clarification Staff may classify clarification differently
Does work begin sooner? Time from request receipt to useful work Changes in staffing or complexity may explain a difference
Is the form creating friction? Abandoned submissions and alternative-channel requests Lower form volume may reflect avoidance
Is the process economical? Staff processing time Extra manual cleanup may hide behind faster response

The measures and cautions are FrameworkWorld’s application advice. The team needs consistent definitions and comparable records to interpret them.

Do means run the planned trial

One group uses the form while the support team logs departures from the plan. If a question confuses people, record the confusion and any mid-trial change. Do not quietly revise the form several times and later describe the period as one stable test.

Support should remain available for unusual or urgent requests. The trial plan needs a way to handle those cases without losing them from the record. Exceptions can reveal important design needs rather than inconvenient data to discard.

A small trial limits exposure and helps reveal practical issues. It does not remove the need for appropriate controls when the change affects important services.

Check the prediction and the tradeoffs

At the review, Linden should compare results with the baseline and the prediction. It should inspect request mix and staffing before concluding that the form caused a change. A quieter fortnight can make almost any process look quicker.

If clarification falls but chat requests rise, the team should investigate whether people avoid the form. If useful work begins sooner with no obvious additional burden, that supports a broader trial, while leaving uncertainty about other groups.

Checking is a learning discussion, not a presentation assembled to defend the initial idea. Negative findings can prevent an expensive rollout.

Act on what you learned

Linden can adopt the change with guidance and ownership, adapt confusing fields and test again, or stop if the form does not address the problem. The act step is the decision following the check, not a synonym for carrying out the original plan.

If adopted, define who maintains the form, how exceptions are handled, and when the measures will be reviewed. A process can deteriorate as the service changes. Improvement remains a cycle rather than a completed installation.

Use SMART Goals to make the intended improvement and review point explicit. Use a RACI Matrix when responsibility for the trial or the resulting process is unclear.

PDCA does not supply a statistical evaluation method or identify the cause of every problem. Linden’s immediate task is to run a bounded trial with enough evidence to decide whether to adopt, adapt, or stop.

Sources and further reading