Marketing PDCA: Plan, Test, Review and Decide with a Practical Worksheet
Shusaku Yosa
Marketing PDCA is a cycle of planning a hypothesis, implementing a change, checking the evidence and deciding whether to continue, revise or stop. Its value is the learning and decision record, not the number of changes made each day.
Choose a review rhythm that fits traffic volume and the time to purchase or close. Plan defines the test, Do records execution, Check examines outcomes and alternative explanations, and Act specifies the next decision.
Follow one initiative through the cycle
Consider a fictional B2B resource-request page. The hypothesis is that explaining the resource and what happens after submission will increase useful completions. Do not assume that removing fields is always the answer.
Stage | Record | Example |
|---|---|---|
Plan | Problem, hypothesis and evaluation rules | Clarify the resource and follow-up process |
Do | Scope, date, change and incidents | Change the explanation while retaining fields and targeting |
Check | Counts, denominators, quality and confounders | Review completion, valid inquiries, devices and traffic mix |
Act | Decision and reason | Continue, revise, stop or gather more evidence |
Set the decision rules first
Document the page, traffic, device scope and exclusions; the primary count and rate; guardrails such as errors, target-account fit and sales acceptance; the cost ceiling; start and review dates; and the comparison method.
Required sample size depends on the baseline, meaningful difference and accepted uncertainty. There is no universal “100 visits is enough” rule. For a simultaneous comparison, define assignment and the primary decision criteria before viewing outcomes.
Record implementation and tracking
Keep release time, changed elements, scope and rollback instructions. Test success, failure, reload and duplicate submissions. Changing the explanation, audience and form at once makes interpretation harder.
Use consistent UTM naming for external ads or email, following Google's campaign URL guidance. Do not attach acquisition UTMs to internal related-article links to relabel the source; use a separate event design if internal interaction is the question.
Separate an observed difference from proven effect
If a fictional before period has four completions from 100 sessions and an after period has six from 100, the rate rises from 4% to 6%. That is a 50% relative increase but only two additional completions. It does not by itself establish an effect. Check assignment, traffic mix, weekdays, tracking and valid-lead quality.
When sales take months, follow early requests separately from mature opportunities. Current-month bookings should not all be attributed to a current-month page change.
Decision | Evidence needed | Next step |
|---|---|---|
Continue or expand | Credible improvement without unacceptable quality loss | Define rollout and monitoring |
Revise | A plausible benefit with unresolved fit or message issues | Specify the next hypothesis |
Stop or roll back | Failure or unacceptable deterioration | Recover and investigate |
Defer judgment | Insufficient data, tracking issues or immature outcomes | Define missing evidence and a new review date |
A deferred decision is useful when the reason and deadline are explicit. Keep the hypothesis, execution, result, uncertainty, decision, owner and deadline on one record so the next cycle builds on what was learned.
Related practical guides
- Marketing KPIs: How to Build a KPI Tree and Metric Dictionary
- Marketing Data Analysis: From a Business Question to a Tested Decision
- How to Build a Marketing Dashboard: Metrics, Layout and Data Quality
- Marketing Insourcing: In-House, Agency and Hybrid Models Compared
Connect execution, costs and reviews
Xtrategy supports project management, monthly budgets and actuals, and customer and deal records. Use a 30-day free trial to evaluate the workflow against your team’s needs.
References
Official sources checked September 29, 2026. Product pricing and availability may vary by market, plan and contract.