CRM campaign measurement and experiment workbook

Send volume, clicks and completed work answer different questions. More clicks require separate evidence before concluding that the desired behavior increased. Record each metric’s unit, denominator, source and observation period, alongside adverse outcomes such as opt-outs and duplicate reminders.

In fictional example IXC-DEMO-01, the goal is shared work completed after accepting an invitation. Match shared_task_completed to invitation_accepted using the same business correlation rules. The completion definition for measurement must agree with the exit condition in the Journey design.

Even when a message targets an individual, teammates may influence each other. Consider workspace-level assignment and analysis. Record both units and decide whether a returning participant keeps the same group.

Separate the recruitment period from each participant’s observation window. If the worked example uses seven days after acceptance, someone entering on the final recruitment day also needs seven days before a comparable result is available. Do not selectively remove late entrants or missing events from one group. Record incomplete observation and its treatment.

For a copy A/B test, check that audience, timing and other exposures are comparable. When comparing an additional reminder with no additional reminder, retain required communications unrelated to the experiment. A before/after comparison also needs consideration of product changes, acquisition mix and seasonal effects.

OneSignal Split Branch separates Journey paths. Check reassignment on re-entry and overlapping Journeys against the official Actions documentation and the actual configuration. Define sample, duration and stopping criteria before execution; do not stop simply when a favorable result appears.

The following synthetic values are for calculation practice and assume complete observation for each unit.

54÷200=27% and 42÷200=21%, an observed difference of six percentage points. This calculation alone does not establish that the campaign caused the difference. Check assignment stability, cross-team influence, other campaigns, missing data and sampling uncertainty. A non-revenue completion event does not establish revenue or ROI.

Server completion events and channel reports have different sources. Record how repeat completions are counted, how user/workspace/invitation IDs are matched, and when late events stop updating the report. Compare available reporting fields with OneSignal Analytics.

An absent numerator or denominator leaves the result unavailable. A measured zero is different from a missing value. A zero denominator or completions exceeding the denominator requires another scope check. Record observed values, interpretation constraints and the next questions together.

The English PDF and CRM Excel workbook contain experiment, observation and interpretation forms plus a completion-rate calculator. Record observations, further checks, a keep/change/stop decision, and an owner with a review point. Link results to the configuration version to reproduce the conditions later.

IXC’s CRM strategy consulting and campaign operations use current goals, reports and event gaps to define instrumentation, experiment design, analysis and subsequent operations.