Voice of Customer
Part of Customer experience prioritisation
Deciding which CX finding needs more evidence
Decide when to investigate a CX finding by identifying the uncertainty that could change the action, then choose a check that answers it.
Gather more evidence when an uncertainty could change the next action, especially where the possible customer consequence is serious or the proposed change is costly. Name the decision first, then identify the missing fact that would make the team act differently.
Separate the finding from the conclusion
A finding should state what was reported or observed, in which task and under which conditions. For example, “Some participants could not locate the next step in a prototype session” would be a bounded observation, not a finding from a test reported here. It would not establish that all customers are blocked or that a particular design element caused the difficulty.
Record the proposed conclusion beside the finding. The team may want to revise navigation, change a service rule or leave the route as it is. Identify the assumption connecting the observation to that action. If the action depends on an unchecked cause, that cause is the evidence question.
Decide whether uncertainty matters now
- Possible serious consequence, weak evidence:take appropriate case or protective action, then investigate cause and wider exposure promptly.
- Consequential change, uncertain remedy:check whether the proposed change is likely to improve the customer’s result before committing heavily.
- Small, reversible change, clear observation:consider a limited change with an outcome check.
- Low-consequence finding unlikely to change a decision:record it and define a trigger for review.
These are decision prompts, not a substitute for the organisation’s specialist escalation process. A finding may justify attention before its population frequency is known, while claims about its scale remain bounded.
Match the check to the missing fact
Choose a check that can distinguish competing explanations. If customers report no confirmation, an authorised case record may show whether the request processed and whether a message was sent; it may not establish that the message arrived or was understood.
If a prototype control appears hard to use, observe relevant people attempting the task and note any assistance. If the decision depends on prevalence, define the eligible population and use suitable records or a sampling approach that supports that inference.
An open feedback box can reveal a problem, but its responses do not provide a dependable rate for all customers. More comments from the same route do not necessarily repair that coverage gap. More interviews may deepen an explanation without estimating prevalence.
Before collecting more, state the question, evidence source, result that would favour each available action and point at which the decision will return. The aim is a better decision, not complete certainty.
Evidence Sources and Their Limitations
- Open feedback box
- Reveals issues but does not show frequency across all customers
- Prototype usability test
- Shows task difficulty but may not reflect real-world use
- Case records (e.g., ATO or ABS data)
- Confirms processing success but not message delivery or understanding
- Sample survey
- Supports prevalence estimates when properly designed
Report what the check supports
Keep the original finding, new evidence and contrary cases together. Say whether the check established an event, narrowed a cause or estimated a frequency; these are different claims. If evidence remains mixed, decide whether a limited, reversible response is appropriate or whether the possible consequence calls for stronger investigation.
A decision note could say: “Check relevant cases to determine whether the problem is in the message or the processing step.” It gives the owner a specific next action without claiming that either cause has been established.
