Back to all articlesCustomer feedback

Diagnose missing feedback before treating silence as satisfaction

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A manager receives a small set of positive survey responses and concludes that customers have few concerns. But most customers did not answer. Some may never have received the request; others may have abandoned a difficult form, doubted that feedback would matter, or simply chosen not to participate.

Interpret customer feedback nonresponse as missing information, not as a satisfaction score. First establish who had a realistic opportunity to respond. Then inspect collection friction, compare the known characteristics of respondents and nonrespondents, and run one limited test of a plausible obstacle. None of these steps reveals what an individual silent customer thinks, and no small test eliminates response bias.

The worksheet below turns that distinction into a practical management decision: what can be reported now, what remains unknown, and what should be tested next.

1. Separate the feedback finding from the collection finding

A feedback exercise produces two different kinds of evidence. The answers describe what respondents chose to report. The collection record describes how successfully the organisation created opportunities to answer. Combining the two carelessly turns an operational gap into an unsupported customer conclusion.

A defensible summary might say: Respondents mostly described a straightforward experience, but new customers were less represented among completed responses. We have not established whether that difference reflects collection friction, different experiences, or different willingness to participate.

That is more useful than either “customers are happy” or “the survey is useless.” The responses can still identify real issues and useful experiences. Their limits affect how broadly the findings should be applied.

Keep three questions separate:

  • Coverage: Which intended customers had a chance to receive the request?
  • Participation: Which invited customers started and completed it?
  • Experience: What did those who answered report?

Do not fill gaps in the third question with assumptions drawn from the first two.

2. Reconstruct the opportunity to respond

Start with a defined customer population and a fixed collection period. For example, an organisation might want feedback from customers whose onboarding ended during a particular period. That definition should exist independently of whether a usable contact record is available.

Then trace distinct collection states. A customer can be eligible but excluded from contact, selected but never sent an invitation, sent an invitation without a confirmed delivery outcome, or recorded as delivered without ever reading it. A form start is not a completed response.

Use only states your records actually support. If delivery information is unavailable, label it unknown. Do not substitute a prepared invitation, an approved message, or a successful submission to a sending system for evidence of delivery.

Also check what each count represents. Customers, invitations, form visits, and submissions are different units. Repeat invitations can inflate apparent coverage; repeated form visits can inflate apparent participation. If you cannot connect stages reliably, retain separate counts and explain that limitation rather than inventing a precise funnel.

The aim is not a perfect tracking system. It is to stop saying “customers declined to answer” when the evidence only says “we do not have an answer.”

3. Filled hypothetical nonresponse diagnostic worksheet

Fictional worked example: All organisations, counts, percentages, observations, and test choices in this worksheet are hypothetical. They illustrate a diagnostic method, not Alactic customer results.

Scope and invitation coverage

  • Decision to inform: Whether the current onboarding process needs further investigation before the next revision.
  • Population: 240 customers whose onboarding ended during a defined four-week period.
  • Contact eligibility: 210 had a usable contact route and recorded permission for the intended channel. The remaining 30 were excluded because contact details or permission were missing.
  • Invitation coverage: 180 unique customers were invited. Another 30 contact-eligible customers were omitted because their onboarding completion status was entered after the invitation list was generated.
  • Recorded receipt: 150 invitations had a delivery record; receipt remained unknown for 30. Delivery was not treated as proof of reading.
  • Coverage conclusion: Only 180 of 240 intended customers were invited. The uninvited 60 cannot be described as people who ignored a request.

Completion friction

  • Participation: 60 invited customers started the form; 36 completed it and 24 left it unfinished. All recorded starts came from customers with a delivery record.
  • No recorded start: 90 customers with a delivery record and 30 with unknown receipt had no recorded start.
  • Possible obstacle: The form required an order identifier before customers could submit feedback.
  • Review observation: A fictional internal walkthrough found that the invitation did not tell customers where to find that identifier. This establishes a plausible burden, not the reason for every abandonment.
  • Unknown: The records do not show why customers stopped, whether nonstarters opened the invitation, or whether anyone attempted to submit and encountered an unrecorded error.

Missing customer groups

  • Established customers: 120 invited; 30 completed, giving a hypothetical invitation-based completion proportion of 25%.
  • New customers: 60 invited; six completed, giving a hypothetical invitation-based completion proportion of 10%.
  • Composition: New customers represented one-third of invitees but one-sixth of completed responses.
  • Boundary: This comparison covers invitees only. It does not establish the composition or experiences of all 240 intended customers.

Competing explanations and bounded test

  • Collection explanation: Finding an order identifier may make completion harder, especially for customers unfamiliar with account records.
  • Alternative explanations: New customers may have less time, different channel habits, less confidence that feedback will matter, or different experiences. The collection period may also have reached them at an inconvenient point.
  • Test question: Does removing the mandatory order-identifier field produce a useful completion path for previously silent invitees?
  • Test group: Randomly select 24 customers from the 90 with recorded delivery but no recorded start, after rechecking permission and opt-outs.
  • Comparison: Randomly assign 12 to the original form and 12 to the same form with the order identifier made optional. Keep the feedback questions and invitation wording unchanged.
  • Contact boundary: One follow-up invitation per selected customer, a seven-day observation period, and no further reminder within this test.
  • Owner and record: The feedback lead records invitations, known delivery outcomes, starts, completions, identifier-related questions, and complaints or opt-outs by group.
  • Decision rule: Review whether either route exposes a reproducible obstacle and whether the optional-field version permits useful responses without unacceptable follow-up burden. Treat any completion difference as exploratory, not proof of a population-wide effect.

4. Compare groups without guessing their satisfaction

The worksheet shows a participation gap between new and established customers. It does not show a satisfaction gap. New customers might be more dissatisfied, more satisfied, less engaged with the channel, or simply less able to find the required identifier.

Compare characteristics already recorded for a legitimate operational purpose, such as customer tenure, service type, or onboarding route. Check that the definitions are consistent and that the records existed before the feedback request. Avoid building speculative profiles of silent individuals.

Use the invited population as the denominator when comparing participation among invitees. Use the intended population when assessing coverage. Those comparisons answer different questions. A group can have a reasonable completion proportion among invitees while still being largely absent because few of its members were invited.

Prefer a few decision-relevant comparisons over many tiny segments. Small groups can show large percentage changes from very few responses. Report the counts alongside proportions, and do not rank teams or customer groups on unstable differences.

Finally, matching respondents and nonrespondents on known characteristics does not establish that their experiences match. Customers with similar account histories can have very different reasons for answering—or remaining silent.

5. Inspect friction before spending another customer contact

A follow-up is not always the first useful action. Walk through the collection route using ordinary customer conditions: the invitation text, the destination page, the required fields, error messages, and the final submission step.

Check whether the request explains what feedback is wanted, whether participation is optional, and what happens after submission. Look for unnecessary account lookups, unclear labels, required fields unrelated to the feedback question, and errors that leave people unsure how to continue.

For forms, the W3C guidance on labeling controls explains that labels should identify each control’s purpose and be programmatically associated with it. A visual label alone is not enough to establish that the form works for everyone.

A walkthrough can reveal a defect, but it cannot explain all missing responses. Fixing an obvious broken submission route may be justified without testing whether customers prefer a broken route. Record the defect and its correction, then treat feedback collected before and after the change as having different collection conditions.

Where no defect is found, retain competing explanations. “We could not reproduce a problem” is not equivalent to “there was no problem.”

6. Keep the follow-up narrow and respectful

The hypothetical test changes one burden: whether the order identifier is mandatory. Shortening the form, switching channels, changing the wording, and adding an incentive at the same time might increase participation, but would make the cause harder to interpret.

The optional-field design has a trade-off. It may reduce effort while making it harder to connect a response to a particular transaction. Before testing, decide whether that connection is genuinely necessary. Do not remove a field needed for a stated purpose without considering how the resulting feedback will be handled.

A follow-up should not imply that silence is suspicious or that an answer is owed. Give customers a straightforward way to decline, honour opt-outs, and avoid repeatedly pursuing people because their views are statistically desirable.

If using WhatsApp, check recipient permission and the current WhatsApp Business Messaging Policy. Approved templates are used to initiate Platform conversations and outside the 24-hour customer-service window, which opens or resets with a user message. Approval does not guarantee delivery. If automation is involved, provide a clear human escalation path.

The test plan is an operating recommendation, not evidence that a particular workspace can execute it. Confirm available collection and reporting functions before committing to the design.

7. Decide what different outcomes would actually mean

Set interpretation boundaries before reviewing results. Otherwise, almost any outcome can be made to support the manager’s original belief.

  • More completions on the optional-field route: The pattern is consistent with identifier friction in this small follow-up. It does not establish why earlier nonrespondents stayed silent or prove that the new responses represent everyone.
  • Little participation in either group: The test may have been too small, poorly timed, or aimed at the wrong obstacle. Low participation is not confirmation of satisfaction.
  • More starts but few completions: Inspect the remaining steps and submission process before sending more invitations.
  • New concerns appear: Investigate their substance. Do not estimate their prevalence across all customers from a small, selected follow-up.
  • Complaints or opt-outs occur: Honour the requests, pause further contact where appropriate, and review whether the follow-up burden is justified.

If a sending or form problem disrupts one group, do not quietly treat its lower participation as a customer preference. Mark the comparison as compromised, preserve the record, and decide whether a corrected test merits another contact. Restarting is not automatically worth the additional burden.

8. Make the next management update auditable

Before the next feedback review, assign one owner to complete the worksheet with a fixed observation period and explicit unknowns. The immediate deliverable is not a higher response count. It is a defensible account of who was included, who was missing, and what the next test can answer.

Report coverage and completion separately, show the counts behind group comparisons, and keep respondent findings distinct from population claims. End the update with one bounded decision: fix a demonstrated obstacle, run the specified test, or accept that the available evidence remains limited.

The useful shift is from “we heard no concerns” to “these customers reported these experiences, these customers are missing, and this is how we will investigate one plausible collection gap.”

Customer feedbackNonresponseSurvey designCustomer operations

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications