Voice-of-Customer Feedback Automation: A Blocked Response Is Never Counted as a Zero

A consulting engagement often collects real customer feedback — a round of interviews, an NPS survey sent after a launch, a handful of satisfaction scores — and then lets it sit in a separate document, disconnected from the branding, journey and offer decisions the feedback should actually be informing. Nobody deliberately ignores the feedback; it simply has nowhere structured to go. A theme that shows up across five interviews stays a vague shared impression among the people who read the transcripts, rather than becoming a named, trackable item that a specific part of the strategy has to respond to. And when an NPS number does get calculated, it is rarely clear afterward exactly which responses fed it and whether every one of those respondents had actually agreed to be counted.

Collecting feedback and acting on it are usually treated as two separate jobs done by different people at different times — a researcher runs the interviews, a strategist writes the positioning, and nothing structurally forces the second person to revisit the first person's findings. Consent is easy to forget under time pressure: a quick phone call gets logged the same way as a properly consented survey response unless the system itself asks the question and separately gates the calculation on the answer. And converting a raw quote or a repeated theme into a specific, actionable, trackable item takes a deliberate extra step — writing it down as a named finding, deciding which specific part of the strategy it affects — that a busy team skips unless the tool itself creates a place for exactly that step.

Every voice-of-customer effort in a project starts as a study — one of five types (interview, survey, NPS, satisfaction/CSAT, or free-form/qualitative), each with its own free-text objective, audience, question set and consent text, and one of four response scales (NPS 0-10, CSAT 1-5, CES 1-7, or none for purely qualitative work). Individual responses are recorded against a required pseudonymous "respondent alias" — the input validation itself instructs "provide a pseudonym, do not use personal data" — along with the collection source (manual entry, interview, form, phone, email or other), an optional numeric score, open text, notes, and an explicit consent_given boolean captured per response, not once per study. The aggregation logic enforces that consent boundary directly: the summary view filters every study's responses to only those with consent_given before computing anything, and reports the excluded count as a distinct "blocked" figure sitting next to the real total — an NPS calculation counts only consented promoters (score 9-10), passives (7-8) and detractors (0-6) using the standard (promoters minus detractors, divided by total) formula, and a CSAT calculation counts only consented scores toward the mean and the percentage scoring 4 or higher. A response withheld from consent is never treated as a zero and never silently vanishes — it is counted, visibly, as blocked. Separately, an insight is a named finding — a theme, one of nine fixed kinds (pain, desire, objection, reason to buy, reason lost, reason to renew, reason to cancel, real customer language, or general theme), a description, supporting evidence, and a target area choosing exactly which part of the engagement it should inform: branding, journey, offer, data, plan, reports, portal, or general. An insight optionally links back to the specific study it came from, and moves through an explicit four-stage status — proposed, approved, applied, or discarded — so a raw idea, a vetted finding and an already-acted-on change are never presented as the same thing. Every study, response and insight write is checked against the acting user's actual organization membership before it is allowed, so one organization's customer feedback is never visible to or writable by another.

A consultant sets up an NPS study for a client three months after a product relaunch, with the scale set to NPS 0-10 and a consent text describing exactly how responses will be used internally. Fourteen responses come in through a mix of phone calls and a shared form; twelve include explicit consent, two do not — one respondent answered quickly without being read the consent text, the other declined when asked directly.

The summary reports an NPS of 25 computed strictly from the twelve consented responses, with "14 respondents (2 blocked)" shown alongside it — the two non-consented answers are visible as excluded, not silently missing and not dragging the score down as if they were automatic detractors. Reading through the twelve open-text responses, the consultant notices three separate promoters independently mention slow order tracking as a minor gripe despite their high scores. That pattern becomes a new insight: kind "general theme," target area "journey," confidence "medium," with the three response quotes attached as evidence. The insight sits as "proposed" until the strategist reviewing the engagement approves it and later marks it "applied" once the journey map is actually updated to address the tracking gap.

A reported NPS or CSAT number becomes defensible after the fact, because exactly which responses fed it — and exactly how many were excluded and why — is visible rather than assumed. A theme repeated across several customers stops being a private impression among whoever read the transcripts and becomes a named, evidenced, trackable item pointed at one specific part of the engagement. And a respondent's feedback is captured under a pseudonym by construction, not as an afterthought someone has to remember to redact later.

A team that needs to actually SEND a survey to customers and collect responses through a public, self-serve link will not find that here — every voice-of-customer route lives inside the authenticated project workspace, and every response is entered by the consultant from feedback gathered through some other channel (a call, an interview, a third-party form). The value here is in structuring, consenting and routing feedback that already exists, not in distributing the survey itself.

A spreadsheet of interview notes and survey scores can hold the same raw information, but nothing forces a distinction between a consented and a non-consented response, nothing enforces a consistent NPS or CSAT formula across studies, and nothing turns a repeated theme into a trackable item pointed at a specific part of the strategy. Centriu Oracle's structure adds exactly those three things without adding a data-collection channel that was not there before.