Definition
A feedback taxonomy is the structure a brand uses to categorize unstructured feedback. It names the themes, groups them into categories, and attaches tags and sentiment at the record level. It can be flat — a single list of 40 themes — or hierarchical, with categories like "Product Quality" containing themes like "enclosure cracking" and "button sticking." It can be auto-generated from clustering on the corpus, human-owned and edited, or a hybrid where the machine proposes and a human edits.
Taxonomies also have scope. A global taxonomy applies across the full catalog; a per-SKU taxonomy lets a blender have different themes than a toaster. Most consumer brands need both layers: a global top level for aggregate reporting and SKU-level branches for detail.
Why it matters
The most common failure mode in VoC is taxonomy drift between teams. Product labels the same complaint one way in their roadmap tool, CX labels it differently in Zendesk, and QA uses a third vocabulary in the manufacturing tracker. When leadership asks, "What are the top product issues?", three different answers arrive and the meeting turns into a vocabulary fight.
A shared taxonomy, applied consistently across every feedback record regardless of channel, ends that fight. It also makes trends meaningful. Week-over-week theme movement only has meaning if the themes are stable. A taxonomy that changes every run is useful for exploration; it is useless for decisions. The taxonomy is also the artifact that survives staff turnover — a Consumer Insights lead who leaves takes the institutional knowledge with them; a locked, documented taxonomy stays.
Example
A home appliance brand runs six months of Amazon, Walmart, Best Buy, and Bazaarvoice reviews plus Zendesk tickets through Indellia's Theme Agent. The agent proposes a three-level hierarchy: Category ("Product Quality," "Fit & Finish," "Onboarding," "Value"), Theme under each, and SKU-specific sub-themes beneath. The Insights lead edits the draft, merges two near-duplicate themes, renames a category, and locks the taxonomy. That taxonomy is now the shared vocabulary: Product reads it in their roadmap reviews, CX tags new tickets against it, QA references the same theme names in factory escalations. When leadership asks for the top three issues, the answer is the same for all three teams because it is drawn from the same labels on the same records. A quarter later, two new themes appear organically from Theme Agent's drift detection; the Insights lead approves them into the taxonomy, and the history rebuilds so week-over-week trends remain comparable.