What does a Voice of Customer FTE own?
A Voice of Customer FTE on ftes.ai owns the signal in what customers write, in the Product Quality department of the e-commerce org chart. It is judged on defects found that were then fixed, and accountable for product defects being found in what customers say and then fixed. It exists because a seam that fails, a colour that differs from the photo, or a listing that promises what the product does not do are all described by customers weeks before they appear as a return rate, in sources nobody reads together.
It does not own the reply, the review request, or the product. It owns the reading, the clustering and the handover.
Where does it read, and what does it look for?
The Voice of Customer FTE reads product reviews from the store and the marketplaces, support tickets in the help desk, the free-text comments customers leave in the returns portal, and marketplace questions and answers. It looks for statements about the product rather than about the service: what broke, what did not match, what was missing, what did not fit as described. Service complaints, about delivery or a refund, are routed to the Order Support and Logistics Officer FTEs rather than clustered as defects.
Where a review or ticket carries a photo, the photo is attached to the cluster, because a vendor conversation with photos ends differently from one without.
How are clusters built and attributed?
A cluster is a set of mentions that describe the same defect on the same product: 'the handle came off' on one mug, across reviews, tickets and return comments. The Voice of Customer FTE attributes the cluster to the variant, and where the store's receipts allow it, to the inbound batch, by matching order dates to the receipt the units shipped from. A defect that appears only on units from one receipt is a batch problem for the Supplier Officer FTE; one across all batches is a design or listing problem.
Listing mismatches are their own kind of cluster: customers describing a product that is not what the page says. Those go to the Product Catalog FTE with the page and the mentions side by side.
Where does each cluster go, and what needs a person?
Each cluster is routed as an evidence pack, the mentions, the photos, the variant and batch, the order numbers, to the role that can act: the Supplier Officer FTE for a vendor or batch problem, the Product Catalog FTE for a listing mismatch, the Fit Officer FTE for a fit theme, the Buyer FTE for a product to stop reordering. The receiving role proposes the fix for the owner's approval; the Voice of Customer FTE tracks whether the defect stops appearing.
Anything touching safety, a product that overheated, a part a child could swallow, goes to a named person the same day, never into a routing queue.
What does the Voice of Customer FTE never do?
It never replies to a review or a ticket, never solicits reviews, and never contacts a vendor; the Social Officer FTE, the Order Support FTE and the Supplier Officer FTE do those on approval. It never reports a sentiment score as a result, because a sentiment score is not a defect fixed. It never clusters service complaints as product defects, and it never presents a single mention as a finding.
It also never edits the sources it reads. Reviews and tickets are read-only to it, and its evidence packs quote them verbatim.
What does it report, and how honestly?
The Voice of Customer FTE reports defects found, clusters routed and to whom, fixes proposed by the receiving roles and approved by the owner, and defects confirmed fixed, meaning the mentions stopped in later reviews and returns over a window the owner set. It reports the sources read and the volume, so a quiet month is visible as a quiet month rather than as a silence.
Where a defect stopped appearing because the product was discontinued rather than fixed, the report says so.
How does it work with the other roles, and in which tools?
The Voice of Customer FTE feeds the Supplier Officer, Product Catalog, Fit Officer and Buyer FTEs and reads receipts from the Warehouse Officer FTE for batch attribution. It reads Yotpo or Judge.me reviews, Gorgias or Zendesk tickets, Loop Returns comments, and Amazon reviews and questions, with orders from Shopify, WooCommerce or BigCommerce.
It earns its place at any store with enough reviews and tickets that nobody reads them all, which is most stores past their first year.