Every question a shopper asks is a datapoint your POS can't see
POS records what was bought. It says nothing about what shoppers asked for, couldn't find, or rejected — that demand evaporates on the floor. ShopVoices turns it into three dashboards for three different readers — store operations, head-office strategy and brand teams — without a single recording or transcript ever leaving the building.
Shelf verification: the stock your systems think is on the shelf — and isn't
POS tells you what sold. ShopVoices tells you what shoppers asked for and didn't find — and ranks it into a Shelf verification queue for every store. When a shopper asks for a product the inventory record says is in stock and walks away without it, that record is the prime suspect: phantom stock, a mislocated item, or a shelf that needed filling hours ago.
The industry context makes this the most valuable list in the building: the landmark academic audit of a US retail chain found 65% of inventory records inaccurate, and a 2025 study of ~24,000 SKUs across 11 grocery stores found that systematic shelf auditing produced an 11% store-wide sales lift — concentrated exactly where the system showed more stock than existed.
Sources: DeHoratius & Raman inventory-record accuracy study; 2025 shelf-audit study (arXiv 2506.05357). A verification queue, not an automatic restock instruction — your team confirms at the shelf.
Built for the person reading it, not the database behind it
Store Operational
For the store manager: intent mix, unanswered-question rate, peak interaction times, per-device breakdowns, satisfaction from on-screen 👍/👎 — and the ranked Shelf verification queue built from what shoppers asked for and didn't find.
Head Office Strategic
For the chain: category demand trends, store-vs-store benchmarking with click-through, a Customer Friction Index, and attribute demand — what shoppers ask about, not just what they ask for.
Brand-Level
The retail-media story: brand demand share, brand-vs-competitor consideration, product-level demand, stock-friction by brand, and price-perception signals — spoken in-store demand, collected without retaining audio or transcripts. Product-level depth is a retailer-controlled configuration, enabled where the applicable legal basis is in place.
Every number opens up — until you're looking at one device on one day
Seven drill-down chains run through the dashboards with breadcrumbs the whole way: category → store → device, brand → product, trend → day. Every figure carries a prior-period delta, and rule-based 💡 insight cards surface the findings worth acting on before anyone goes looking. Regions roll stores up for the managers in between.
- Multi-level drill-downs with breadcrumbs — no dead-end charts.
- Prior-period ▲/▼ deltas on every figure.
- Automatic insight cards that flag what changed and why it matters.
- Interactive at millions of records per store.
Demand data without the data-protection bill
The analytics pipeline ships an allowlist: timings, counts, category tags, a language code, a thumbs-up or -down. Audio never leaves the building, and neither does the transcript — the record structure makes it impossible, not merely forbidden. Richer catalogue-level tags are a retailer-controlled option where the legal basis exists. Every surface feeds the same pipeline, so kiosk, counter, handheld and mobile demand all land in one place.
From a spoken question to a merchandising decision
A shopper asks for a product that isn't on the shelf. The question lands in the store's Shelf verification queue within the hour, shows up in the category trend at head office, and moves a brand's stock-friction figure — every step visible in the same tool.
Measure your floor's spoken demand
A pilot ships with the full analytics suite from day one — your stores, your categories, your questions.