Whitepapers
Papers, written for the person who has to approve this.
Consent architecture, the measurement method, a buyer's guide you are welcome to use against us, and why capture visibility belongs in hardware rather than in a policy document.
Most are gated. We ask for a few details, a person sends the paper, and one follow-up happens. There is no drip sequence.
Library
Abstracts in full, so you know before you ask.
The consent architecture for in-store conversation capture in the United States
A working design for obtaining and evidencing consent from both parties to a captured retail conversation across US jurisdictions. The paper sets out the four-layer customer consent stack — entrance signage, counter signage, a QR notice in plain language, and a scripted verbal disclosure inside the greeting — and explains why the verbal layer, the one retailers most often try to omit, is the layer a reviewer trusts most. It then treats associate consent as a separate problem inside the employment relationship, distinguishing acknowledgement at hire from the ongoing visibility that makes acknowledgement meaningful. A section on zone design covers default exclusions: fitting-room interiors, optical exam rooms, finance and insurance offices, staff areas. A final section addresses the states that require all-party consent, why several of them are large retail markets, and why we advise against opening a first deployment in the most demanding jurisdiction available. Includes the signed zone map template and the signage specification. Not legal advice; every deployment follows the retailer’s own counsel.
The matched-cohort method for proving coaching lift
Store-versus-store comparison is the standard way retailers evaluate coaching programmes and it cannot produce a result worth acting on, because the variance between two doors exceeds the effect being measured and pilot stores are almost never selected at random. This paper documents the alternative we run: cohorts inside a single store, where coached and capture-only associates share footfall, catchment, manager, inventory and promotional calendar, with a group swap at the midpoint to test whether the effect follows the coaching or the people. It specifies the preconditions that make the readout auditable — point-of-sale access at ticket level, cohort assignment fixed and written before counted measurement begins, and capture coverage published weekly so a thin week cannot be excluded after the fact. It also states the method’s limits: small cohorts, seasonal contamination, and what we do when coverage in a store degrades. Includes the readout template used at the day-90 decision review.
A buyer’s guide to conversation intelligence for retail leaders
Written to be used against us as much as by us. The paper separates the category into contact-centre software being repositioned for stores, camera-and-footfall analytics being described as conversation data, and systems that actually score in-store speech, and explains why the distinction changes both the legal review and the achievable coverage. It provides the questions that expose a weak product: whether any voice embedding is created at any stage including model training, whether capture can run without a visible indicator, whether the associate sees what the manager sees, what the dispute path is when a score is wrong, how coverage is reported, and what the vendor requires in order to prove a revenue result. It covers the two budget lines these programmes are typically funded from, the co-funding routes available in watches, beauty and automotive, and the failure modes we see most often — no point-of-sale access, a taxonomy imported from another market, and a manager dashboard shipped before an associate view. Includes a scoring sheet for vendor evaluation.
Hardware and consent: why visibility is a feature
A policy that says capture is always disclosed is worth less than a device that cannot capture without disclosing. This paper describes the capture device as a compliance instrument: an always-visible indicator wired so that capture cannot run with it unlit, cryptographically signed firmware that causes a device to refuse an unsigned image, no camera and no image sensor of any kind, and no on-device speaker enrolment. It explains why the no-covert-mode guarantee is placed in hardware rather than in configuration — a software setting requires a reviewer to trust the vendor’s change management and every future release — and what that costs us commercially when a buyer asks for an evaluation mode with the indicator off. Further sections cover on-device redaction of payment and personal identifiers before transmission, encryption in transit and at rest, key custody, the audit log of every access including Gaincraft staff access, and the physical estate questions procurement asks: mounting, power, network posture on a store LAN, and replacement logistics.