SealTrust — Product Verification, Sealing & Evidence Platform

Wrong-item deliveries, return swaps, internal shrinkage and false claims all share one root cause: at the moment of dispute, there is no independent, tamper-evident record of what was actually in the box and who touched it. Refund decisions therefore default to the customer's word, and the loss is absorbed.

SealTrust closes that gap with a two-sided control: a verification gate on the way out (scan, weigh, photograph, seal) and an evidence-matched adjudication gate on the way back (seal, serial, mass and photo comparison). Both sides write to an append-only, hash-chained audit ledger that no operator, agent or administrator can alter.

The commercial case is deliberately simple: a per-parcel fee, netted monthly against a finance-ready loss-avoidance ledger. The pilot proves net-positive before any scale-up spend is committed.

Wrong product deliveries generate refunds, replacements, courier costs and negative reviews. Return fraud lets high-value items be swapped for low-value substitutes. Theft and false claims — internal and external — are indistinguishable without custody evidence. The cumulative losses escalate with volume, and the reputational cost compounds: a customer who receives a swapped or short parcel rarely gives a third chance.

Each of these is addressed by a specific platform module rather than a policy statement. The mapping is shown on the platform's landing page and drives what the consoles actually do.

South African online retail turnover is forecast at R159bn in 2026, growing 22.5% and crossing 10% of total retail for the first time (World Wide Worx / Mastercard, September 2026). Volume growth of that order multiplies every unresolved loss category rather than diluting it.

Internationally, retailers report 19.3% of online sales returned against 15.8% of total sales, on roughly $850bn of returned merchandise (NRF & Happy Returns, October 2025). A material share is fraudulent or abusive, and it is indistinguishable from genuine claims without evidence captured at dispatch.

Locally, retail shrinkage is estimated at 1.6%–2.1% of turnover in 2026, up from 1.4%–1.8% through 2023–2024, and recorded supply-chain theft rose more than 50% year on year with hijacking the dominant method (BSI, June 2026). None of these figures improve on their own.

Verification Line is the operator console at the inbound lane: scan against the order, capture certified mass, photograph six frames, apply the void-pattern seal and issue the buyer certificate. Committing the record is a server-side action that mints the Seal ID, hashes the evidence and appends to the ledger.

Return Validation is the returns-agent console: the claim is placed side by side with the original verification record, and each evidence line is marked match or mismatch. The agent's approve or decline is written with its confidence score and evidence hashes.

Control Tower is the executive view: verified volume, loss avoided, seal integrity, decline rates, per-DC risk and a coverage model that projects fee against net benefit.

Audit Chain and Notifications are the assurance and communication surfaces: chain re-computation on demand, and a complete delivery history for every operational and buyer message.

High-risk units are diverted to a verification booth on the existing inbound lane. The operator keeps their current scanner; the booth adds a certified scale, a fixed camera rig and a seal printer. Average handling time at pilot measurement is under 40 seconds per unit for the categories in scope.

On a return, the claim arrives in the Returns console already matched to its verification record. The agent reviews four evidence lines and rules. Declines automatically produce a dispute-ready evidence pack; approvals route the recovery to the responsible party — courier, DC shift or supplier.

Six least-privilege roles are defined: Platform Admin, Operations Supervisor, Verification Operator, Returns Agent, Buyer/Customer and Read-only Auditor. Roles are stored in a dedicated table, never on the user profile, and are resolved by a security-definer function used inside row-level policies. An operator sees the verifications they committed; an agent sees the rulings they made; oversight roles see the network; a buyer sees only their own parcel and messages.

All meaningful activity is written to the audit ledger with timestamp, authenticated identity, role, action, parcel and seal ID, decision and evidence hashes, plus the previous record's hash. A database trigger rejects any update or delete on the ledger, for every role including administrators. Any party can re-compute the chain and see immediately whether — and exactly where — it was disturbed.

Pilot instrumentation reports mis-pick rate, seal integrity, claim volume versus evidence-backed declines, intercepted fraud value and net benefit after fee. Because each intercepted event is priced and posted to the loss-avoidance ledger, avoided loss is reconciled with Finance monthly rather than asserted annually.

Phase 1 (weeks 1–8): Kempton Park DC, high-risk categories only, running alongside existing workflows. Success measure is baseline versus verified claim rates across 60 000 parcels.

Phase 2 (months 3–9): all national DCs and cross-docks, API integration into WMS, returns and dispute tooling, automated sealing lines with per-unit costing, monthly ledger reconciliation with Finance.

Phase 3 (month 9+): customer-facing Verified Seal badge on product pages, a Marketplace seller programme, and public trust reporting — making Takealot the first major South African retailer with end-to-end product verification.

MRHS supplies booth hardware, consumables, trained operators and platform operations under an agreed SLA covering throughput, seal integrity and evidence-pack turnaround. Takealot retains all customer communication, refund authority and commercial policy; SealTrust supplies the evidence and the recommendation.

Commercially: R 4.80 per verified parcel, a platform subscription from R 210k per month covering the consoles, retention and dispute evidence packs, and enterprise pricing across the DC footprint, category risk tiers and Marketplace onboarding. The per-unit fee scales exactly with volume, so Takealot carries no fixed exposure.

Adoption: booths bolt onto existing lanes and operators keep their current scanners; Phase 1 changes no WMS master data. Cost: the per-unit fee is netted against the loss-avoidance ledger monthly, and the pilot must prove net-positive before scale-up spend. Scalability: seal printing, mass capture and photo capture are machine-driven, so throughput grows without proportional headcount.

We ask for lane access at Kempton Park, a nominated Takealot operations sponsor, read access to the relevant order and returns feeds, and a joint weekly review. At week eight we present the measured baseline comparison and the reconciled loss-avoidance ledger, and Takealot decides on Phase 2 with evidence in hand.