Anti-counterfeit + loyalty on one QR: two problems, one code

Counterfeiting in India is not a dark-web crime; it is a wholesale-market business. Fake house wire with under-gauge copper sells through the same Bhagirath Palace lanes as the genuine article, at 15–30% below list, with packaging good enough to fool a distributor. Engine oil drums are refilled with base oil and resealed. Agrochemical fakes — often inert powder in perfect pouches — cluster in exactly the districts where a brand's genuine sales are strongest, because that is where the brand name has pull. Practitioner estimates across these categories put counterfeit penetration at 5–20% of branded volume in affected geographies, which is simultaneously lost revenue, warranty and liability exposure, and — for wires and agrochemicals — a genuine safety hazard attached to your logo.

The traditional counters — holograms, colour-shift inks, tamper seals — share one fatal weakness: they are verified by nobody. A hologram works only if someone looks at it, knows what the genuine one looks like, and cares. Counterfeiters copy holograms well enough to pass a glance, and the person best positioned to check — the retailer or mechanic handling the unit — has no reason to bother. Detection therefore happens years late, through a field-team accident or a customer complaint, long after the fake network is entrenched.

Serialisation gives every unit a unique identity: a QR encoding a one-time code, generated by the brand, printed on the label, mapped in a database to SKU, batch, plant and dispatch. A counterfeiter now faces an unpleasant arithmetic. He has three options, and every one of them is detectable:

• Invent codes. Random codes fail validation instantly — the series was never issued. Every failed scan is a flare with a GPS coordinate attached.
• Copy one real code onto many packs. The first scan claims it; every subsequent scan of the same code — in Kanpur, then Indore, then Guwahati — is a duplicate-scan event. One genuine coil cannot be sold four hundred times.
• Buy genuine product to harvest codes at 1:1. At this point counterfeiting stops being profitable, which is the actual goal.

But serialisation alone has the hologram problem in digital form: if nobody scans, nothing is detected. Verification-only QR programs — "scan to check authenticity" with no reward — see scan rates of a fraction of one percent. The database is perfect and blind.

Attach money to the scan and everything changes. When a coil scan pays ₹10–15 and a lubricant-carton scan pays ₹20–40 through an influencer loyalty program, scan coverage jumps from under 1% to 40–80% of secondary volume within quarters. India has an estimated million-plus electricians, and lakhs each of mechanics, painters, plumbers and agro-counter staff — the reward converts this population into a continuously reporting, self-financing inspection network with better geographic coverage than any enforcement team a brand could hire.

The member does not even need to know he is inspecting. He scans because he is paid; the platform does the detecting:

Serialised codes catch cloned and invented codes. The remaining gap is the sophisticated counterfeiter who prints plausible unique codes, or the refill operator reusing a genuine container. This is where computer vision closes the loop: a single phone photo of the label can classify original vs duplicate, with no holograms, special inks or packaging changes.

The principle: every print process leaves a fingerprint. The genuine label is produced on known presses with known plates, inks and substrates. Its micro-texture, dot structure, colour profile, font geometry and print registration are consistent in ways a counterfeit run — different press, different plates, redrawn artwork — cannot exactly reproduce. Individually these differences are invisible to the eye; to a vision model trained on genuine samples (feature keypoints, frequency-domain texture, colour statistics, geometric layout), they are a stable signature. The scan flow simply adds a camera step: scan the QR, photograph the label when prompted, get a verdict in seconds. Field teams use the same check during market visits, and high-risk verdicts queue for expert review rather than auto-accusing anyone.

The practical beauty is deployment: because the fingerprint lives in the genuine artwork itself, photo authentication works on stock already in the market — no packaging changeover, no new security feature for the counterfeiter to study, and quiet model updates whenever a new fake variant appears. Combined with serialisation, the counterfeiter must now beat both the database (unique valid codes) and the physics of his own printing press. Few bother; they move to an unprotected brand — which is, bluntly, the goal of brand protection.

• Wires and cables. Fakes are under-gauge copper — a fire risk sold under your name. The electrician scanning for ₹10–15 a coil is exactly the person unrolling the wire; duplicate and invalid alerts concentrate on the wholesale hubs feeding fakes into tier-2 markets. Inner-flange or under-label QR placement stops shelf harvesting.
• Lubricants. The refill problem: genuine drums refilled with base oil. Batch-timeline anomalies (year-old codes scanning fresh) plus photo checks on cap seals catch it. Mechanics scanning at ₹20–40 per carton give near-total coverage of the workshop channel.
• Agrochemicals. The highest human stakes — fake pesticide destroys a farmer's season and the brand's district reputation for a decade. Agro-dealer counter staff scan at the till; invalid-series clusters map fake distribution with tehsil precision, and seasonal spikes (kharif sowing, June–July) tell enforcement exactly when to act.
• Cosmetics and personal care. Salon and beauty-advisor programs double as authentication at the point of application; fakes here are a skin-safety liability, and counters report willingly once reporting pays.

Worked example. A lubricant brand does ₹300 crore of secondary revenue and estimates 8% counterfeit penetration — ₹24 crore of displaced sales, before warranty claims and reputational damage. It already intends to run a mechanic loyalty program at 1.5% of secondary revenue: ₹4.5 crore a year in rewards, platform and operations, justified entirely by share gain and secondary visibility. Serialising the packs adds ₹0.10–0.30 per unit — say ₹60 lakh on 3 crore units — and the anti-counterfeit capability rides on infrastructure the loyalty case already paid for. If detection and enforcement recover even a quarter of the displaced volume, that is ₹6 crore of recovered revenue against ₹60 lakh of incremental cost — a 10x return on top of the loyalty program's own ROI. Compare that with standalone brand-protection projects (holograms, mystery shopping, investigators) that cost ₹1–3 crore a year and see nothing between field visits.

The reverse framing is just as useful when selling the budget internally: the CFO who hesitates at a loyalty program's cost often approves it the day counterfeit recovery enters the business case. Two problems, one code — and one line item. Model the combined cost in the QR cost estimator.

Compliance note. All of it — scan rewards, informer bonuses, gifts at mechanic meets — aggregates per PAN for Section 194R: once a member's cumulative benefits cross ₹20,000 in a financial year, 10% TDS applies at payout. One platform ledger across loyalty and reporting rewards keeps this automatic.

Anti-gaming note. The same fraud engine protects both faces of the program: geo-fencing and velocity caps against dealer bulk-scanning, device fingerprinting against harvesting rings, name-match on payout handles against mules, and human review before any counterfeit accusation reaches a channel partner — a false fake-flag against a genuine distributor is a relationship you do not get back.