Ask who scans a product QR and most people picture a shopper with a phone. The real scanning population is broader and more consequential, and designing the code for all of it, rather than for the imagined shopper alone, is what separates a Digital Link strategy from a marketing QR.
The humans
Shoppers pre-purchase check provenance, allergens and authenticity, concentrated in categories where trust is the purchase: food origin, supplements, wine and spirits.
Owners post-purchase scan for instructions, registration, reorders and disposal guidance, the moments where the pack is the only interface the brand reliably owns.
Professionals scan too: retail staff checking recalls, inspectors verifying claims, recyclers reading passport data. Each expects the code to answer their question, which only works when resolution is audience-aware.
The machines
Point-of-sale imagers are the growth story: Sunrise 2027 makes the till a first-class scanner of the same symbol, extracting the GTIN for pricing while the URL waits for phones.
Warehouse and logistics systems read codes for receiving, inventory and traceability records, increasingly asking the URL for structured data rather than just parsing digits.
Software with no scanner at all, marketplace bots, compliance crawlers and AI shopping agents, consumes the URLs directly. A code that resolves to JSON-LD and a linkset makes the product legible to that entire population; a marketing page makes it invisible. This is the machine-audience argument in the comparison.
Designing for the whole population
The design consequence is single: never print a code that serves only one audience. A Digital Link behind a conformant resolver gives each scanner its native answer from one symbol, and the analytics then report the populations separately, humans as engagement, machines as operations, which is how you finally learn who was scanning all along.
Common questions
Who actually scans product QR codes the most?
It varies by category, but the machine population, tills, warehouse scanners, systems, grows steadily, and consumer scanning concentrates on moments: provenance checks, recalls, instructions, reorders.
Do AI agents really scan product codes?
They increasingly consume the URLs: shopping assistants and crawlers fetch product data programmatically, and a code resolving to structured data is how a product stays legible to them.
Why distinguish humans from machines in analytics?
Because they mean different things: human scans are engagement, machine scans are supply-chain and platform activity. Mixing them corrupts both readings.
Can I see who scans my products today?
Not who as identity, but which populations: Onpack's analytics split humans from machines per product and country, live, with no tracking of individuals.