Reading QR analytics: scans, sessions and what happens next

A scan is a redirect, with the same caveats

Scanning a dynamic code produces a request to your short link, so everything true of click analytics is true here: bots and preview fetchers can appear, repeat scans by the same person count more than once, and the number will exceed analytics sessions on the destination. Treat scans as a relative measure between placements rather than an absolute count of people.

The three numbers that matter together

Scans, sessions on the destination, and one downstream action. Scans tell you the physical placement generated interest. Sessions tell you the phone actually reached the page. The action tells you the content was worth reaching. A large drop between the first two points at a slow page or bad connectivity; a drop between the last two points at the content.

Time-of-day patterns are the most useful dimension

Physical placements have sharp temporal signatures — lunchtime in a canteen, the evening in a restaurant, weekday mornings at a transport hub. Comparing the scan curve against when the placement is actually visible tells you whether the code is being seen at the right moment, and whether a placement is producing scans from staff rather than customers.

Separate first scans from repeats where you can

A code on a reusable item like a loyalty card or a piece of equipment gets scanned repeatedly by the same person. That is a good sign for utility and a bad thing to include in a reach metric. If your platform reports unique-ish visitors, use that for reach and total scans for engagement, and say which you mean.

Compare placements, not campaigns

The comparison that produces decisions is between placements within one campaign, because the audience and the offer are held constant. Comparing this campaign's scan total with last year's tells you very little unless the placements, sizes and locations were the same.

Investigate zero and near-zero immediately

A placement with almost no scans is usually not a demand problem. It is a code that is too small, too high, obscured, printed with poor contrast, or pointing at a broken destination. Go and look at the physical object before you draw a conclusion about the audience.

Keep the print run date in the report

Scan volume decays from the moment material is distributed, and different placements are distributed at different times. Without the date, a placement installed two weeks late looks like a failure. Record when each placement went live next to its numbers.