A/B testing with dynamic QR codes: how to compare two destinations and measure which works better

A/B testing is a controlled comparison technique: you split your audience into two groups, show each group a different variant and measure which achieves better results. Applied to QR codes, the process is straightforward.

Variant A: dynamic QR pointing to the current version (control). Variant B: dynamic QR pointing to the alternative version (test).

You distribute the two codes under equivalent conditions: the same print quantity, the same relative location, the same target audience. After a measurement period (7-14 days is typical), you compare the scan data and conversions at the destination.

The difference from classic web A/B testing is that here there is no algorithm automatically splitting the traffic: you control which QR goes to which physical medium. That requires care in designing the experiment, but also makes the analysis more transparent.

These are the four cases where an A/B test with dynamic QRs gives the clearest results:

1. Two versions of a landing page: You have an event poster with a QR. You create two versions of the landing: one with a video and one with only text and a buy button. You print two identical batches of posters except for the QR and put them up in equivalent locations (same type of street, same pedestrian footfall). After 10 days you compare the conversions.

2. Two price offers: You want to know whether a 10% or a 15% discount generates more net sales. You use the same physical medium but with two different QR codes, one per venue or per zone of the establishment.

3. Two restaurant menus: You test whether a menu with photos or one with text only converts more into orders. You assign QR-A to the indoor tables and QR-B to the terrace tables (making sure the customer profile is similar).

4. Two poster creatives: Two poster designs for the same campaign, each with its QR. You distribute them in two neighbourhoods or two shopping centres and measure which has more scans.

The analytics at codigo-qr.es log, for each dynamic QR, the total number of scans, the daily and hourly breakdown, the approximate country and city, and the device type. With that data you can carry out the basic A/B test analysis.

Metrics to compare:

| Metric | QR-A | QR-B | Which wins? | |--------|------|------|-------------| | Total scans | 148 | 203 | B (+37%) | | Scans week 1 | 62 | 89 | B | | Scans week 2 | 86 | 114 | B | | Peak hour | 1 pm | 7 pm | Different patterns |

Scan analytics measure interest (how many people reach the destination), not conversion at the destination. To measure conversion you need to combine QR data with your web analytics tool: Google Analytics, Plausible or the native CMS stats. UTM parameters in the destination URL help to connect the two sources.

See the article on QR codes with UTM parameters to see how to combine both metrics.

A poorly designed A/B test gives results that are useless for decision-making. These are the most common mistakes:

Mistake 1: Non-equivalent groups. If QR-A is at the main entrance and QR-B is in a side corridor, the traffic is not comparable. Each QR must have the same opportunity to be seen.

Mistake 2: Too short a period. With fewer than 7 days or fewer than 50 scans per variant, the difference may be statistical noise. Let the test run for at least 7-14 days and reach 100 scans per QR before drawing conclusions.

Mistake 3: Changing the destination during the test. If you modify the destination URL of one of the QRs mid-experiment, you mix two different variants in that QR's data. Decide the destinations before starting and do not touch them until the test is complete.

Mistake 4: Measuring scans only, not conversions. More scans does not mean a better result if conversion at the destination is worse. Combine QR analytics with landing page analytics for the full picture.