Designer QR Codes—Good Idea or Not?
The AIDC lab at the Russ College of Engineering and Technology at Ohio University has just released a months-long study of QR Code performance. Volunteers were asked to test a collection of QR codes with various forms and degrees of “customization”. All the major smart phone manufactures and operating systems were polled. The idea was to determine, with some amount of scientific rigor, how accommodating was the QR Code to graphical manipulation by each of the popular types of smart phones.
As with all bar code symbologies, the issue of performance is paramount. The OU study examines the effect of graphical manipulation of the QR Code image on how well the symbol can be decoded by the intended scanner—in the case of a QR Code, a variety of smart phones. QR Codes can be graphically modified to include brand-related color schemes, logos or other artistic touches that allegedly make the QR Code more appealing. It is possible to do this because a QR Code contains an amount of built-in error correction that permits a portion of the symbol to be colorized or “damaged” by the insertion of a logo or other graphic image in a portion of the symbol.
This same type of damage is possible, albeit to a lesser degree, with other types of symbols. For example GTIN-12 barcodes on consumer packages also have a degree of built-in error correction, but in this scanning environment, where the barcode structure and image, as well as the scanners that read them are tightly controlled by industry accepted standards, no such “designer damage” is allowed. Although there is an ISO specification for the image quality of a QR Code, there are no comparable standards on the scanning performance of smart phones—hence the importance and interest in a study such as what OSU has conducted.
The results of the study, which can be read here, are that designer QR Codes can be read between 9.6% to 88.6% of the time, depending upon what kind of smart phone you have and specifically how and how much of the QR Code has been damaged. In other words, designer QR Codes can be successfully scanned an average of 61.5% of the time. Smart phone scanning success or failure can be attributed to differences in smart phone camera optics, operating systems and decoding software, not to mention a wholly unpredictable universe of scanning situations and ambient lighting conditions. This is made worse by graphics designers who don’t really know how much of the symbol can be damaged, or where the designer modifications are least damaging to the functioning of the symbol. We have heard numerous accounts of QR Codes being modified on a trial and error basis, the “verifier” being the designer’s personal iPhone. This is like using a scanner as a verifier—something we have written about extensively in earlier articles. But with QR Codes it is worse—smart phone cameras are not manufactured to any performance standard for their use as a QR Code scanner.
Unlike other symbologies, QR Codes have a dual function because of the role they play in advertising. While this might not be as mission-critical as a patient wristband or a barcode on a prescription drug where a scanning failure could get somebody injured or even killed, advertising is often a major cost where a return on investment is expected and needed. Why run the risk of crippling or even killing QR Code performance by modifying the image?
The very presence of a QR Code on an advertising piece says “scan me”. Unless the graphical modifications make the QR Code stand out from the rest of the graphics, they might—in an effort to make it “prettier”, make it less obvious–or worse. If the graphical extravagances make it any more likely that the consumer will scan the QR Code, the OS study makes it abundantly clear that they make it less likely that the QR Code will perform its intended purpose. The motivated consumer will be disappointed when their smart phone is unable to land them at the intended web page—and probably less likely they will be as motivated to scan that QR Code again, or any QR Code, customized or not, in the future.
John helps companies resolve current barcode problems and avoid future barcode problems to stabilize and secure their supply chain and strengthen their trading partner relationships.
John – You omitted to inform your readers of two important caveats. Firstly the study was performed with QR Codes displayed on a screen and for reasons that I outline here http://bit.ly/Z4YOx4 cannot be extrapolated to print. Secondly there was no control for which scanning software was used. Since there are more scanning apps than people who took part in the study the results obtained are about as useful as a chocolate teapot.
Roger: Thank you for your comments. Here is how Dr. Kevin Berisso, Director of the AIDC Lab at Ohio University and author of the white paper responded:
“You are correct in that the extrapolation from scanning on the screen to scanning paper is difficult, but I would respectfully disagree that it cannot be extrapolated at all. In a worst case situation (from the point of view of the report), the paper might result in better success rates. However, having said that, keep in mind that the report is trying to warn designers of the potential problems – not conclusively prove that one performs better than another.
To your second point, that is part of the issue (and maybe it should have been better expressed in the white paper). The designer cannot control what is used for the scanning and thus needs to keep the least common denominator in mind. If they develop a symbol that can only be scanned by RedLaser (as an example), then they are going to have a large portion of the population out there not be able to scan their symbols.”
OK.
First point “I would respectfully disagree that it cannot be extrapolated at all”.
Extrapolation is a well defined statistical term and does not apply in this case. If Dr. Berisso means speculate then yes what he suggests is possible but yet to be proved.
Second point “If they develop a symbol that can only be scanned by RedLaser (as an example), then they are going to have a large portion of the population out there not be able to scan their symbols.”
On the other hand the opposite may be true, it may be that every app except RedLaser is able to scan the specific QR Code. In which case a very small portion of the population will experience a problem. Unless you control for what is probably the most important independent variable (the app) then the rest is meaningless.
Dr. Berisso is to be admired for attempting this survey but it is seriously flawed. I hope he conducts another on a larger population set and with a more rigorous methodology.
John – You omitted to inform your readers of two important caveats. Firstly the study was performed with QR Codes displayed on a screen and for reasons that I outline here http://bit.ly/Z4YOx4 cannot be extrapolated to print. Secondly there was no control for which scanning software was used. Since there are more scanning apps than people who took part in the study the results obtained are about as useful as a chocolate teapot.
Roger: Thank you for your comments. Here is how Dr. Kevin Berisso, Director of the AIDC Lab at Ohio University and author of the white paper responded:
“You are correct in that the extrapolation from scanning on the screen to scanning paper is difficult, but I would respectfully disagree that it cannot be extrapolated at all. In a worst case situation (from the point of view of the report), the paper might result in better success rates. However, having said that, keep in mind that the report is trying to warn designers of the potential problems – not conclusively prove that one performs better than another.
To your second point, that is part of the issue (and maybe it should have been better expressed in the white paper). The designer cannot control what is used for the scanning and thus needs to keep the least common denominator in mind. If they develop a symbol that can only be scanned by RedLaser (as an example), then they are going to have a large portion of the population out there not be able to scan their symbols.”
OK.
First point “I would respectfully disagree that it cannot be extrapolated at all”.
Extrapolation is a well defined statistical term and does not apply in this case. If Dr. Berisso means speculate then yes what he suggests is possible but yet to be proved.
Second point “If they develop a symbol that can only be scanned by RedLaser (as an example), then they are going to have a large portion of the population out there not be able to scan their symbols.”
On the other hand the opposite may be true, it may be that every app except RedLaser is able to scan the specific QR Code. In which case a very small portion of the population will experience a problem. Unless you control for what is probably the most important independent variable (the app) then the rest is meaningless.
Dr. Berisso is to be admired for attempting this survey but it is seriously flawed. I hope he conducts another on a larger population set and with a more rigorous methodology.