This post here should be a guide how to identify good and bad body data.

Over the last few years a lot has been talked about good / bad fit, return rates, sustainability and so on. There are also numerous articles talking about high return rates in online sales. Of course, there are many service providers now trying to help solving the fit issue, but you have to be careful what data you can trust.

A while ago there have been those “Fit Engines” using many strong words (computer vision, machine learning, AI etc.) and promising to help reduce returns. Since this is not as easy, today many of those companies just say they help you “guiding more traffic on your website”. The term reducing returns is mostly gone…

Further there are a few companies now saying they have hundred thousand of scans or “big data”.

How can you identify good or bad data quality?

So, the first question you should ask is where are those data coming from? And I mean what is the device the data was captured with? How representative of a country are the data? Where have all the scan data been captured (only in gyms)? How old are the data?, Etc. p.p.

There are some companies that are using scanning devices with depth sensors to capture data. Usually those sensors are the cheapest way of building a scanner but often not the most precise. The next thing is that at some of those scanners, the subjects needs to stand still on a rotating platform for over 30 seconds… Just think about all the small body movements done while being on a spinning platform. Those movements are being captured! Often those scanners have handles to hold people in place. The scanners handles force the subject into an unnatural posture effecting the body measurements, not even to mention wrist and hand measurements… Again, just think about you are on a platform for 35 seconds or so… accuracy goes away!!!!

Often reconstructed avatars of those companies show remodeled generic heads and hands, because the scan quality is too low to capture details accurately or because the hands are grasping some handles.

After you are aware of with what technology scans have been collected with, you should ask about data privacy and if every person being scanned has signed a consent form. If there is a doubt do not use those data. There is a legal risk! And have in mind that some of those sensor scanners are often in a gym and can easily being used by everyone without signing anything.

Also very important for data quality, what is the scanning garment people are wearing. As we did Size NorthAmerica, Hanesbrands sponsored all the scanning garments for consistent results. You can’t mix different underwear up (specifically not shape changing underwear) and you need consistency. Often companies put scanners in Gyms to collect data, people in gyms are usually wearing very tight / form changing clothing and underwear (sports bra, tight short, etc). You can’t use a scan if the breast has been deformed entirely by a sports bra. Just think about it…

Demographic data, one of the most powerful assets of a size survey are Socio Demographic data. Customer insights… check, if “those companies offering scans” have those data and I mean more than age and ethnicity. Size NA has asked and collected over 60 demographic questions.

Why should you use questionable data if there is was a national size survey @SizeNA that was funded by a wide range a companies - apparel, retailers, automotive, tech companies, furniture and health? Those companies have contributed to the success of their survey. 

Fig. 1 - Scan postures of Size NA