[VIDEO] Big Data and Sizing: An Interview with WDR

In this interview with Fit Analytics and Zalando, German news outlet WDR takes a look at how apparel retailers leverage big data to help customers find the right size.

Sebastian Schulze WDR Interview

Video | WDR Markt | Aired on 09.09.2020 | By Philip Hinken

The returns problem costs companies money and hurts the environment. That’s no secret. And while online retailers are working to implement various ecommerce solutions to improve the shopping experience and streamline online operations, consumers are also searching for ways they can make better buying decisions. According to a Nielsen study, 3 out of 4 millennial shoppers are willing to spend more for sustainable offerings. With the collective focus shifting from consumerism to sustainability, online shoppers are taking a hard look at their shopping behaviors. In this video report, Westdeutscher Rundfunk’s program MARKT takes a look at how apparel companies leverage big data to solve sizing and shows consumers what sizing solutions are all about. This German-language interview is transcribed in English below.

English Language Transcription

Teresa: You can’t even put it on!

Larissa: It’s really bad!

Teresa: That’s like a dress.

Moderator: [00:06] Online shopping can be a disaster. We all know that.

Larissa: Two people could squeeze in here!

Moderator: [00:14] The consequences aren’t so funny. One out of every two apparel items ordered online is sent back – 250 million returns every year. But artificial intelligence can help. [00:27] The days of the “baggy” or “painted on” looks are over. Online shops compare the body measurements and shopping behaviors of millions of consumers. That’s supposed to help shoppers find the right size. [00:41] We’re going to test it with students Teresa, Larissa, and a bunch of clothing ordered online. [00:49] The claim sounds great: “Say goodbye to insecurity!” More and more online shops use online size recommendation solutions: Tommy Hilfiger, Peek & Cloppenburg, About You, Boss, Bonprix, Zara, and ASOS. [01:04] Teresa and Larissa work their way through the questions. Size. Weight. And… belly shape?

Larissa: [01:11] Flat, average, or curvy.

Teresa: [01:16] How old are you? Ok, it says here “Why do we ask your age?” (reading) Age has an impact on how your weight is distributed. Knowing your age helps us recommend the right size. [01:25] Alright!

Moderator: [01:29] Then the online shops want to know if you prefer to wear your clothing tighter, normal, or a bit loose, and your favorite jeans brand. [01:38] Teresa grabs her favorite Levis. They will be compared to the pants she selected from New Look.

Larissa: [01:44] They want to know the length of the leg. They want to know a lot, but fine – it most likely helps find the best size.

Moderator: [01:54] Two minutes later – a clear size recommendation is produced with an 84% probability. For Teresa’s jeans: size 36. Larissa’s Dress: size small – something she wouldn’t have normally dared for fear that it would be too tight or too short.

Larissa: [02:13] They recommended a size smaller for me. It throws me off a bit because I would have not chosen that size, but I’m curious. We’ll find out later if it really fits or not.

Moderator: [02:22] The two put together a shopping cart with various clothes for which they would have instinctively chosen other sizes. A few days later they can try on a total of 12 articles of clothing. Ideally nothing will have to be returned. 

Teresa: [02:36] These pants were compared to my favorite pants and fit exactly the same. Maybe they’ll be my new favorite pants!

Moderator: [02:49] Teresa is happy with the jeans. [02:53] The safari-dress is for Larissa. Can she move without busting the seams open? Do the shoulders have enough room?

Larissa: [02:58] I would have normally ordered one size bigger, but was recommended one size smaller. I would have probably ordered a medium, this is a small. It’s actually somehow better!

Teresa: [03:08] Yes, definitely. 

Moderator: [03:10] Behind the scenes for most online shops is the company Fit Analytics from Berlin. They run the sizing tool for dozens of retailers.

Sebastian Schulze: [03:20] On average we are able to reduce returns for our partner shops between 4-5%. That means a yearly number of 1 Million fewer returns [in Germany].

Moderator: [03:32] How exactly does the computer do this? We’ll use our tester Teresa as an example. She is 164 cm tall, weighs 52 kg, and prefers slightly looser clothing. The computer finds her body doubles in all of Germany, and sometimes internationally – it locates other shoppers with the same dimensions and preferences. The condition: all of these shoppers have bought this dress. The program can see who kept it and which sizes fit. In [Teresa’s] case, most of her doubles wore a 34. And on Teresa?

Larissa: [04:06] That looks like it was made for you. Teresa: Exactly, yeah!

Moderator: [04:10] It works with millions of customer data inputs. 

Sebastian Schulze: [04:14] Last year alone we had over 300 million users of our sizing solution who went through the process and answered the questions. Naturally that’s a hugely significant foundation. We cover 18,000 brands and have their size charts on record – that’s the second important point. And third – returns.

Moderator: [04:36] But there’s another major player in the fashion industry: the biggest online fashion retailer in Germany – Zalando. Zalando doesn’t work with Fit Analytics and instead has had their own big data project for the last 3 years. Stacia Caar is the boss.

Stacia Caar: [04:52] I would love to offer my customers an experience where they wouldn’t even have to select a size. A Zalando customer should trust that we will simply send them the correct size. 

Moderator: [05:06] The manager shows us a prototype of the questionnaire that Zalando will soon launch on its website. Body dimensions, fit preferences – looks similar to the competition. But in the future at Zalando, the artificial intelligence will even generate a 3D model. The customer will be able to see the tightest areas of a piece of clothing. They will know where the clothes will be too snug before they order.

Stacia Caar: [05:32] When we can show a customer the areas where it will be too tight, then they will be able to say, “Yeah, tighter in the hips, I like that.” It’s a completely new type of information that we can offer through such images.

Moderator: [05:44] Teresa and Larissa can’t try the 3D view yet, but Zalando already offers concrete size recommendations. They work with customer feedback and returns data among other things.

Larissa: [05:57] The whole time on the side it shows all the suggestions. Here they suggest a small. It’s definitely practical. 

Moderator: [06:06] Practical, and practiced as well. This fitting department is one of a kind in Germany. Every week models try on 800 different pairs of shoes and 500 new pieces of clothing. According to the brand, this leopard-print dress should be a size medium, but is that really correct? It’s pretty loose around the waist. And on the shoulders and hips? Plus one size. [06:42] Zalando sells 2000 different brands. There are a lot of deviations to make a note of. According to Zalando, 20% of clothing fits differently than stated on the tag. Every model embodies an exact clothing size, for example a perfect 36. [07:06] It’s a huge undertaking – but one that pays for itself since returns cost the company an extreme amount of money. But does the system pass our test? Larissa selected a jean dress from Zalando. 

Teresa: [07:20] That looks great.

Larissa: [07:22] Yeah it fits really well. I’m especially surprised because it suggested I buy an extra small, one size smaller, and I thought it would never fit. But this is almost too loose. 

Teresa: [07:34] Yeah, right? Imagine you had ordered a small. Then it would have really been too big. 

Moderator: [07:39] Teresa’s short blazer also fits. She was worried that it would look like a circus tent in the suggested size – it doesn’t. Even so, the system isn’t flawless.

Teresa: [07:50] I just tried these pants on and there was no way I could fit in them. Look, they look super small. There’s no way that will work. 

Moderator: [08:00] Wishing for a wasp-waste doesn’t help. By the way, that was the only exception in our test. 11 out of 12 items fit. Impressive progress compared to the usual returns quota.

Teresa: [08:17] You definitely save the time, especially when later you would have to go to the post office to send the items back.

Larissa: [08:22] And it’s much more sustainable when you don’t have to buy 3 or 4 sizes, or even 2 or 3, but simply one that fits in the best case scenario.

Moderator: [08:30] And then what you see here will hopefully not happen anymore.

–END–