E-tailer Zazzle has faced consumer criticism over selling black-themed t-shirts. The problem: They are modeled by white women and men.

Companies like Amazon have gotten into trouble over racism charges for products found in their marketplaces. The sellers were third-parties. Get rid of one set and others come in and it becomes next to impossible to keep up.

But although Zazzle runs a marketplace for third-party sellers, they aren't the cause of grief in this case. Instead, technology automatically takes actions that consumers see as bad and company management hasn't fixed the issue. Even after a year of getting bad PR because of it.

Product companies can sell their wares on Zazzle. Artists and designers license their images imprinted on a variety of items, whether cards, clothing, mobile phone cases, posters, or more.

To display goods on the site, Zazzle uses technology to project designs onto blank items. Customers can see what they would get without the need for the company to undertake individual production and photo shoots every time a new design came in.

The use of technology is smart. The more you can automate display images and print products on demand, the more readily the concept works with reasonable prices. However, automation has risks, as Zazzle has shown. You don't necessarily have human input to keep a computer from doing something really dumb.

When computers are in charge

Automation is the first part of Zazzle's problem. For shirt designs, the images are projected onto a physical shirt worn by a model. People can see how the item looks on a human being.

But there's no built-in sensitivity for context. The mismatch between the messages on the shirts and the appearance of the white models was striking. Here are some examples:

(Yes, that is a white man, opening the door to gender complaints as well.)

But clearly they have models of color.

image 3

Twitter has had some responses.

Management doesn't take action

What is particularly surprising about this issue is that the story is at least a year old. Someone should have fixed this.

(I have questions in to Zazzle and will add their response when I receive it.)

[Update 23-August-2018 5:20PM: The company responded with a statement that said, in part, "Our designers and users are free to choose from several different t-shirt styles, each style with its own pre-posed selection of model, or in some instances a laydown option with no model at all." It also said, "In the case of the t-shirts in question, each designer is shown a series of pre-posed randomized model shots upon which their design is placed. It's always possible that gender, age, race and other attributes of the model do not match up to the specifics of the design, given the ratio of millions of designs to the 100 or so t-shirt styles we offer."

The company said "the product page photography contains a message indicating the above on most of the designs," but the statement doesn't seem as clear and also doesn't seem to ameliorate the feelings of some consumers.]

Everything is ultimately the responsibility of management. That includes monitoring coverage of the brand, identifying problems, and correcting them, whether an issue of process, people, technology, or all three.

To not realize a problem because of the complexity of how technology works and the impossibility of considering all potential outcomes is understandable. Any business will likely run into something analogous.

But you can't ignore things and hope they go away. Too often they return, time and again, to hound you, affect customer relations, and drain time and attention.

Fixing a problem like Zazzle's takes focus and resources. Solutions have their own issues. Do you hire enough staff to review every single design that comes in and decide how it should be displayed, losing the efficiency that automation promised? Will you give designers who need models a chance to specify preferred color of model, potentially opening an even bigger racial discrimination issue? Do you show the shirts without showing a human in them, possibly lowering sales effectiveness?

Making the right decision isn't easy, but that is why management gets paid.

Published on: Aug 23, 2018