github linkedin email rss

That time I tried to sell graphic t-shirts on the Internet

Apr 12, 2019

I had a phase where I bought way too many graphic t-shirts. Every morning as part of my interneting routine, I would check up on the latest offerings from a handful of trendy sites[1]. Fortunately, a t-shirt hobby is much cheaper than a car hobby. Mostly.

Some of the early sketches of the Nifty logo. Brendan Marnell knocked the graphic design out of the park for the logo, site, and printed materials.

Around the same time, an idea combining game theory and on-line shopping started consuming an unreasonable number of my brain cycles. The shopping experience of most online retailers boils down to browse a catalog->add to cart->buy. This closely mimics the traditional brick and mortar experience as we are, at heart, our parents. Thus just like their parents, online retailers find themselves locked in the same tough differentiation battles on price, marketing, product, and execution. We have this whole new medium of technology and no burden of physical presence, and online retailers largely stay within the safe confines of the practices defined in a physical world. This screamed at me as an opportunity to differentiate by using technology to turn the online shopping into an experience in itself rather than just purchasing a good[2]. I did not expect to topple Amazon, but I might be able to carve out a reproducible niche using a technology twist as my edge. Given my background, I naturally turned to thinking about making a shopping experience more game like.

Games have a way to win. Most retailers use some combination of promotions like coupon codes or “buy 1 get 1 free” or a clearance sale to help move inventory. Might there be a way to turn those promotions into a game which savvy customers earned free items rather than just handing out the discounts carte blanche? The retailer would still give away the same amount of margin in the end, but this way a group of customers emerged as winners and thus hopefully more engaged and excited.

Most ideas come and go, but the its-so-out-there-it just-might-work feeling persisted with this one. I left my job at Maxis in 2011 and started work on what would become Nifty.

The Nifty homepage before a new batch of t-shirts launched.

Nifty would post 5 uniquely designed graphic t-shirts for sale for a week. At the end of the week, I would tally the total sales of each design. If you ended up purchasing a t-shirt with the least popular design, your next t-shirt was on me. I thought of it as a way to reward people with unique taste. Even better, you could turn your free t-shirt into another free t-shirt if you managed to pick out the least popular design again in the next sale. With a particularly unique taste or knack for game theory, you might never need to pay for a graphic t-shirt ever again[3]!

To make the game more interesting, I would regularly update the site with an arrow pointing to the least popular design at the moment. Over time, I was hoping to add even more features to fuel the game theory, such as showing polls on what people thought would be the least popular design or putting on display what past winners were purchasing with their free credit.

From an operator’s perspective, I was essentially offering a buy 4 t-shirts get one free promotion except I did not care who bought the 4 t-shirts. If you bought all the t-shirts during a sale, you were guaranteed a free t-shirt a.k.a 20% off. However, if you were savvy at the game, you could get maintain a 100% discount.

Mockups of a few t-shirts sold on Nifty.

Creating and running Nifty was a nifty experience. This was the largest scale endeavor I had ever undertaken by myself[4]. Along the way, I built and ran the website, tracked down and made royalty agreements with artists for t-shirt designs, worked with a vendor to print the t-shirts[5], arranged photography with models wearing the t-shirts, managed the t-shirt inventory in my studio apartment, ran advertising campaigns, and manually packed and shipped so many t-shirts. I learned once physical goods come into the picture, the amount of work and financial commitment needed to get something out the door really takes on a whole new level.

I chose t-shirts as the first item to try on Nifty because of my familiarity and their good margins. Browsing any t-shirt site with inventory, you’ll notice frequent discounts. Since so many t-shirt sites seem to stay afloat, this means the t-shirts which do sell compensate for the duds discounted to clear inventory. If you look into costs of making your own graphic t-shirts, an American Apparel base shirt with an intricate multi-color print and tagless labels will run you around $5 when purchased at a high enough quantity. When you manage to sell t-shirts for $25, there’s some room to play around.

I was particularly happy with the website and backend[6]. E-commerce sites, such as Amazon, consistently write home about high page load times and unnecessary friction significantly reducing sales. Yet, so many of the sites we use, even today, frustratingly ignore the advice. I went to great lengths to build a fast and painless shopping experience. The site looked great on any device, served a very small and optimized requests footprint, took heavy advantage of caching, and ran a minimal amount of Javascript. The front page would load completely from scratch in a couple hundred milliseconds on a bad day.

Making sure the site looked good on as many devices as I could get my hands on.

I am still very proud of taking an idea in my head and seeing it through to making sales. Unfortunately, Nifty was financially untenable in the end. I tried numerous experiments, and nothing worked well enough to keep going without bankrupting myself. You can never be entirely sure why something failed, but here’s some of my favorite speculation:

  • Buying a t-shirt is really discretionary. You either really want the t-shirt and will pay any reasonable listed price. Or you don’t want the t-shirt and aren’t even willing to cover the shipping. Ever turned down a free t-shirt? I know I have. I played with pricing a good amount. I even ran an experiment where I changed the game to lower the price of a t-shirt every hour until the inventory sold out. The inventory did not always sell out 😜. In the end, none of my pricing experiments seemed to move the needle.

  • Not discerning enough with the t-shirt designs. This one bothers me the most. I still wear many of the t-shirts I printed and really dig them. However, I selected and printed too many stinkers. Many of the designs ended up not working very well as t-shirts or just weren’t good designs in the first place. I also wish I would have stuck to a more tight aesthetic amongst all the designs and focused more on what I would wear instead of trying to appeal to others.

  • Advertising completely failed. I went in expecting really conservative conversion rates on advertising. While I do not remember the exact numbers, they were abysmal and well below comparables. I might as well have lit the money on fire. I was either doing something fundamentally wrong or people really did not want what I was selling.

  • Too many inventory options. I offered too many sizing options which made purchasing, managing inventory, and selling through really difficult at my small scale. I often think I should have started with only the most popular sku, a Men’s medium, and tried to slowly grow from there.

  • A more simple MVP. There’s a world where I tried out a couple iterations of the concept a bit quicker without building out so much. But out of everything above, I don’t lose any sleep here. I’m happy with how the site turned out, the time to launch, and the amount of money I invested.

Nifty failed me, but I still have a strong suspicion there is an opportunity to play with the online shopping experience to differentiate from competitors. Penny auction sites are an example of a mechanic which works but feels really grimy. I have high hopes for social shopping sites which feel like the right combination of real world human behavior brought online in a unique way. For now, I continue to watch on the sidelines.


[1] Threadless, Design By Humans, TeeFury, and Woot to be precise.[/f]

[2] I think of huge successes like the Rainforest Cafe, the Hard Rock Cafe, or Dave & Busters as really emphasizing the experience to differentiate themselves in the very competitive restaurant industry.[/f]

[3] This model raised “fun” legal concerns. Each state in the United States has differing rules concerning contests. The hope was this would be considered a game of skill, else I would be required, in most states, to provide a way to enter the contest without purchase. Free entry would destroy the business model. I consulted a lawyer versed in contests for his take on the legal landscape with Nifty. Nothing troublesome ended up happening as I failed quickly, but I have always wondered how this might have played out should the site have taken off.[/f]

[4] Thankfully I had the help and support of many good friends and the resources to employ a couple contractors along the way. In particular, Brendan Marnell created all the branding and site design work.[/f]

[5] Forward Printing in Oakland was top notch.[/f]

[6] Not surprising this received undue attention given my background. Always hard to get away from your strengths when wandering in uncomfortable territory.[/f]

Back to posts