In September of 2008, one of our designers, Greg Washington, began discussing a frustration he had. When starting any design project he would assemble a collection of images. These images are the foundational design research he would do before opening Photoshop or sketching. Over years of projects he collected dozens of folders of loose images on his hard drive.
Most designers maintain private collections of visual inspiration and research to fuel their creative process. Their typical workflow involved going out on the Web and downloading images to a local folder. There these images would sit with a bunch of other badly named files like img4065.jpg and fre566ba01.tif. If highly motivated, designers might bring those images into another program like Illustrator, InDesign, Photoshop or Keynote to create a “moodboard”. These files—both the individual files and the moodboards—might get printed once, or thrown into the depths of a server only to be lost.
Greg was looking for an easier way to find, organize and store images that would serve as creative inspiration. He wanted to be able to easily share these images and see what other designers were using as inspiration. He wanted to be able to network those folders that he knew were sitting on everyone’s computers.
We loved the idea and decided to put some effort behind building a site that could do this. Through October, November and December of 2008 about 4 people worked on a site that we called Image Spark. It launched in January of 2009.
As of December 2012 Image Spark has over 1,152,487 images, 21,996 moodboards and 48,947 user accounts. It had 155,339 visits last month, 89,816 of which were unique. Those users generated some 968,057 page views.
In 2012, it had over 1.1 million unique visitors viewing almost 9 million pages.
So why are we considering shutting it down?
How 37Signals made us all believe
In 2008, when we were building Image Spark, we discussed, how we could commercialize it. We discussed different business models, including display advertising, premium features and membership. We thought we had an idea that could be commercially viable. But how would we know what to invest in it? Should we pull people off client work? Should we take some of the cash we had saved as a business and hire additional people?
We convinced ourselves that we would know very quickly whether it was successful or not. People would either love it or they wouldn’t. The “people” would tell us. Usage would give us evidence about whether to put more resources into it. In retrospect, this seems insanse. The “Field of Dreams” approach to product development… build it and they will come. Why did we believe 6 weeks after launch we would know whether to sacrifice part of our services business for Image Spark? Because 37Signals made us believe it was possible.
37Signals was the worst thing to happen to services businesses trying to make products. They fucked it up for all of us, because they made it. For those of us old enough to remember, 37Signals was a services company like Teehan+Lax. They had clients and did work for hire. Of course, 37signals isn’t a services company anymore. They make amazing digital products and their success is enviable.
What’s more enviable is how they made the transition from services to products. They seemingly did it with very little disruption. One day they had clients, the next day they were running Basecamp. Cash started rolling in.
The great lie, agencies tell themselves is that it can be done. You can make the transition. Look, 37Signals did it. So can we. Image Spark might have been our Basecamp. Several years later, a startup called Pinterest proved there was a business in image sharing. Why wasn’t that us?
Let me explain why, I think, 37Signals were able to do it and why we failed with Image Spark.
Clayton Christensen, the father of disruptive innovation, says, “you can’t start a disruptive business from inside an incumbent one.” The incumbent business will always take the resources from the disruptive one. He argues that if you want to create a disruptive business you need to isolate it from the incumbent business. The disruptive business needs its own values, processes and resources to be successful.
In our example, the incumbent business, Teehan+Lax, which gets its revenue from charging clients, was too powerful for Image Spark. As soon as Image Spark launched, the people who worked on it (the resources) went right back on to the work that paid the bills. In the spring, Greg (a driving force behind Image Spark) left for an opportunity on the west coast. Image Spark became the Christmas toy that sat neglected in the corner.
So how did 37Signals do it? Basecamp was a project management tool that 37Signals was using to manage their clients.
“We built Basecamp because we needed it. I’m a big believer in investing in what you know and what you need. We invested our time, energy, and focus into building a product that we knew we needed to run our own business. When you build what you know, and when you use what you build, you’ve got a head start on delivering a breakout product.” [source]
Basecamp got the resources (capital and human) it needed because it fueled the incumbent services business of 37Signals. It was able to take root because it wasn’t disruptive but highly complimentary.
This allowed them to mature a product and then realize, “hey others might like this”. This is why, I believe, 37Signals succeeded in their transition, and most agencies will fail.
Image Spark had some complimentary aspects but it wasn’t like its existence was crucial to Teehan+Lax’s services business. After we launched we got positive response from the design community, but not enough to gather the evidence we would need to invest heavily in it.
This is actually what happens with most products. They are rarely complete failures nor are they complete successes. They limp into the world on launch and then the real work starts. The real work of maturing and scaling.
Image Spark dies a horrible death and becomes a zombie
We did the worst thing you can do for a product: we left it alone. Over the past 3 years we’ve made only a few incremental updates to Image Spark, all on the backend. We moved it over to AWS from a dedicated server and then later re-engineered the database to improve performance.
There was never enough evidence that we should invest in it and so we just left it out there. It cost relatively little to keep open (about $2500 in server costs) and some people use it. But the product has become, what Joe Olsen at Phenomblue calls, “a ‘zombie idea’ not dead, yet not alive either.”
Image Spark is our zombie and now we need to decide whether to kill it.
In December, MongoHQ informed us that our database requirements have grown to such a size that we need to upgrade to a Large server. Our AWS bill, while not huge, is getting larger not smaller. Next year, we estimate Image Spark will cost us about $9,000.
We aren’t sure it’s worth it.
What do we do now?
We are faced with a decision about the future of Image Spark. It has become this zombie walking through our business. A small asset is growing into a increasing liability.
Stock investors use a simple model for making investment decisions… buy, sell or hold.
We adapted that model to make investment decisions in our business.
Improve – Invest the required resources to improve the product so that it is valuable to users.
Maintain – Do nothing more than is required to keep it alive
Kill – Shut’er down
Sell – Can someone else find value in what’s been built?
Right now, Image Spark is on the kill list. We aren’t interested in maintaining it, finding a buyer and negotiating a sale feels like a distraction. Improving it, means diverting resources from money making activities to Image Spark which in a post-Pinterest world seems foolish.
We’d be interested to know from you, what do you think we should we do with Image Spark?
Zombie Photo Credit: e_monk