In January, I was transitioning from molecular neuroscience to freelance web development.  To help pad my portfolio (and learn new technologies), I set upon building a number of side projects.

ThemeSquirrel was one such project.  It was a project so I could get my feet wet with jQuery and javascript.  ThemeSquirrel (TS from now on) is an infinite scroller for ThemeForest, a marketplace for website and wordpress themes.

ThemeForest is a really great resource – you can find pretty awesome themes for $10-$20.  Unfortunately, the method to browse themes is really, really awful.

Themes are listed in rows with tiny icons.  Mousing over the icon displays a larger “preview”…except theme authors typically “editorialize” these previews with marketing copy and flashy images.

ThemeForest hover “preview”

You don’t really know what the theme looks like until you click through. And you don’t even have a general idea unless you hover over each icon on the page.  Ughhhh.

ThemeSquirrel was a simple project to build an alternate browsing interface for ThemeForest.  You can see the current, live site here (although it is no longer managed by myself, so things have changed a little).

What I did right

I built ThemeSquirrel fast.

This was a side project meant to pad my resume.  I only wanted to get it to the point where it was functional, pretty and highlighted my ability to ship products.

I built TS in about a weekend, with a little extra work over the next week.  Total time commitment was probably around 5-6 hours.

You can check my original post for technical details.  I can assure you nothing fancy was going on.  Some simple web scraping to build the initial database, jQuery Masonry and jsTree to build the user interface.  Looking back, the JS that wired everything together is iffy at best.

I picked a project that had an affiliate program.

I was inspired to build TS because of Scrollsy and the resulting Hacker News post.  However, I distinctly remember the author lamenting the lack of a referral program:  Etsy doesn’t do affiliates.

There are plenty of crappy interfaces on the internet.  I made sure I picked one that had an affiliate program.  ThemeForest does have an affiliate program (albeit a very weak one) and it was in the tech niche.

I was pretty sure I wouldn’t make any money…but at least there was a chance that I could.

I sold it.

Yep, I think this is one of the things I did correctly.  How many little projects do you have sitting around on your server collecting dust?

ThemeSquirrel had become one of those projects.  I built it in January, showed it to Hacker News and then promptly forgot about it.

A few months later I logged into ThemeForest and discovered that it was (slowly) making money for me.  What?  I hadn’t even looked at it in months.  Was the cron-job even updating the database anymore?

ThemeSquirrel refused to die

Selling ThemeSquirrel was a good idea.  I was neglecting it, but clearly it held value.  I tossed it up on Flippa with a very candid description and netted myself $500.  (TS was sold in early April)

Total money earned by ThemeSquirrel to date: $679.

Astute readers will notice that my “hourly rate” on this was project was actually pretty poor.  Five to six hours developing, another hour or two compiling stats for Flippa, another hour or two helping the new owner to get everything set up on his server.

Probably 15 hours sunk into the project total.  That’s only $45/hr, less than a good freelancer would make.  Luckily (or not), I was a lowly biologist at the time making $32k/year in Boston.  I’d just quit that to become a freelancer, which meant I had zero effective income at the time.  An extra $700 was pretty swell.

What I Did Wrong

I ignored feedback.

I posted to Hacker News and got some great feedback.  The masonry layout is simply too difficult to browse.  Large previews should be default. Links should go directly to the “theme preview”, instead of the theme sales page.  Suggestions for an entirely new layout.

All great, actionable feedback.  Most of the changes would have been really simple to make.  I was still in “Side Project” mode though, and mentally I was done with ThemeSquirrel. I didn’t make a single change after I put it live.  In retrospect, I feel this reduced the overall value of TS.

I sold it…too soon.

This goes hand-in-hand with the above point.  Look at my traffic numbers (grouped by month, otherwise the HN spike in January makes everything else impossible to see):

Google Analytics for ThemeSquirrel

And now look at this table of stats:

January sales were clearly a product of the Hacker News post.  It isn’t surprising that the purchase percentage was poor – people visiting from HN were looking, not shopping.

However, as we move into Feb and Mar it becomes pretty clear that ThemeSquirrel held some kind of underlying value.  Unique visitors are down to reasonable numbers, CTR remained in the double digits and actually purchases stayed high (relatively speaking).

I sold in early April, so it’s difficult to judge that month.  We were moving servers and my referral code was taken down.

Looking back, if I had implemented some of the requested features I may have been able to capture a few more purchases per month.  I believe the high CTR but low purchase ratio meant people were using the site…but not finding a theme they wanted.

Some improvements to the UI may have helped keep the funnel from leaking so much.  In particular sending, clicks directly to the “template preview” page since this would have been the single largest reduction in friction.

Ultimately, I still would have sold the app.  But had I put a little more effort/time into it, I may have upgraded $50/month revenue to $150 or $200/month.  That would have commanded a far higher price on Flippa.

More marketing and SEO.

I did not have a marketing plan, and that was on purpose.  It was a side project.

That said, man, I wish I had thought of it as a real project.  I think I could have really marketed this around the designer/WordPress circles and gotten some good attention.

Furthermore, I had a database of content just sitting around.  I really should have generated a page for each theme, added some useful metadata (graphs, statistics, themes-like-this-one, etc) and let Google index them.

Of course, this is a tricky proposition.  Google isn’t fond of autogenerated content.  I think, with enough finesse, you might be able create pages that provide value-add and aren’t spammy.

Paid advertising would not be possible given the low conversion rates.  At 0.3% conversion and a lifetime value of $11, you would only be able to pay four cents per click on a CPC basis.  If you can average closer to 2% conversion, things start to get a little more feasible (22 cents).

Go sell your side project

I learned a lot from ThemeSquirrel.  I learned a bit of javascript and jQuery, obtained a nice side project for a resume and learned how to sell an app.  I also made $679 in the process.

I strongly encourage you to look through your side projects, polish one up and see if you can sell it.  Sure, you’ll be making New Shoes money, but you’ll learn a lot and hey, everyone needs new shoes.

Discussion at Hacker News