Entries by Des Traynor

Bigulo sold today, actually it sold a couple of weeks ago, but today the money arrived in my account, and I split it with Andrew. Since announcing it on Twitter, I've been flooded with congratulations, from the most familiar and most surprising of places. I've received a lot of emails asking for details, back-stories, etc. So it looks like I should say something more about it. Where to start…

Maynooth, 2006

I'm sitting in the research lab in NUI Maynooth, beside Andrew Page. If you cast your mind back, this was the time when the web 2.0 movement was at its peek. Twitter didn't exist. Reddit, Youtube, Digg, Delicious, had all received amazing valuations. 37Signals had released Getting Real. Usability had suddenly become popular and I was slowly coming to terms with the fact that I was far more interested in the web, creating software, usability, and writing than I was in my academic pursuits.

Andrew had always been a good friend, and also a serious academic rival. I guess at this point Mr. Traynor should congratulate Dr. Page, and admit defeat on that one. Anyways, I knew if I needed help to make something happen, Andrew would be the guy to turn to. As it happened, he turned to me.

Code Speaks

Bebo had taken Ireland by storm over the summer, for reasons I'll never really understand. Walking around the college, all you would see on anyones computers was that ugly logo, as boys wearily clicked through girls holiday photos. Perhaps the most frustrating thing about Bebo, at the time, was that you couldn't find anyone. For example, Andrew knew his sister was on Bebo, knew her email address and even her Bebo username, yet still couldn't find her to friend her. This was a problem. Andrew had the beginnings of a solution when he called me over.

I had spent a lot of the previous months learning scripting, writing the usual stuff, image gallery scripts, command line tools, and a fancy command line web text program. I used Perl at the time, as I had an abhorrence of PHP. I was all set to have round fourteen of Perl vs PHP when I realised that Andrews code actually worked.

Lesson #1: Working code speaks louder than words.

Andrew wanted help with presenting the information he had collected, in a stylish "web 2.0" manner. This gave me my first chance to work on an interface that people might actually use, and I made an absolute balls of it.

We published the site on Mikado, the college forum at the time, and within days we were getting hundreds of hits a day. We published it on Boards.ie, Politics.ie, and asked our friends to tell everyone. The Irish tech scene, whilst small, were very helpful, Damien started the ball rolling and with weeks we were getting thousands of hits a day. We added "Google Adsense" and waited, and waited.

Pay per click

Weeks later we were approved to display adverts, and within days we had made our first dollar. We tinkered with ad placement, aided by friends, and soon the site was earning $20 a week. Surely things can get better than this?

Lesson 2: You'll be a long time waiting for advertising revenue.

Stickiness

Andrew asked me how could we get the users to stick around. I had no idea, the purpose of the tool was search. The user comes along, finds what they need, and leaves. The answer came to me one day in the foyer of the computer science building. It sounds crude to say it, but most people on social networks are there to check out other people. You can dispute that if you like, but I've been watching this space for quite some time. Boys look at photos of girls. Girls look at photos of boys. Like everything else, Bebo sucked at this, so we fixed that.

We added Babes and Hunks of Bigulo, so that users can just look at pretty people, view their profiles, get talking, whatever. This meant we'd get more page views per visit, and increased our percentage of returning visits. These days as an analytics consultant, I refer to this as the "stickiness" of a site and Bigulo had gotten very sticky (no pun intended).

Lesson 3: It's easier to tap into existing behaviours than it is to create new ones.

Lesson 4: To paraphrase Jamie Zawinski - if your code can get people laid, you're off to a good start.

Media whores

Andrew had some experience getting sites into national media, so we started off with a press release announcing the site. That went nowhere, but our follow up article on how Bebo leaves profiles of children public by default hit the headlines. Won't somebody please think of the children. (Side note: Bigulo never searched for under 18yr olds)

We followed this up with a press release announcing which government officials were on Bebo, during the elections. It's funny to see this same issue come up in every single election since then. Enda Kenny was a "guru" of Social Media long before Barak Obama. At this point not a day would go by without someone ringing me looking for a soundbite about anything vaguely related to social networks. This was social media, before the term had been coined.

Lesson 5: Don't believe every "expert" you hear on the radio

Lesson 6: You need to see sales opportunities everywhere, not just via reddit and twitter.

1,000,000 hits

At one point around Christmas 2007 Andrew told me we'd passed the 1 million mark for searches performed. It's hard to know what to make of this. The first thought a naive businessman would have would be "imagine if they each paid a dollar". A slightly more mature would ask "Is each search worth a dollar?" or "How can we make it valuable to people, so they'll pay for it". Neither I, nor Andy, could ever answer that question, so we stuck with advertising.

Lesson #7: If you can't imagine any user paying you money, then they won't. You can take 1 million users, and be ultra conservative and say .5% conversion, and very easily earn yourself $5,000 in Excel. Getting that money into your bank account is a different kettle of fish.

It was about this time we stopped working hard on Bigulo. Why did we stop? Quite simply the tool ran itself, we were out of ideas and had other things to do. For Andrew this meant journal paper after journal paper. For me it meant learning Javascript, more about XHTML/CSS, writing blog posts all of which ultimately landed me a position as a senior analyst in iQ Content, and as luck would have it got me talking to a Dublin blogger called Eoghan McCabe.

Letting go

Bigulo moved from strength to strength, we recently hit 25 million searches. Bigulo got more traffic than any site I've ever worked on. But we simply didn't know what to do with it, so when approached by a lovely lady from the US who had money to buy it and good plans for it we realised we needed to work out what was best for Bigulo. Maybe it'll go on to be a huge success selling for millions, maybe it'll die in obscurity, we've had our fun with it.

Lesson #8: Remain objective and dispassionate about your properties ; when the right time comes, let go.

The lessons I've listed out here, are what I learned, your mileage may vary. One lesson that applies across the board is the following:

You're better off doing something, than sitting around complaining.

This is as true for two bored PhD students, as it is for out-of-work freelancers, graduates struggling to find a job, or people who simply aren't happy in their job.

Like most fun projects I've been involved in Bigulo started off as 800 lines of broken code, that we were always told that "anyone could have written". But we did, and that's usually the difference.

Just the facts

• Launched: September 2006
• Traffic at time of sale: 150,000 uniquevisitors/month
• Searches performed to date: 27.4 million
• Monthly Revenue: Never enough to sustain 2 guys on reasonable salaries
• Amount of sale: I love my job at Contrast. But even if I didn't, I'd still be showing up there tomorrow. Put it that way.
• Bought by: I'm waiting for confirmation to announce our buyer and her plans. We're keeping things under wraps until all is sorted. I'll let y'all know as soon as I can.

If you've been keeping score here at all, you'll know I don't like meta posts, but none-the-less here we go. Since we last talked, I have joined Contrast and with David, Eoghan and Paul I've been working hard to try and make a really good company. It's going really well. You can follow my posts over on the Contrast blog, which is damn sight prettier than this one. You can follow me on twitter, too: @destraynor. Let me know if you want me to follow back (following absolutely everyone gets noisy)

I will of course continue to write here, they will be posts regarding programming education, and rants about little usability quirks. I'm going to do a re-design though, as I'm sick of this look, and I can now do better. For now, if you're interested, please subscribe to the Contrast blog. Here is the RSS. And thanks for sticking around.

Vilfredo Federic Damaso Pareto was born in 1848. During his notable career in the field of micro-economics, he observed that 80% of the income in 19th century Italy went to the richest 20% of the population. This became known as the Pareto prinicple, and later on as the more general “80/20 rule” As you can imagine, it’s a significant stretch to relate an observation about wealth distribution in Italy to the design of user interfaces, but it does happen. I have three main problems with Pareto being used to justify decisions…

1. The data rarely backs it up
   Even with tonnes of usage data, or Google Analytics data, it’s very tricky to find this mythical 20% that accounts for 80% of the usage. Also, if/when you do find it, you have to make sure that it reflects what the users want, not what they click. I’ve worked with clients who were adamant that their “About page” was their most popular, only to see all the users end up there whilst hunting for a non-existent contact number. Analytics only tells you the what, not the why.
2. Some vital features are rarely used
   From what I’ve seen, less than 1% of users will change their password arbitrarily. If you’re just going to focus on the 20%, you can annoy a lot of people. There are a lot of high priority features that don’t see much usage.
3. What does it really mean?
   If I told you that Paste, Save, Copy, Undo and Bold account for 32% of the commands used in Microsoft Word, would you really alter the UI based on that? Paste alone is responsible for 11% of the commands, so it should have a super big button, right? At least 25 times the size of word count? Nonsense.

I know we use Pareto’s principle as a little nugget to capture clients ears and make them think about focussing on what's important rather than expanding their product to become the sum of all desired features. That's understandable. But in this new age of Less Software(tm), lets not get carried away. You can’t design software based solely on quantitative research. Especially when that research comes from 19^th century Italian economics.

The search engine world has found itself in a weird position. Weird, yet strangely familiar to me. Cuil.com launched just 20 hours ago, and is already the whipping boy of web applications. There are about a hundred blog posts showing side by side screenshots of Google vs. Cuil.com, with all sorts of hilarious search results. Cuil brought this on themselves, but that's only part of the problem.

Whenever I’m designing an application in Omnigraffle, I use the YUI stencils which are really great. There is one whole stencil devoted to Advertising styles, leader boards, banners all the usual stuff. They follow all the advertising standards (125x125 etc), and for the most part they do the job, that is they allow you to indicate where the ads will appear. They lull you into a false sense of security however …