Frédéric Lalonde is a serial entrepreneur in the travel business. In 1997, he co-founded Newtrade Technologies, raised $7 million of venture funding, grew the company to 70 employees and eventually sold it to Expedia in 2002. Newtrade built one of the world’s largest connectivity platforms for the hospitality industry. Frédéric then stayed at Expedia for four years as Vice President.
Now Frédéric is changing the travel industry once more with Hopper. Hopper is a search engine for planning trips, currently in closed alpha testing. We wanted to know more about Frédéric’s background and what he has learned in creating Newtrade and now Hopper.
Riku: How was Hopper started and where did the inspiration come from?
Frédéric: We started Hopper because planning a trip online is unnecessarily tedious and time consuming.
Online travel is a unique segment because the information on the Internet is incredibly fragmented and messed up. There are thousands of websites that offer product search (flights, hotels, cars, etc.) and hundreds of thousands of travel content sites. Then every local operator offering tours, activities, events, etc. has a website – that’s a couple million more. The space is also plagued by SEO tactics and pricing volatility.
There is no Amazon, iTunes or even a real Google for travel. I am not even sure it’s entirely possible to build one. So naturally that’s what we decided to do with Hopper.
Riku: Hopper was founded in April 2007. What stages has Hopper gone through since then?
Frédéric: Because we are tackling a ridiculously hard problem we have pretty much been in “stealth mode” since we started in 2007. That is to say that we haven’t built a company yet, we have mostly been a small bunch of guys hacking away at the problems, looking for a viable solution that scales and that people will want to use. I think we have it now, so we are just about to start the process of scaling up, which means we are hiring and I have begun fundraising again.
Riku: Hopper has pivoted a few times. What signals did you follow that made you pivot and how were those decisions made?
Frédéric: When we started we had this idea of a building a hybrid between a “semantic wiki” (something like freebase) and a search engine. This was back when Jimmy Wales had just raised all that money for Wikia and Calacanis released Mahalo Search. Nobody had figured out that you couldn’t mix search and wikis yet.
In the summer of 2008 we did a small release with just a little bit of content under the openplaces.org brand. The site had a really strange user experience – when you think of it, asking users to edit search results is a really clumsy idea.
It quickly became apparent that we didn’t understand how editorial communities work, so I met with guys like Evan Prodromou (Wikitravel) and Jack Herrick (wikiHow). The answers came quickly. I remember sitting with Jack in a Korean restaurant listening to him explain what motivates people to write articles and the driving forces behind Wikipedia’s Spanish fork. I just kept thinking: “We can’t do that!”
We pulled the site, dropped all the contribution features and scrapped 90% of our code. Then we went all in into search, even if we barely understood what TF-IDF was back then. It hurt like hell at the time, but it was the absolute right thing to do. (TF-IDF – ‘term frequency–inverse document frequency’ – is the best known weighting scheme in information retrieval. This weight is a statistical measure used to evaluate how important a word is to a document in a collection or corpus.)
When I think of it now, it’s clear that one of the reasons we wobbled so much in the beginning is because we were scared to admit we where building something that competes directly with Google – which we are.
Riku: What have you learned from these pivots?
Frédéric: Simple is better. YouTube started out as a dating site where you could upload videos. Flickr started out as an online multiplayer game with a picture sharing option. Even Twitter started as a feature of Odeo.
It seems to me that the things that take off – I mean really take off – tend to be simple solutions to relatively small problems that a lot of people have. Entrepreneurs, myself included, have a tendency to overthink things. Often the most elegant ideas are the simplest ones, and they are just in front of us. The hard part is learning how to see through all the surrounding muck.
Riku: Is there something you would do differently now if you would start it over?
Frédéric: Move to the Valley.
Let’s be honest: the state of venture in Canada is abysmal and entrepreneurship, in software at least, is not much better. Just look at the CVCA data; they have been compiling Canadian venture industry data since 2001 and I can’t find a single report that shows positive returns – even in the years where the S&P was up 30%. Since 1997 I have watched a lot of people that shouldn’t be VCs fund bad ideas from people that shouldn’t be startup entrepreneurs, and fail. Most of those people, on both sides of the table, are doing something else now.
I don’t actually blame the VCs for the current mess though. There are much bigger problems on the startup side that need to be fixed. The local schools aren’t teaching any of the core startup skills and the talented developers aren’t really learning them in the market. Product management skills are almost impossible to find locally and all of the smart entrepreneurs I meet are feeling their way around, getting most of their advice from SlideShare and Vimeo because there just isn’t anyone here who can really help.
Entrepreneurship can emerge anywhere. Even in the coldest, dampest most remote locations on earth. And it does, but the more I do this the more I am convinced that you need to evolve in a real ecosystem if you want a honest chance at being successful. There are some obvious counter examples: Groupon is in Chicago, Hotmail came out of India and Marcel Lebrun has built an amazing company (Radian6) out of Fredericton, NB. But I think that in the big picture those are the exception, not the norm.
I am hitting this problem for real now that we are scaling up Hopper. Most days it feels like running a marathon on the moon: in low gravity it’s an easy run, you just need to hold your breath for three and a half hours.
Riku: Newtrade also took about three years to build before you got funding and five years before you were acquired. Has the Hopper experience been similar or very different from creating Newtrade?
Frédéric: At Newtrade we raised $7M and grew to seventy people without a working product or any customers. I was 25, I had no real experience in software or business (or anything else). We had no idea what we were doing. I was clueless about how to build software, what product development was, how to hire properly and I can tell you no one in the company had a clue what Product/Market fit was. That’s a really dumb way to run a startup. Those where very different times, you couldn’t do that today – or at least that’s what I like to tell myself.
We did well exiting Newtrade because we worked around the clock (literally) and because – probably out of sheer luck – we were solving a real problem that real people had.
Riku: Have these companies included lots of quick iterations or have you had to spend a long time building a working product?
Frédéric: Two things I have learned about search: there is nothing quick or iterative about it and it’s not for the faint of heart. Building a search engine is really hard. Part of the problem is that when you are working with gigantic amounts of data you are front-loading all the scalability issues. Most startups don’t need to worry about scalability until they have a million people using their service – and by then they know what the product should be because it’s already catching on.
In search you are manipulating insanely large datasets, running your code on hundreds of machines, usually executing dense algorithms against a totally unpredictable corpus. You need to control everything perfectly: your code quality, you data quality, even the hardware stack, or nothing is going to work. And search is not about features or tweaks: it’s just a box where you enter words and the results are amazing, or not.
The reality of it is that a search project takes years to execute and there aren’t many shortcuts. Moreover, you can do everything right technically just to realize after you launch that nobody cares about your site and they would rather just keep using Google. Cuil raised $33M before publicly imploding in a supernova of founder ego. SearchMe raised $43M, launched beta in 2008 and hit the TC Deadpool by October 2009. Powerset burned through $22M before Microsoft picked up the pieces. Even the Blekko guys have raised $24M and have been at it since 2007. I am a big fan of what they are doing – I think it’s really clever – but it’s still unclear if they will get to P/M fit with their current concept.
Riku: What are your recommendations to our readers thinking about starting their own business or current entrepreneurs?
Frédéric: Product/Market Fit is everything. When you are in a startup it’s like everyone and everything around you is conspiring to distract you – even your own brain. You get sucked into all sorts of things that seem important at the time: UI design, conversion, marketing, partnerships, fundraising, buying office supplies. Until you get to P/M Fit, none of this really matters, it’s all noise.
Iterate as much and as often as you can. And if you can’t iterate quickly, figure out a way to get input from real users for your product during the development process. There’s always a way.
One of the hardest parts of this is killing your darlings. Between all the ubiquitous uncertainty and the rejection you encounter trying to build a company, startup entrepreneurs get used to taking abuse and criticism. But even then, most of us all have this one idea we cling on to, even when everything around you is telling you that it’s not going to work. In some rare cases it’s the right thing to do, but more often it’s like an onion (a really nasty, smelly, onion): you need to peel away all the layers of bad ideas until you find the right thing to do.
Don’t get distracted, don’t let your ego get in the way. Getting to P/M fit is the only thing that matters, everything else will take care of itself once you do.