The Web Needs an eHarmony for Travel

When Dev and I began exploring ideas after PlayCafe, I considered what I would personally want to use. One idea immediately came to mind: an eHarmony for travel.

I am seriously considering living abroad. I’ve been in the Valley for 12 years now and while I love it, I feel some wanderlust. My criteria for a new place are:

  • Within 5 minutes of a beach, preferably warm-water
  • English-speaking, since the only other language I know more than ten words in is dead
  • Relatively low cost of living
  • Safe and somewhat modern

The question is, what are all the cities in the world that match this? The answer is surprisingly hard to find.

Travel booking sites like Expedia and travel guides like LonelyPlanet assume you know your destination. Travel social networks like Tripwolf have people to ask, but that’s manual and hit-or-miss. You can Google terms and guess like I did – New Zealand and Australia fit – but that’s inefficient. It took my roommate to suggest Costa Rica.

What’s needed is a data-driven, travel-matching system that shows you which cities match your needs eHarmony does this for dating: tell it who you are and it shows you matches. Travel sites do the equivalent of asking you possible names of who you’d like to date. When a search engine asks you for more information than you have, it’s not doing its job.

Travel is a $100 billion market and not going away even in a recession. A friend in the industry says about 70% of travel is for business and non-discretionary, and of the 30% consumer market, about 70% goes to top 20 cities. Assuming this search engine wouldn’t change top 20 behavior much, that means the long tail of consumer travel is about 9% of the total travel market, or $9 billion. That’s still pretty big.

Note this will be useful for both the vacation and permanent moving market. The latter is especially valuable since people spend tens of thousands moving and that decision must be vetted more.

The site could collect however much information travelers want to give:

  • Environment: temperature, humidity, rainfall, landscape types
  • Culture: languages spoken, religions, ethnic diversity, openness to foreigners
  • Government: tax rates, type of system, economic and social freedoms
  • Safety: crime rates and types, natural disaster patterns
  • Things to do: popular sports, activities, night life, cuisine options, tourist spots
  • Price: cost of airfare, hotel, car, food, rent, activities, schooling, housing prices, health care

Some travelers will only need one or two search filters; some will have highly detailed needs. The site can offer wizards to guide choices,  wikis for user reviews and content, and forums to connect with other searchers.

Business model
Anything that gets people to travel is lucrative. A typical one-person, three-day trip costs about $1,000 in airfare, hotel, car, food, and activity packages. Longer or family trips are at least several thousand. Each booked item can earn a commission of $20-200+; lead-generation in travel is big business.

The site’s search and data APIs could create demand and convert uncertain buyers for several types of partners:

  • Travel booking sites that can increase purchases by showing travellers their best matches. An Expedia VP said their users visit 10-15 times before buying. Reduce that just a few visits and Expedia will be thrilled.
  • Transportation vendors such as airlines, hotels, cars, and cruise ships that want to spark demand.
  • Travel guide sites that want to suggest your best destinations to sell relevant guides and ads.
  • Even weather sites that want to monetize their information better. “It’s 72′ and sunny in Hawaii today. See if Hawaii is right for you!”

Partnerships are essential for this idea to gain scale. Travel is too crowded a market to compete without a lot of partners or a ton of funding. Fortunately this offering is unique and valuable enough to gain partner interest. When I interviewed a VP at Expedia on this, he was ecstatic at the prospect, offered access to Expedia data, and began selling me on why I needed to start this.

Because travel is big business, it’s very crowded. Travel keywords cost tens of dollars per click on AdWords. Hundreds of sites focus on SEO to get a sliver of Google juice. There isn’t that much innovation in travel but it’s still a ruthless market.

There isn’t much out there that is a direct competitor.

  • Uptake has some matching features and categories of vacations you can choose, but still requires you to name a destination city, defeating the whole purpose of matching.
  • has a lot of detailed data on cities and a surprising 6 million monthly visits, but no matching system and an awful interface straight from the playbook of Craigslist.
  • MyIdealBeach is a nice matching pilot from Orbitz, but only shows beach destinations from a limited set. Says Orbitz’s press release:  “Our research has shown us travelers want a different, better way to search for complex trips than by dates and destinations.”

Has anyone seen anything else like this?

Starting this is a bit time-consuming but fairly straight-forward.

  • Find reliable sources for the above data. Start with a few main traits, then expand as users tell you their needs. The Weather Channel, Expedia, Fodor’s, the CIA factbook, and many other sites have this data and some already have APIs.
  • Build a basic search engine to query the data on keywords and pre-set options.

That would yield a usable beta. You could then add social networking and wikis, build APIs, integrate commission programs, develop partnerships, and watch the dollars fly in.

Why we didn’t do it
Dev and I were open to starting this, but had a few key concerns:

  • Most people don’t move or take vacations often. We’d be lucky to have visitors return every six months. High user churn means constantly having to find new ones or targeting the small sliver of frequent, high-end travelers.
  • Search engines are a pain to market. This one is actually a little word-of-mouth viral in that people often travel together and may share search results and itineraries. Still, gaining search share is a bitch when Google looms.
  • Barriers to entry are low. Any of the major travel sites could copy this in a few months if they woke up. If this site got traction, they might be more likely to buy us than build but that still sets a ceiling on potential acquisition values. The site would need to cultivate sticky content like reviews or a social network to create a barrier.

I do believe just moderately good execution could make this a $5-20M business. What do you think?

P.S. I’m openly releasing this idea into the wild. If you steal it and strike it rich, you owe me lunch.

Start-up co-op mission statement

Create a productive and inspiring space for high-quality, early-stage internet entrepreneurs to share knowledge, resources, and support.

Most internet startups share the same early tasks: refining a business plan, developing a scalable infrastructure, creating a marketing strategy, vetting vendors, recruiting employees, courting investors, and managing logistics like legal, accounting, and operations.

Thousands of startups have repeated these tasks, often without benefiting from the mistakes and lessons of others. While advisers and venture capitalists can help with high-level decisions, we have found that sharp entrepreneurs facing the same daily decisions are often most informed on the many low-level decisions that shape execution. We want to maximize this benefit of shared experience, resources, and support with smart and driven peers.

Mistakes to avoid
Despite the theoretical cost- and knowledge-sharing benefits of incubators, they have an unimpressive history in the valley and typically make one or more of the following mistakes:

  • Equity-for-space: the incubator host often takes an equity percentage, spurning some high-quality startups to whom shared space is not worth the trade-off.
  • Passionless founders: some incubators brainstorm ideas and spin their execution off to an external team, which usually have insufficient passion.
  • VC or corporate oversight: VCs and especially corporations can have insufficient operating experience to competently guide (or leave alone) budding startups. Further, the VC/corp’s immediate goals can cause conflicts
  • Lack of focus: some incubators accept startups across all industries, such as biotech and internet, reducing the relevance of sharing.
  • Too much distraction: some incubators throw several startups into a small space to the point of sharing rooms and tables. This distraction can impede execution and lead to more of a fraternity than a productive office.
  • Too little sharing: on the opposite extreme, many incubators have no process for knowledge-sharing; they are simply office parks of individual startups.
  • Charging for profit: some incubators charge premiums for the benefit of plug-and-play space. While this is a reasonable business, it does not maximize the quality of potential tenants.
  • No marketing: despite numerous incubators, few do consistent marketing and none are recognized as a leader that attracts top talent.
  • Low standards: landlords typically only vet startups for credit and personal fit, not entrepreneurial experience or business plan. Thus, most importantly, incubators fail because their overall proposition does not sufficiently attract the highest caliber entrepreneurs.

To this end, we will build a start-up co-op with the following features:

  • No equity cost: tenants will only pay monthly rents (or possibly no rent if sponsors are secured).
  • No corporate or VC oversight: all involved parties will be entrepreneurs. Beyond being a good tenant, no startup’s decision will be beholden to a co-op authority.
  • Internet focus: we will focus exclusively on internet startups. This is where the co-op founders share the deepest expertise and the plurality of valley startups deal.
  • Limited distractions: each startup will have a semi-private space with barriers to retain focus. Guidelines will restrict interruptions between startups during work hours.
  • Cross-pollinating: to encourage brainstorming and build relationships, all tenants will be under one roof, share a comfortable leisure space, informally share meals, and be invited to weekly social events.
  • Not-for-profit: the co-op will aim to nurture its individual for-profit companies, not make money itself.
  • Quick up-time: spaces will be plug-and-play: furnished, wired, priced full-service, ready-to-go.
  • Bay area marketing: the co-op will hold monthly entrepreneurial events to raise awareness and attract top talent. These events may include mixers, industry speakers, and workshops on how to build a successful internet startup.
  • Highly-selective screening: we will set a high bar for admission. Ideal candidates will have entrepreneurial experience and join the co-op with a high-quality team and idea ready to develop. New tenants will need to be approved by all but one of the co-op committee’s members.

Our goal is to maximize the success of our entrepreneurs and become the premier Bay Area location for launching an internet company.

If we succeed in recruiting top talent, the co-op may be an attractive sponsorship opportunity for service firms. Firms that offer venture capital, legal, web design and development, IT, accounting, banking, space brokering, and other services derive much of their deal flow through networking and relationships. These firms are incentivized to find high-quality, funded or likely-to-be-funded entrepreneurs, especially firms that take on deferred fees or equity stakes.

For access to our screened teams, invites to exclusive mixers, speaking event opportunities, and association with our unique brand, these firms may be willing to sponsor rent for the co-op. Many of these firms regularly spend $5-25k for one-off, low-ROI sponsorships at events and conferences. Investing a comparable amount in high-value, long-term relationships may be a wiser investment. Free rent in a corporate-quality space would differentiate us from other incubators and greatly enhance the co-op’s appeal to entrepreneurs, thus attracting further sponsor interest and creating a virtuous cycle.

Physical tags: an experiment worth trying

How often have you been at a mixer, conference, or social event and thought, “I know the type of people I want to meet. Where are they?”

The search for relevant people is woefully inefficient. It is essentially random; it takes several minutes for two people to cycle through social graces, backgrounds, interests, and eventually mutual needs. In a place driven to match the world’s best minds and solve inefficiencies, the most common solution to this significant problem is a plain white nametag with name/title/company.

We can do better. I wish conference organizers would offer physical tags that let participants self-describe their haves (background, skills, interests) and needs (job openings, funding requirements, partnerships). The tags could be graphical, text, and color-coded to quickly convey information. Most importantly, the tags would not be mutually exclusive; people would be encouraged to use multiple tags to convey granularity. This recognizes that no one tag can convey all potentially useful information to a room of diverse people. For instance:

Role – graphic – color
Founder – light bulb – yellow
Executive – tie – white
Software engineer – code – blue
Hardware engineer – wrench – red
Product manager – conductor – purple
Investor – dollar sign – green (of course)
Business development – handshake – brown (of course)

Need software engineer – code with question mark – blue with dots (“holes”)
Need investor – dollar sign with q-mark – green with dots

Blogger – megaphone – orange
Staff – smiley – pink

(Right now, I’d wear three tags: purple, yellow, and blue with dots.)

The type and detail of the tag taxonomy could of course change for different events. A web 2.0 conference could let participants self-tag by web application; e.g. ‘social network’, ‘newsreader’, ‘video aggregator’ (thus helping investors know who to avoid). 😉 Even non-professional events like meetups and speed dating could benefit from physical tags. Wouldn’t you want to know if that certain someone is single, straight, and likes tennis?

If anyone is organizing a social event and wants help setting up a physical tag experiment, email me at

P.S. Did I mention I am hiring? 🙂

How to tap the collective before we’re Borg

If you had to choose the biggest ineffiency in the world, what would it be?

My current vote would be the poor distribution of arguably our most valuable resource: our experience.

Think of any decision you have to make soon, crucial or trivial, and there are hundreds of millions of people that have already made it. Every second, someone somewhere is choosing a soft drink, a TV show, a car, a profession, a spouse. Yet, their knowledge and opinions are mostly locked up in their heads, languishing unused and depreciating into hazy memory.

The idea of aggregating and sharing this collective intelligence is a staple of sci-fi. Futurists have long salivated at the idea of networking our individual brains into a massive neural net. Of course, neuroscience will need a bit more time to work out the kinks.

Is there a good solution in the meantime? The internet is the best yet, but most efforts to aggregate knowledge – the current buzzword is “wisdom of crowds” – have been mediocre:

Epinions: users write reviews and used to be paid for every read, but the system was gamed too much. Now there is little incentive to contribute and the content is stagnant.

Yelp: an Epinions-like site within a social network (founded by former PayPal colleagues) where users get recommendations from friends. There’s extra value in reviews from trusted sources but sadly there isn’t much content yet – most of my queries returned few reviews.

They face a double barrier: getting people to sign-up and invite friends, then getting friends to write reviews. I thought there would be a synergy there, that users would be more likely to write reviews for friends, but it so far hasn’t spurred the growth I expected. Bessemmer Ventures is still bullish.

Wikipedia: the most successful knowledge-sharing site to date. Any user can contribute and moderators control quality. However, the site is more knowledge than opinions and not geared toward buying decisions.

Why is there no solid opinion aggregator? I think no site has yet done the following:

Make feedback frictionless: it’s still too much of a pain to share opinions. Users of Epinions have to visit the site, begin rating with four text fields, sign-up with another four fields, choose two more radio buttons, then preview comments – that’s five pages and twelve decisions for one little opinion. Yelp requires even more: seven pages and seventeen decisions before I stopped counting. This causes sites to overrepresent extreme opinions that skew the results – especially negatively – because those users are willing to endure the friction.

A better model would be eBay’s simple numeric system (+1 / 0 / -1) or Netflix’s one-click star system (though both still require sign-up). This reduces the system’s precision and depth, but I think the volume of opinions grow enough to justify that.

[Ed: As Chris Law writes in a comment, truly frictionless feedback would mean no change in behavior from users; feedback could be gleaned implicitly. He cites Amazon’s recommendations as an example but those are more targeted cross-selling than feedback; they are based on sales before feedback is possible (i.e. “here’s what others bought” vs. “here’s what others enjoyed)”. His company, Aggregated Knowledge, may be on to something.]

Allow time-of-trial ratings: users should be able to share opinions when they are freshest and this is often soon after a trial experience (though this may taint objectivity, especially on something complex or emotional like a car). Otherwise, opinions can be cloudy or tainted by third parties.

Kaiser Permanente gets it right with their Opinionmeter feedback kiosks. The kiosks are placed near the exits, ask two simple questions, and use an intuitive button interface.

Of course kiosks aren’t always practical. One fantastical alternative would be a credit card that allows a buyer to change the last digit at the point-of-sale to rate the purchase – e.g. 1234 1234 1234 123X, where X could be a numeric rating (1-5). This would require some safeguards to protect privacy and prevent merchant tampering, but if feasible, the system could easily aggregate a massive amount of opinions, deliver them for free to consumers, and earn nicely scalable revenue by charging businesses for advertising or in-depth consumer feedback. Of course, this system would only work for products and services tried before purchase, like restaurants.

Give compelling incentive: Epinions used to pay opinion writers by page view, but the system was gamed (I wonder if some anti-fraud PayPal folks could create a more foolproof solution). Yelp offers the fuzzy reward of helping friends, but I don’t think that’s enough. Marketing contests that reward a few users or points programs that reward heavy users may help, but an ideal solution would incentivize every opinion, especially useful ones. What if every positive review gave a 10% discount on another product from the same company, while a negative review did the same for a competing product?

These are not easy problems, but markets less juicy and more complex than aggregating opinions have been tapped. Why has this problem gone unsolved so long?

P.S. For a solid feedback system, check out Slashdot’s comment moderation. The interface could be better but the concepts are well thought-out.