Why search is such an exciting business for startups
|Why search is such an exciting business for startups
by Jaideep Singh | March 9, 2008 | PDF of Article
Guest Post Written by Jaideep Singh
A well known VC once told me “Every business has its pros and cons, you just have to pick your poison”. Search takes a long time and costs a lot to get right. But once you’ve got it right, you’ve built something utilitarian and of lasting value. This is in sharp contrast to many web social web applications that can get be built quickly and cheaply and get mass adoption.
Any company that truly becomes a utility also becomes very valuable. In software, this is the story of Microsoft, Yahoo, Google, etc. We use these companies daily to be able to do things that we think are basic needs. For the same reasons, telecom carriers and cable networks also are valuable utilities.
Yahoo (Overture), and Google have also proven that search monetizes better than other web applications. This is not a surprise, because the search utility tends to captures user intent well.
Just as directory based approaches gave way to general full text search as web content grew, now we are seeing the transition from general search to vertical searches such as people, product, travel, health, source-code, etc. These new category search engines can improve the user experience dramatically.
However, these business take time, talent, and investment to bring to fruition. They all start with crawling web documents, filtering for the category, extracting and disambiguating entities, and then building a rich user interface.
To attain success, any new search startup needs to think about 3 main things:
1. How can you build something defensible
2. How do you get distribution
3. How to attain maximum leverage given the high fixed costs
Technorati built a good blog search solution. It took them a few years to get it to work well, but at the end of the day as the market for blog search became clear, Google quickly implemented a good solution, thereby ending Technorati’s search ambitions.
The lesson is that competing with Google head-on and on technology alone is very tough. We say a similar fate for companies in the 80s and 90s that tried to compete directly against Microsoft.
On the other hand, Kayak took on travel search but built a custom application for finding and booking flights and hotels. Its model is somewhat of a blend between Expedia and Google, which gives them a unique core competence. The deep interfacing with ticket sites, and building a simple but custom search application on top of it not only provides users a rich and convenient user experience, the strategy is defensible.
The same is also true with Spock, a people search application that Jay Bhatti and I co-founded in 2006. Users can search for people and get the latest information and news on them. The search is also human-powered so that people can add information such as pictures, tags, related people, etc. that are not easily found on the web. This aggregated search result ends up creating the default place of record on the web for people information. Also the human powered search strategy is highly defensible while providing a great user experience.
Startups have limited budgets and therefore can’t spend much to advertise their services. Thus, most successful web startups rely on zero cost viral growth. However, unlike social networks and other web applications, most search engines are not inherently viral. This makes distribution and growth a challenge.
In general, few users will discover a new search engine, and in the first few years growth will be slow.
Some search applications, especially in product search and travel have found ways to aggregate content in a way that not only is valuable to users, but also adds value to search engines like Google. Oodle and theFind are able to use SEO to grow. This can be a good early growth strategy for such search engines.
Building a search engine has a lot of fixed costs. It is not possible to have a team of 10 people build something of value. Most search engines need 30 to 50 top notch engineers and at least 2 years to get to a reasonable place. In addition to talent one needs a lot of hardware to crawl, index, and serve up a large data set.
It is therefore important to limit the scope of the problem, go after low hanging fruit, and find a way to deliver a good user experience. This is how Google did it too. This is also the strategy that Spock, theFind, Kayak, Kosmix, and others have taken.
Raising money is never easy especially when you need to raise more money than most consumer web initiatives. Be prepared to hear “why do you need to raise $7m when facebook only needed $1m” or “that facebook application only needed $100k”. You certainly need to assemble a strong technical team with search experience, and build a prototype for $1m, which is a very good way to start.
But you also have to prepare your investors. You’ll need to raise $25-$50m and spend a minimum of 3 years before getting to healthy profitability if you’re going after large fragmented categories like people or product. If you can find a way to dominate a smaller category quicker and cheaper (with a smaller end outcome), that is wise too.
With Google, Yahoo, MSN, and Ask, we’ve really only seen the first phase of an enormous and ever expanding field. The second wave of category-search companies have the opportunity to build large stand alone businesses that deliver tremendous value both upstream to Google and downstream to the category that they serve.