It seems entrepreneurs would rather join the race to build the fastest and the greatest technology than build one that fills a void in the market.
Waikit Lau co-founded his Boston-based company ScanScout because he wanted to fix a mounting problem — monetising the growing popularity of online videos. While he probably would have struggled to raise funding had he been based in Malaysia, his Boston start-up has managed to raise some US$20 million (RM69 million) since 2005. This includes angel investors who had previously bet on Google.
In comScore’s July Video Metrix Report, ScanScout is ranked the No 1 in-stream video ad network for reaching the biggest audience, at more than 80 million unique viewers a month. His clientele includes big names like Toyota, McDonald’s and Warner Brothers.

But ScanScout is not just about delivering ads on video. Its proprietary SE2 Engagement Engine uses statistical machine learning to target the advertisements according to the video’s context and content so that the ads are relevant to users. This ensures higher engagement rates.
Indeed, Lau has come a long way since growing up in Ipoh. He left Malaysia in 1993 to pursue a degree at the Massachusetts Institute of Technology (MIT). After attending the inaugural Multimedia Super Corridor (MSC) Open Source Conference in Malaysia last June, Lau talked about the growing online video trend, and why we should not write off the MSC as a failure just yet and what he thinks can be done to improve it.

[email protected]: Tell us the story behind ScanScout.
Lau: My co-founder and CTO, Steve Lee, an American, and I have known each other since our freshman year at MIT. We had always talked about starting a company together. In 2005, when we saw broadband penetration in the US crossing the 50% mark, we were sure online video would be the next inevitable killer app after email. There were some online videos being posted back then but certainly nothing mainstream. It was still pre-YouTube.

We realised the advent of online videos would see the same kind of challenges web pages faced before. What I mean is that all technologies created up until then were ways to search and monetise text on web pages, but nothing for online videos.
Additionally, we felt we had the right skills to solve this particular problem, given our background in statistical machine learning and large-scale data mining. So, we bought some servers using our money and started building the foundation to solve the monetisation problem in 2005. We built a video search engine as well but it was never publicly released because we decided to focus on the monetisation part. We felt this was a more ‘deterministic’ problem. If you can truly solve it, advertisers and publishers will reap obvious values.

Currently, TV ad budgets are much larger than the oft-heard search engine marketing budget. For example, in the US alone, annual search engine marketing spend is US$13 billion to US$15 billion (Google’s domain) while TV ad spend is US$80 billion. Even if a small percentage of that US$80 billion gets moved into the online video ad market, it translates to a multibillion-dollar market for a sector that frankly did not exist four years ago. That’s pretty good growth.

Did any of your funding come from angel investors?
We raised some from angel investors who had previously funded Google and understood the particular problem we were trying to solve. Since then, we have grown the company to five offices in the US. We are one of the largest video ad networks in the world today, reaching more than 100 million unique users a month globally. We work with many of the leading publishers and advertisers, such as P&G, Nissan, Toyota, Universal Studios, Sony Pictures, Coca-Cola, Pepsi, McDonald’s, Holiday Inn and EA.
ScanScout is ultimately a technology company that builds targeted and optimised online video advertisement technology.

Comparing the low affinity of viewers to video ads with the high growth of online videos, how does ScanScout solve this problem?
We believe overlay ads are great for short-form (one to three minutes) content while pre-roll ads are great for long-form (more than 10 minutes) content. Overlay ads are laid onto the streamed video while pre-roll ads run before the streamed video. From an ad unit standpoint, I think all the ingredients and technology for a great user-experience are already there.

For example, we are seeing more publishers adopt overlay ads for their short-form content because they see less user drop-off, which makes for a much happier audience and more revenue for them. Pre-roll ads have their place as well. For a user watching a TV episode online, having a pre-roll that is 15 seconds long is a fair user-experience if it is a valuable piece of content. After all, this is already what all users are accustomed to on television.
Now, we believe the major problem to be solved is ad targeting. Ultimately, the biggest factor in ad effectiveness is whether the user finds the ad relevant. This is a tough problem to crack for online video because it is difficult to figure out who the user is and what video the user is watching. This is the problem that we have solved, using our proprietary technology.

What is the basic idea behind the algorithm to target ads? Also, doesn’t this weigh down streaming?
It does not weigh down streaming at all. If our servers were to go offline, the streaming doesn’t get affected at all. This is because our code works in such a way that it is not in the data path of the streaming content. The basic idea behind the solution is understanding the video content. We index the video using audio and speech recognition, visual analysis and metadata crawl. We then use a lot of machine learning-type algorithms to figure out what the video is about.

Machine learning is a type of advanced statistical technique that allows us to relate a word or a set of words to other sets of words. For example, for a word like ‘Kobe Bryant’, which is a proper noun and meaningless to a computer, that we might see in the metadata of a video, our system is able to correlate that to other words like ‘NBA’, ‘basketball’, ‘sport’. This allows advertisers to target their ads granularly and this is all done without any human intervention.
The beauty is that we don’t touch the actual content delivery. For example, publishers stream their own content. They do not need to change anything they do. All they have to do is integrate a few lines of code from us into their video player. We then deliver the ad content into the video player through our own content distribution network.

In Malaysia, in-stream video ads are not really picking up, partly because they are still new. What about the US? Which countries are seeing growth and which are not?
Well, all ad units take a while to get adopted. In the US, online video ads are now getting more mainstream. They have gone past the experimental buy stage to becoming a part of many advertising budgets now. I think the US and major EU countries are leading the adoption cycle.
China is lagging behind, not just in video ads but most online ads despite the large volume of inventory for a few reasons. In general, online advertising hasn’t really caught on in China unlike in other large markets. One reason is that the ad agencies there are still going through the education process of how online advertising can help clients.

Another is that most consumers in China do not have credit cards unlike in the US and Europe. A lot of online advertising, especially on the direct response side (think Google Adwords or any search-based marketing), predicates the consumer owning a credit card and buying a product by clicking on the ad and going to the website to complete the transaction. The fact that credit cards are not widespread in China (as a percentage of its overall Internet user population) might be holding back online ad spend.
Japan has a more vibrant online ad market. Unlike other countries, where the ad agency landscape is fragmented, the one in Japan is actually very concentrated. Dentsu, Japan’s largest ad agency, owns about 70% of Japanese ad spend market share. So, for anything meaningful to happen in Japan, Dentsu has to buy in.

Which Southeast Asian country is witnessing a high consumption of online video streaming relative to developed markets?
I think online video consumption is growing fast in Southeast Asian countries. For example, in our network recently, Malaysia’s video viewing traffic moved up between 20th and 30th place in terms of volume. The US is at  No 1, followed by Canada, the UK and other EU countries. Previously, Malaysia was nowhere on our list. We have seen that broadband penetration is the main growth factor in many countries. I suspect a lot of it has to do with increasing broadband penetration among Malaysian consumers in the past two years.

Industry observers have often said Malaysian entrepreneurs do not think global enough. After meeting some of  the Malaysian-based entrepreneurs, what is your general impression of them or of our technology start-up ecosystem?
I have some strong opinions on entrepreneurship and can probably fill a book with them. I think there are a few factors that encourage or discourage entrepreneurship:
•    There has to be an ecosystem of capital sources, readily available human talent pool from both the technology and business sides and lucrative exit sources in the form of M&A or IPO.
•    There has to be a successful ‘initial wave’ of companies where aspiring entrepreneurs can witness their journey to success and learn how to scale a start-up. These initial success stories are extremely important to foster subsequent generations of entrepreneurs. A good example is PayPal, whose alumni have created a host of successful companies like YouTube. But this problem goes beyond Malaysia. Even in Boston, we have frequent debates among entrepreneurs and venture capitalists (VCs) on how to make Boston more like Silicon Valley. Ultimately, these ‘feeder’ companies represent a set of people who have been there and done that. They can help grow the next generation of companies. Remember that it took Silicon Valley decades — 40 years or so — to get to where it is today, where the initial wave of success stories included the likes of Intel. That’s why I think AirAsia is such an exciting and rare story and one of those potential feeder companies for Malaysia.
•    Where Malaysian entrepreneurs are concerned, I don’t think it’s so much about global thinking per se. Maybe it’s more a lack of bravado and confidence. But I think a lot of times, confidence comes from having done it before or having seen it done before. I think these are basic ingredients if you are trying to foster a start-up mentality. I don’t think Malaysia is any different from other countries or locations that are trying to kick-start the start-up culture. While private capital in Malaysia is scarcer than in the US, it becomes imperative for the government to step in to fund the right opportunities. The best way to implement this is if the government became a limited partner but allowed the funds to be run by professional investors. The fund should not be government-run but private, with the government as the majority limited partner.
•    It’s important to be passionate and not have a culture of just following the money, which unfortunately is prevalent in Asia. It’s a cliché but nowhere else have I seen this work better than in start-ups. Winning start-ups are always staffed by people passionate about what they do. Even if they don’t get paid a sen, that’s pretty much what they would still be working on. Sometimes, it’s important to ignore history. Sometimes, it pays to go your own way. Google started when everyone who was an expert thought search was a dead game and everyone was doing the web portal strategy. Facebook and MySpace started after many social networking companies failed in the late 1990s and no VC wanted to touch these until they saw growth.
•    Finally, to be successful, we all have to take the long view. There is no overnight magic bullet even if you look at Silicon Valley. This is about changing the culture and it takes decades, if not longer, to do it, even in the best of scenarios. I think a lot of people fall into the trap of being impatient or setting false expectations where they think change can happen in a few years. If it doesn’t, it is declared a failure. In Malaysia, this touches a lot of areas of day-to-day life, from the availability of high-quality tertiary education, basic infrastructure, funding sources and setting public examples by fostering open competition to a vibrant exit market. Currently, I see China’s listing market becoming  increasingly more vibrant for high-growth foreign start-ups to raise capital.

Comments: [email protected]
This article appeared in [email protected], the technology section of The Edge Malaysia, Issue 775 Oct 5 - 11, 2009