Business Plans are characterized by low shelf life. More so, if the authorship is done on a paper napkin in a Mumbai coffee shop. Pubmatic is case in the counter-point. The 10 year old plan of making ad servers is still alive \u2013 and kicking ass. So what makes the company tick?<\/p>\n\n\n\n
To start with the founders. The product idea was probably from Amar Goyal, an HBS alumnus and serial entrepreneur, based out of California. He roped in Mukul Kumar \u2013 an IIT KGP engineer \u2013 with solid tech experience in Veritas. Mukul was a big fan of Paul Graham\u2019s articles \u2013 and was looking at starting off on his own. Btw, Paul Graham, is the brains behind Y Combinator. (Must google for his articles some time.) Paul\u2019s article that caught the fancy of Mukul \u2013 \u2018How to make wealth.\u2019<\/p>\n\n\n\n
What is an ad server? It is the gizmo that is responsible for the display ads that you see on websites. Pubmatic\u2019s primary customers are the Content creators \u2013 the publishing houses, The New York Times and the like. Though the company has also developed tools for media buyers in ad agencies, the bias has continually been to look at problems from a publisher-centric point of view. This has helped in maintaining focus.<\/p>\n\n\n\n
In 2007 when the company started, there were 300 ad networks that existed in the US. It was difficult for an ad-agency to pick and choose which one they wanted to work with. Pubmatic\u2019s first product helped the publisher to optimize their ad yields \u2013 by using ad networks which paid the highest. The product also helped the publisher with decisions like ad space, color et al. The rates then hovered around $ 1 per thousand impressions.<\/p>\n\n\n\n
In 2009, Pubmatic came up with a product that changed the market dynamics \u2013 Real Time Bidding. Earlier the ad rates were static \u2013 and with RTB \u2013 you could change rates every second. This product soon made the ad-networks redundant. Product churn continues to happen. Soon machine learning was being used to work out pricing decisions. In the last one year, the display ad market shifted significantly towards Mobile. So Pubmatic now designs the ads first with mobile, and then later on works out the desktop version.<\/p>\n\n\n\n
A very interesting discussion on metrics happened. Mukul mentioned that when Pubmatic started the key internal metric was number of customers. That soon changed to number of ad impressions. Pubmatic\u2019s servers today serve 50 billion ad impressions every day! With click-frauds galore, the metric was later revised to sell-through rates. The top management at Pubmatic takes a hard-look at its internal metric every quarter. The current one, btw, is cash-flow!<\/p>\n\n\n\n
Being a product company, one of the challenges for Pubmatic has been feature-management. At a macro level, Pubmatic has to decide whether a feature will help them move revenue from X to Y. Mukul has a very simple definition for an excellent product \u2013 It is one that a customer likes! Pubmatic focuses on the needs of its top customers \u2013 and add features accordingly. I asked Mukul about features deletion \u2013 and his somewhat diplomatic answer was \u2013 that once added, it is a very difficult task to take away a feature.<\/p>\n\n\n\n
The products are re-architectured once every 24 months. The first re-architecture involved re-writing the ad-server code into C. There were latency issues involved in the Java platform that was used earlier. Sometimes when not financially viable, some products are also scrapped. An interesting feature related example he gave was Drop Down lists. The disadvantage of using standard off the shelf modules is that we always start off by looking at Afghanistan on a Country Drop Down List, no matter which country you are located in!<\/p>\n\n\n\n
Mukul recommends that the Alpha should be up and running in about 3 month\u2019s time. An Alpha is what you can show to the customer for feedback. He gives another 6 months for the Beta. His definition of Beta being \u2013 going to the customer \u2013 and saying now pay up! Amar had previous experience in fund raising. This helped with the Series A funding of 7 m $ that came in during the Beta stage.<\/p>\n\n\n\n
Till that time the company was bootstrapped out of Mukul\u2019s house \u2013 with a small 4 person team. Most of them had been students of his second partner, who is also based out of India. He was teaching at Fergusson at that time \u2013 and these 3 joined fresh out of college. Of the 4, 3 are with Pubmatic even today. At this stage, funds came in from the F&F route (Family and Fools). The narrative changed soon enough though, Pubmatic has till date raised 63 m $ in total funding. Mukul feels that their investors have helped significantly in hiring and networking.<\/p>\n\n\n\n
His life-gyan for new entrepreneurs \u2013 Keep Learning!<\/p>\n","protected":false},"excerpt":{"rendered":"