E-Commerce Done Right: Five Keys to Snapdeal’s Success

As e-commerce continues to rapidly expand on a global scale, the numbers reflect just how powerful this market has become. In the U.S., online marketplace eBay saw $175 billion in sales in 2012 with all signs pointing full speed ahead. On the other side of the world, Internet retail in India accounted for $3 billion in 2013, and is expected to reach $22 billion in the next five years.

Capitalizing on this massive growth opportunity is Snapdeal.com, India’s largest online marketplace.

Snapdeal Stats

  • 20 million subscribers
  • Ships products to 4,000 cities and counting
  • 4 million products for sale on the website
  • Up to 10 million users per day, and 17 million per month

With 500% growth in 2013 and revenue projected to exceed $1 billion by 2015, Snapdeal is what many people would consider an overnight success story. Through a combination of solid technology infrastructure, a collaborative working environment, and allowing engineers the freedom to take risks, Snapdeal has been able to grow to its present size in just over two years, with a 100-member team.

In this webinar, Snapdeal Vice President of Engineering Amitabh Misra shared five keys to how his company has been able to achieve its incredible success in a short amount of time:

  1. A Services Oriented Architecture: What has allowed us to run the business, and grow ten times in two years, has primarily been the technology that we built. Our entire technology infrastructure has been built in house. It’s custom-made. That’s a mighty feat considering that we are a full-fledged e-commerce site. The vast majority of this technology has been built in two and a half years by 120 engineers. One of the primary things that allowed us to do this is the architecture. We have a fully distributed service-oriented architecture made up of mostly standalone subsystems.
  2. Engineers as Recruiters: One of the biggest challenges in building this complex system very fast is building a team that is able to deliver. We grew from 20 employees to over 200 in a little over two years with a very young field of engineers. Hiring in India is incredibly difficult because a lot of the good engineers leave for Europe. So, we did a couple of cool things. First of all, we have an HR team that specializes in technical recruiting, and are themselves engineers. We also have managers who are very hardcore techies, a lot of them are engineers too. That helps attract people because they look forward to the learning experience, and they know they will make a real contribution to the company. “We are able to convince people that once they join a company like Snapdeal, as opposed to a very big multi-national company, what they are building directly affects their life and the lives of people around them”.
  3. Engineers as Entrepreneurs: Every team member is encouraged to be an entrepreneur. Teams do not own technology deliverables, they own business metrics. Metrics as in how many clicks, in how many scrolls can people get to the search results that they’re looking for? Things of that nature, things that directly impact the business.
  4. Standups: We align business priorities in a very effective manner. One of the key things that we have done to make it more effective is the interaction between teams is very informal. All of the decision making in terms of technology, road map planning, decision making, what needs to go in, what doesn’t need to go in, how much you get from other teams – all of this interaction is very informal. A lot of the meetings happen in the corridors. We stand there, make a decision in five minutes flat, we make the call, then we move on.
  5. Divide and Conquer: Each team has complete freedom on the technology choices that they make in terms of adopting a technology that appeals to them. A very small team, like two or three people, can rapidly explore a whole bunch of technologies that are out in the market and then apply them with very effective results.

Relief for Snapdeal Growing Pains

Last year, Snapdeal took a look at their foundational technology and realized that it was not keeping up with the rapid pace at which their business was growing. Snapdeal initially deployed 10 MongoDB NoSQL database servers with 5 gigabytes of data in DRAM as a cache in front of MySQL. When they saw that response times were pushing a full second, they decided to look for a robust database that could handle millions of transactions with minimal latency.

Amitabh shared:

  1. Our primary criteria was that the technology should be able to support many reads and writes concurrently, and at least ten times more than what our requirements were then because that’s how fast our business had grown. We didn’t want to look at another technology for the next year.
  2. The second criteria was that while it handles these concurrent reads and writes, while performing at 5 millisecond burst response times, we wanted that to be the case 95% – 99% of the time. So while many technologies were able to give an average response time of sub-15 milliseconds, when it came to the last 10% – 20%, it sometimes very rapidly degraded. For us, the second criteria was very important.
  3. Finally, cost-effectiveness was important. Cost-effective, both in terms of if we were to get support for the technology, as well as the cost of hardware. We evaluated a bunch of SQL technologies and finally voted on Aerospike because it gave us the best consistent performance.

Watch the full webinar:

Read the Snapdeal Case Study