Rocket Science in Click-To-Call

You have used the magic of Click-To-Call if you have ever been at the receiving end of food deliveries or deliveries of goods from e-commerce companies. The delivery boy calls you when he is at the doorstep. He clicks your name on his mobile app and somewhere on the web, our servers call two numbers - the delivery boy's and your's. And put them on the conference call. You talk with each other but none of you know each other's numbers. Instant connection and instant privacy!

However, when you are doing it a million times in a day, it becomes a little complicated. A whole bunch of things happens when the delivery boy clicks on that app - dealing with our servers and making phone calls through our archaic and now fast consolidating telecom networks. It does not always work as expected. This means if it does not work, you will have to figure out alternate routes. If it works, you have to make sure that the connection quality is good enough.

If you do this several million times a day, you will discover thousands of ways this process does not work. Marrying telephony with the web is not that easy.

But the magic of the high-quality products lies in doing seemingly simple and straightforward things millions of times with minimum failures. What do we do? This is when our engineering comes in.

The Knowlarity tech team - Ajay, DP and the engineers - went through at "Industry standard" 4 seconds Click-To-Call to a 400 Milliseconds Click-To-Call in the last 30 days. You shave off half a second from those queues, another second from databases and then optimize the error rates and suddenly you uncover another set of problems. You are peeling layers and layers off the onions one at a time. The part that is in your control - your software - is easy. What is difficult is dealing with the telecom infrastructure as you do not control it. But a bit of push and pull, some ingenuity and some workaround, you find the way. You have to.

This must be the journey that repeated itself hundreds of times in making the fast and reliable cars we drive and super fast computers we use. These intense optimization efforts, that involve getting hold of that one number - the crash rating, the fuel efficiency and that number of transistors per inch - that must drive the engineering innovations and give the consumers these amazing products that we take today for granted. The lessons from these journeys are the most important part of building great products:

Incremental improvement is boring but that is what makes great products great. We do not want products that work most of the time. We want products that work every time. First 80% of the task takes only 20% of the time. Most of the efforts take the last 20% of the improvements. Everyone has copied Apple iPhones. But only very few have come close to the reliability and quality of the iPhones. It is easy to do 90%. It takes everything you have to do the last 10%.

Most innovative ideas look silly at first. The innovations are out-of-the-box. It is important to think free and think radical to see what will move the "output metric". Nothing should be out of scope. This is what gets the new breakthrough and really moves the needle on the "output metric".

Belief precedes rationality. You have to win in your mind before you win in reality. The same is true about engineering excursions. 400 Milliseconds is a very short amount of time. There definitely was a bit of back-of-the-envelope analysis for the right target but after the high-level analysis, it all went down to the belief in the team.

We believe that we will do world-class engineering innovations out of India. It is one firm step at a time.