SINGAPORE (Aug 13): Bill Lee, a former business consultant at the now-defunct accounting firm Arthur Andersen, has seen more than his fair share of data analytics companies. Many of them, run by data scientists, propose to help companies manage their businesses with the help of better data. The results can sometimes be disastrous.
Lee gives the example of a university that wants to optimise its teaching staff’s class schedules so that they do not have long breaks between each class. On paper, this sounds beneficial for both the institution and staff. In reality, most teaching staff would resent back-to-back sessions or find it difficult to adhere to a schedule that is too tight.
“[Successful implementation of data analytics] depends on the art and not the science,” says Lee. “When science produces the output, how do we use the output and implement it in such a way to help the organisation progress and not panic?”
At Azendian Solutions, Lee aims to do just that. The data analytics start-up, where Lee is regional managing partner, aims to marry the hard science of data analytics with the art of implementing a model that will help drive a company’s success.
“Organisations are made of human beings. [Data] models are perfect because of mathematics — one plus one will always be equal to two. However, human beings are not perfect. I cannot measure a day and say a person takes only 45 minutes for lunch, goes to the bathroom only three times for not more than a minute and 45 seconds, and takes five minutes and 30 seconds to change a brake pad,” says Lee.
“If you follow that tightly, the model will calculate that tightly for you. But three to six months down the road, you’ll see your attrition rate go up because the human cannot take it. Humans cannot follow the hard regime that the model crunches out,” he adds. Lee says it is better to sacrifice up to 15% productivity than to suffer a high attrition rate.
Nothing new under the sun
Lee founded Azendian with a group of friends three years ago, and attributes the company’s success so far to the diversity of experiences and opinions in the group. “Each of us has different strengths and we all rely on each other,” he says.
Azendian is focused on solutions for educational institutions, hospitals, residential estates and companies in the aviation, maritime and transportation industries. One of its most popular products is a smart campus solution that helps universities optimise teaching staff’s hours and predict when a student might need academic intervention, among other predictive solutions.
“We do not invent new solutions to solve new problems. The solutions we have use new techniques to solve some of these current challenges much more productively and effectively,” says Lee. “We also do not believe in hiring a whole bunch of engineers to build a product and hoping someone will buy it.” Instead, Azendian looks for partners that want to use data analytics to find solutions to specific challenges. “We develop a solution together with them after which the alpha or beta version of the product is ready.”
At times, Azendian’s models might show that companies have more than enough manpower for their needs. However, Lee says he typically advises clients not to retrench employees. “In any system, there is already a natural attrition rate. The advice we always give to our clients is let the natural attrition rate manage the headcount,” he says.
There are also certain lines that Azendian does not cross. While data scientists typically prefer to have as much data as possible, the use of some data to generate models would raise ethical or moral concerns. “In this part of the world we’re living in, with multi-racial and multi-religious contexts, when you build a model, you have to be sensitive. I always tell my managers and data scientists: What’s legal may not be ethical, so be careful. Your ethical code should be tougher than what [the Personal Data Protection Act] requires you to do, otherwise it might come back and bite you at the end of the day,” says Lee.
“Simple example: We can actually predict staff performance. But when you build a model of employee performance, do you put in race and religion as one of the variables? I don’t think it’s going to be helpful to see race and religion as a predictor of performance, as there’s nothing you can do about it. As a CEO, I would rather not know about it.”
Azendian received $4 million in funding from Singapore Technologies Engineering’s corporate venture capital unit, ST Engineering Ventures, in March. The choice to take money from the locally listed engineering firm was a deliberate one, according to Lee.
“When I was looking for funding to grow the business, I actually received a better valuation from other organisations. But we decided to go with ST [Engineering] because it was a very good match. They need people with data analytics skills, and data analytics does not live in a vacuum. We need data and data is generated by systems, and they are very good at that. They are strong in building sensor and [Internet of Things] networks, engineering solutions and enterprise systems, all of which use data that I need,” says Lee.
Azendian currently has over 25 customers utilising its data models to optimise organisations. The start-up has offices in Singapore and Kuala Lumpur, Malaysia, and has plans to expand to the rest of Southeast Asia in the next two years. It has identified Thailand, Indonesia, the Philippines and Vietnam as potential markets for expansion.
“We believe that the rest of Asia holds a lot of promise for us also,” says Lee. “One of the key reasons we started this company is because we believe that an Asian company can be in a hi-tech area and be successful.”