SurveyMonkey data chief: ‘Data science is about expanding horizons’

27 Oct 2017

Data power to the people. Image: SurveyMonkey

SurveyMonkey’s Mehmet Goker explains how data science is augmenting workers’ and organisations’ capabilities.

Mehmet Goker is vice-president in charge of data and analytics at SurveyMonkey, with his appointment part of the company’s data strategy and, specifically, its people-powered data vision.

Before joining SurveyMonkey, Goker was the first data scientist at Salesforce where, in his seven years there, he transformed the way in which the world’s leading SaaS company surfaces upsell and cross-sell opportunities, and drives customer success and revenue.

‘The truth is always in the middle. The truth only comes from merging both human and machine. Relying purely on machine data will never get you the accurate and entire picture’

Before that, he worked at Strands Labs and PwC, where he was research director at its Center for Advanced Research. There, he led a group of researchers, developers, and usability and subject matter experts to develop advanced tools and technologies.

Silicon Valley-headquartered SurveyMonkey employs 50 people in Dublin, where it focuses on customer operations. The company has also begun hiring engineers for a new product and engineering hub in the city.

SurveyMonkey data chief: ‘Data science is about expanding horizons’

Mehmet Goker, vice-president in charge of data and analytics, SurveyMonkey. Image: SurveyMonkey

What are your thoughts on how SurveyMonkey approaches data science?

In general, I’m a big believer that data science and any kind of data science application is there to support the end users of an application.

For us, it is not about figuring out the deepest thoughts of our users and trying to make millions of dollars – that is not the intention. The idea is always to find how we can assist a user, both in creating a survey by figuring out and helping them with what they should be asking and how they should be asking these questions; and helping them to find an audience to whom they can send these questions to.

It also means supporting the people who answer them, by making sure they answer the surveys that are actually relevant to them and that they don’t receive random stuff from people asking irrelevant things.

The ultimate goal is always to support and help the users and, of course, respect their privacy – that’s very important.

What is the data science behind SurveyMonkey’s people-powered data tools?

In my previous work, what we used to do was, we were able to access usage data of customers to understand from the meta data the nature of their transactions to improve adoption.

For example, we looked at how many account queries there were, how many times they logged in and all that stuff. We would understand if users were utilising our service online. But the problem with that is, there is always a situation where people are using a service a lot, but are not exactly happy.

So, what we figured out over at Salesforce was, you have to have a better understanding of what people are actually thinking and get direct feedback from human beings as opposed to only relying on pure data and insights coming from pure data.

The truth is always in the middle. The truth only comes from merging both human and machine. Relying purely on machine data will never get you the accurate and entire picture. You always have to bring the insights from human beings into play and that gives you that crucial 360-degree view.

In essence, what SurveyMonkey wants to do is, we want to help people leverage our experience with our vast amount of data that we have, and merge that and align that with the information we get from individuals to make sure that our customers, the surveys they are creating, the companies that we work with, provide that 100pc understanding.

How are businesses utilising SurveyMonkey?

We have 270 people answering questions every second. And we have 3m survey responses per day. There have been hundreds of millions of surveys created on our platform.

For a large corporation using it, for example, they need to think about it as if they have an employee feedback HR experience survey, for example. If you and I were to go about that, we wouldn’t know what questions to ask if we didn’t know what the best way of answering those questions was.

What we can provide is experience of which questions work the best in what environment, and from there we can provide that information to our end users in the form of catalogue questions designed to produce better results. But also, since we have it across industry, we can provide benchmarks for this and provide information about what similar companies are asking.

The other thing we can do is provide an audience for your surveys – for example, if you want to set up a new brand. If you want to understand what the market will think about the new brand, is it a good name for that brand, you can do market research. The key to market research is having a good audience to answer your questions and SurveyMonkey can provide that audience.

There are many other scenarios and use cases.

How do you think we will be working in the coming years? What will a data-powered job look like?

For me, data science and access to information is all about expanding your horizons – I’m going to refer to this as the information horizon. We all have horizons that limit our ability to view and understand things. All of the IT we have is there to make sure we can expand our horizons, to make sure that we can go beyond what we are able to do under normal circumstances and gain more insights.

SurveyMonkey will allow you to get insights from other human beings, making sure that you are well grounded in your decision-making and have the insights about what you should be going towards. In parallel, a similar thing is happening in search engines with new interfaces, from voice to VR environments, making sure you are getting more and more information.

The ultimate goal of the AI and data mining in the end is about helping us to process the information.

Our horizons are constantly expanding. The wider the horizon gets, the more data we have.

The ability to ensure we can make decisions on this information is pretty much limited. Our brains need to process this data. So, what we need are systems to help us understand all of this information.

Our task is to help people to expand their information horizons in an easy manner and, at the same time, helping them to interpret and understand the data and make the information they gather most useful.

These tools are still quite complicated. Do you think it will come down to data science being taught in schools?

The way I think about it is: data is a discipline, like engineering or computer science. What we will most likely do is put more fundamentals about how to think about data, how to interpret data, how to analyse data, into our education systems.

The core elements of machine learning and data science are, in time, going to become part of the normal repertoire of engineers going forward.

The challenge is, if you want to go really deep and have a customised focus, then you will have to go into more detailed technologies and be able to tune and modify the algorithms, so that is going to make it even more complex.

But there is going to be a broad range of applications and ways of using these technologies.

Even at elementary level, it is going to boil down to helping people understand how to interact with data, access data and interpret various ways in which data is represented.

Want stories like this and more direct to your inbox? Sign up for Tech Trends, Silicon Republic’s weekly digest of need-to-know tech news.

John Kennedy is a journalist who served as editor of Silicon Republic for 17 years