Analytics India Magazine had an exclusive chat with Pratham Hegde who heads the Marketing Science unit for Facebook India. He has 18 years of experience in Data analytics and specialises in Marketing Measurement. In this conversation he talked to us about how data science and analytics is used in marketing measurement at Facebook and how the role of analytics has evolved over the years.
Before his role at Facebook, Hegde headed the Marketing Analytics Practice at Genpact, then moved on to set up a Marketing Measurement team for Amazon for their major global markets.
Analytics India Magazine: How is data science and analytics used in marketing measurement at Facebook?
Pratham Hegde: Data science is at the core of what we do, which is to quantifiably measure the impact of ads on Facebook on business metrics (brand or sales metrics). Hence, the tools and solutions we deploy for marketing measurement are grounded on the principles of experimental design and statistical significance of results — ultimately everything boils down to that. A lot of data science and research goes into our measurement products, they are built and tested to ensure they yield unbiased results and they can be deployed in a scalable manner. We also use a lot of custom solutions working with third party providers; we use advanced statistical models to understand the impact of media in driving business metrics, and isolating the impact signal from the noise.
AIM: How has the role of analytics evolved over the years in driving insights through data and consumer research?
PH: In general, advertisers are getting more sophisticated in measuring the impact of marketing activities on their business. But the journey is hard and requires patience. If you think about it, marketing has traditionally been thought of as a “right brain” centered function where one had to be creative for the brand stand out in the crowd and be noticed. That is no longer the case. With the rapid move to mobile and digital, marketers who have a flair for data and new technology are able to leverage available tools well and drive impact for their business, while those hanging on to just offline channels are still stuck at asking the question “does digital work?”.
However, in India the available tools for measurement are still at a nascent stage and the industry is still at an early stage of the journey is being able to quantify the impact of advertising. For example, the kind of closed loop experiments or panel based solutions that are able to generate single source measurement of TV and Digital are still not available in India. I find that in India we always ask the tough questions but the resource commitment and patience to get to the truth via hard measurement principles might not always be there.
AIM: How has analytics and data science become crucial to the social media industry? Would you like to highlight few use cases?
PH: I think the term “social media marketing” can mislead people. Paid advertising on social media is going on globally at an unprecedented scale and is set to eclipse many traditional channels in terms of both reach and ability to drive impact. I see a lot of folks are still depending on social channels to drive “social buzz” or “social engagement”.
At Facebook measurement we have many studies that prove that intermediate metrics like likes, shares, click don’t necessarily correlate to business metrics such as brand metric movement or sales impact, which is what we ultimately want to drive for the business.
So, we encourage advertisers to think of Facebook just like any other media channel and apply the same measurement principles as they would do for any other marketing spend.
The good thing with digital is that it allows you to precisely measure the impact that a digital campaign had (or did not have) in terms of the business outcome metrics. We have hundreds of examples for Lift studies done in India which are both Facebook and third party measured, that show positive significant impact using control exposed methods. In addition, we also have a number of matched market tests and mix models that provide a read on the direct impact that Facebook advertising had on sales. On the Performance marketing side, we have a number of examples that measured positive lift in add to cart, sales, app install and other outcomes of interest for advertisers. We also have examples where we used innovative ways to close loop and prove that facebook ads drove in-store sales for retailers.
AIM: What are the challenges you face in your current role? How do you overcome these challenges?
PH: In general, there is a lack of understanding of robust advertising measurement approaches, tools and principles in the Indian market. We find many advertisers who are wedded to offline marketing and hence offline measurement tools such as brand tracks that rely on face to face interviews to measure brand metric movement. There is a tendency to use methods that help make marketing look good Vs trying to apply robust measurement principles that measure the true impact of media ; that then help optimise marketing spends. So we spend a lot of time in just educating the advertisers and agencies on fundamental measurement principles, and taking them down the path of continuous test and learn measurement – it’s hard and requires a lot of patience but the light at the end of the tunnel is the ability to accurately measure the impact of marketing monies ; which in turn ensures that valuable funds deployed in marketing are generating adequate returns for the business.
The other big challenge that we face is to keep things simple. While advertisers are asking more complex questions that lead us to deploying more complex tests, it is very easy to get lost in the methodology and forget to translate the outcomes into simple business language. Ultimately if you cannot break the method and results down into consumable bytes that advertisers can use in decision making, the most sophisticated measurement experiment can become useless.
AIM: What is the roadmap for analytics in marketing measurement at Facebook?
PH: Our goal is to continuously build solutions and tools that help advertisers measure the true value of Facebook advertising. Since an absolute “Facebook only” measure can only go so far in helping the decision maker, we also focus on methods that help measure Facebook together with all other channels. Our long term goal is to provide every advertiser the ability to measure campaigns in self-serve mode. Our measurement tools rely on the unique “People based” platform and capabilities at Facebook, and are arguably the most accurate and best grounded in the first principles of measurement in the advertising industry.
Try deep learning using MATLAB