What The Iconic's data and analytics chief is doing to drive stronger business decision making
- 26 February, 2018 07:06
Achieving significant productivity wins via data-driven automation and unifying business insight into a single source of truth are just two big priorities for The Iconic’s analytics and data science chief.
Kshira Saagar joined the online retailer as head of analytics and data science six months ago and is tasked with transforming the way the company imagines and uses data. He boasts of more than a decade in analytics and decision sciences, working across Australia, the Americas, Asia and Europe for Fortune 100 clients in retail, telecoms, insurance, media, healthcare and logistics.
More recently, Saagar was with Fairfax, where he was responsible for institutionalising data-driven analytics across the media group’s core competencies and building next-generation analytical products.
Today, Saagar oversees a federated team of about 20 data scientists at The Iconic. His first decision was to unite staff members operating within different divisions into one team under his leadership, with the common aim of using data to improve business decision making.
Saagar was presenting on how he’s re-imagining data processing and warehousing and bringing agility and scalability into The Iconic at ADMA’s Data Day conferences in Sydney and Melbourne in February.
“At the Iconic, the data team is at the crossroads of the business, catering to every department of the company,” he told CMO, noting he has two people working with buying team exclusively, two in warehousing, two in customer support, two in finance, two in marketing, and so on. “The idea is people belong to one common team, which is my team, but work in a federated model.
“There was no one in my team when I joined. Everyone who raised their hands to say they wanted to be an analyst was christened an analyst, or ‘superhero’ as I call them. By creating a common team, we ensure they upskill and are working with the right people, and secondly, that we’re not solving the same problem in different departments.”
The second priority has been unifying data and reporting into a ‘one database, one platform’ model.
“We currently have a transaction database, plus an enterprise platform largely used for marketing, and so on. Almost all these cause arguments on where data comes from and what it means,” Saagar explained. “We wanted to bring in an abstraction for the wider business so that everyone sees these numbers coming from only one source.
“The way to do this is to get data just from one place. That’s what we’re trying to do with the ‘one database, one platform’ model – we’re removing any discrepancy or sliver of doubt on where these numbers are coming from.”
The new approach sees everyone in The Iconic business logging in via one portal to access real-time numbers on sales, performance, financials and more. The Iconic’s unified data insights offering is being built in-house and based on open source technology, including the Apache Superset business intelligence Web application and Apache Airflow computational workflow modelling tools.
“We’ve already set it up in one form, for example with key metrics on where sales are happening right now, in this second, how long it’s taking to sell goods, and so on,” Saagar continued. “The key numbers are already there and we were able to do it because everyone in the business is looking at just one source of data.”
To get buy-in, Saagar said he had to ‘kill’ everything else already there. “We removed the oxygen on some things, which mean people started asking where the numbers have gone and how they do it now. So then you give them one platform,” he said. Saager and his team then worked to provide training and data-driven use cases relevant to each department.
The Iconic also maintains five ‘star’ levels of data access. For instance, everyone can see revenue but very few people can see profit and EBITDA.
“You don’t have to download software, applications, or buy a licence, you just go online and log in. It’s minimised friction and there are no excuses not to use it,” Saagar said. “Once users log in, they quickly get data and see the value.”
Data in action
To get some quick wins, Saagar turned to productivity and efficiency gains.
“In marketing for example, we’d produce lots of reports every day. As the social media team, they might then apply rules and logic and ends up spending 6-7 hours per week pulling numbers manually. What we did is automate the approach, so they just log in to get those numbers,” he said.
“We did the same thing for our buying team – they’d get reports and painstakingly produce cut and paste in images and numbers into another report, which they take to suppliers to see which brand to buy and do. We automated that and now they don’t have to have three people doing this job.
“Simple things like that which takes up a lot of time helps you get buy in. And it means these employees can do more meaningful work, as they’re not pulling numbers, they’re digging further into them.”
To showcase how data is being used at the point of interaction with customers, Saagar noted The Iconic’s work to firstly personalise email, then broader digital marketing activity via a personalisation algorithm.
“We started small and personalised email. We worked to understand user purchase and visit behaviour, the brands they looked at and what she might potentially like. We then make recommendations with zero understanding of fashion. We built that for email, as it was a quick win that could be applied on existing platform,” Saagar said.
In control tests of personalised versus non-personalised emails, The Iconic has seen a 20-25 per cent lift in clickthroughs and open rates on average. That means more than 25,000 are more likely to click if personalised for them, and 14 per cent are more likely to buy.
The algorithm has since been adapted to personalise SEM marketing and display retargeting.
Getting your data foundations right
For those still struggling to get anywhere near such data utilisation, Saagar said there are two stages to pass through. The first is getting the data in one place, the second is turning data into insight. The problem is too many companies are still stuck on the first step.
“Most people in BI and analytics teams are fighting to get data into a shape that they can use. By the time they have they’re tired and don’t want to do more stuff or go beyond it,” he claimed.
“To then enable data service as a successful unit, the first step is automating a lot of stuff that takes a long time. There’s no excuse and nothing to hide behind. People in the business and team then spend more time looking at data, seeing if something is or isn’t right, and asking smarter questions, instead of producing retrospective reporting.”
It’s this future-proofing that Saagar also believed machine learning and artificial intelligence advancement will deliver.
“Everyone in the industry is starting to use data but it’s a euphemism for looking at the performance for the last five weeks. If you could see what data could look like in the future, you can start making much smarter decisions about what to do for the business. That’s where AI/ML will make a big impact – it will help people understand what possibly can happen and which levers you can pull to make these good steps happen,” he said.
But again, the key is getting the data foundations right. “Either the data is not trustworthy, or these companies don’t have a basic understanding of what or how much revenue they make on a day by day or hour or hour basis,” Saagar said.
“I doubt there are many Australian mid to large sized organisations that don’t have their finances in order – they know how to take purchase orders. If you start with the right foundations, you can start digging deeper to see where you can make more money and that’s the same story with data. If you have the foundations right you can dig deeper and do more with your data, even look into the future and see what could happen.”
With the arrival of privacy laws such as the Notifiable Data Breaches scheme in Australia and GDPR in Europe, it’s never been more important to get the corporate data store in order.
“A few months ago and in line with GDPR coming in, we started to look at Iconic not at clamping down access to data, but understanding how far and wide data is being accessed, an what tools and systems it’s flowing into,” Saagar said.
“The exploratory exercise was like creating the map of our data world. Clamping down or modifying access will be a cake walk once you’ve done that.”