Online retailer The Iconic has experienced a massive surge in growth since it exploded onto the scene in 2011.
Buoyed by a total $100 million in various funding rounds since launch, the company attracted attention by being one of the first fully online Australian clothing retailers, and by offering fast delivery and a big product catalogue.
Australians flocked to the site - evident by its strong social media following and claimed 50 percent year-on-year growth - prompting the company to earlier this year shift its website onto AWS to address business growth that was happening faster than its existing infrastructure could scale.
Part of its value proposition is an “effortless and enjoyable” shopping experience for its customers, and customised offerings that put what the shopper is most likely to buy right in front of them.
But you can’t do that without good use of data.
Until just a few weeks ago, The Iconic was using Amazon’s cloud-based Redshift for its data warehouse, but the overall approach was not up to scratch, according to head of data and intelligence Brent Maxwell.
“The problem with the way it was working before was [Redshift] hadn’t actually been designed, we hadn’t designed a data warehouse,” Maxwell told iTnews.
“People just dumped all of the data from our systems, from the website, the ERP system into the one place.”
The data was also not available to the wider business - staff were entirely dependent on the business intelligence and IT teams to interact with the data and extract it.
This meant the company was largely reliant on Excel spreadsheets, and also resulted in mismatched numbers arising from different employees calculating things in their own way.
“There was a trickle of data coming out of the small amount of techs that could query and extract it,” Maxwell said - hindering use of the information by a self-professed data-driven company.
The goal was to open up the data for the entire business.
Maxwell studied other data warehouse options on the market, but decided Redshift was still the best technology for the job.
What needed changing was the way the data was presented to the business, and how the data warehouse interacted with the company’s other core systems.
Maxwell’s team built an extract, transform and load (ETL) tool he calls the Historian, based on the Pentaho software, to reach up out of Redshift and into the store’s systems to pull in data.
The difference between the Historian and other ETL tools, according to Maxwell, is it pulls data in dynamically based on metadata.
“If something changes in one of the upstream source systems, it automatically figures it out and imports in intelligently,” Maxwell said.
“Instead of saying ‘pull the data out of the website database and put it into this table in the data warehouse’, we say ‘look at the metadata and dynamically construct the query that pulls the data out of that table’.”
It means the business can get near real-tme data in its data warehouse.
Businesses generally prefer nightly loads to avoid impacts on source systems during peak times, but Maxwell says doing more frequent loads means each individual load is smaller, reducing the impact on the performance of the system in question.
However, real-time data means nothing if it can't be visualised to and used by the wider business.
Now, once data from The Iconic's systems is in Redshift, a star schema classifies the data for the newly-installed Tableau business intelligence tool to present it into easy-to-consume pieces for the business.
Tableau has the added benefit of reducing the strain on the BI and IT teams, Maxwell said, who were previously the sole channel through which the business could access any data from Redshift.
“It means BI can make a decision about what needs to change in the data warehouse, and they can change the star schema or they can liaise with the developers to load new systems when they come in,” he said.
Maxwell’s also created an “analyst guild” - a collection of people across every division of the business that have licenses to use the software and can serve as the entry point for their department.
The company’s next target is personalisation: using the pool of data at its fingerprints to make the online shopping experience as personal as possible for its customers.
Part of that initiative will involve the implementation of a new tracking tool for user behaviour, using Snowplow analytics. The Iconic had been using Google Analytics but found it didn’t provide the rich customer information the business was after.
“We’ll be using that to help understand how our customers interact with the website and app,” Maxwell said.
“The second that’s implemented it’s massively important for us to get it into the data warehouse because it’ll provide a huge amount of information to make much better decisions about what our users want.”
The Iconic currently segments its customers into gender groups, but wants to be able to provide a more detailed offering.
Maxwell is looking at segmenting into categories like age and shipping location, potentially even integrating with information from users’ Facebook profiles.
"If we can have more detailed segments we can give people a more customised website and app experience,” he said.
“Personalisation can be done in so many different ways, we need to make some careful decisions about exactly what we’ll focus on.
“Our target market is 25-35, we’re basically looking at product recommendations based on age, shopping type (browser or hunter) - you need to present those people with a different experience.”