Codeinks – a use case in providing the best insights to customers

Profile picture for user snambiar By Sheila Nambiar February 27, 2019
Codeinks provides an insight into the power of insight using Oracle's Autonomous Datawarehouse.

In today’s highly competitive world of modern retailing, any insight into customer behaviour and the performance of products and campaigns is invaluable. It can make the difference between a shopping cart being abandoned and someone checking out.

In online, while offering lower prices, a typically broader product range and convenience to drive sales, increasingly retailers are recognising the need that they too have to go beyond the norm and boost the shopping experience – being online alone is not guarantee of success.

So what if you could actually predict how a particular campaign might perform before it even runs?

This is the kind of insight that Bangalore-based, e-commerce software developer Codeinks wants to provide to its customers. And it is working with Oracle’s Autonomous Datawarehouse (ADW) technology to help bring that vision into reality.

Codeinks is the developer of Cartvines, a cloud-based, ecommerce platform that enables B2B and B2C clients to set up an online store in just minutes, and take advantage of a full set of tools and dashboards for campaign reporting insight.

According to Codeinks co-founder and CEO Manish Verma, the future success of his company will depend on being the best provider of insights to customers:

Any company that is not spending a significant portion of the time innovating and creating business value for its customers is going to go away. Customers want highly-analysed and processed information for quick decision making and what-if analysis. What we are trying to do is get into deep levels insight into user behaviour and past performance to do future predictions of trends.

Getting down to the nitty gritty

So in mid-2018 Codeinks began working with ADW to derive better insights from campaign performance, down to a granular level around categories and even individual products, with the aim of using this to predict the outcome for future campaigns. For instance, if a client was running an e-commerce portal selling shoes, it might like to know if offering a midnight sale offering a 5 percent price reduction on a specific brand of running shoes might drive a surge in sales.

Verma says that previously Codeinks had been prevented from achieving this level of granularity due to the sheer volume of data generated in campaigns. Investigating the performance of a particular product might mean analysing millions of page impressions and even greater number of attributes to deliver specific information on each product:

Doing this with a traditional database takes a lot of time, because of the need to link large impression volumes with actual transaction value to see what happened. We wanted to see if this piece of technology would impact performance or manageability in any way.

Impressive results

So Codeinks began feeding historical transactional and behavioural data into the autonomous data warehouse, and the results were impressive. Average improvements on performance of between 30 to 50% were achieved; even in some cases up to 450%. Verma says Codeinks also noted a benefit in compression of between 20 and 40%.

Codeinks then created dashboards that could quickly show the results of what-if analysis around a range of factors including session data looking at user interactions in a particular timeframe or around particular tasks. Verma says:

We could determine the probability percentage of a conversion by the time spent on the page, by product. If you were to purchase a shirt you may take the decision in less than a few seconds, but if you were looking at a phone and you were comparing different versions of it, maybe you need more time. And we were seeing that even for a particular brand the time spend for a conversion changes.

Aiming to get predictive

Verma says the goal is to mash up large volumes of data to deliver better insights to retail customers, which can be used to provision campaigns on-the-fly. The long term goal is to create a predictive tool that can be offered back to Codeinks customers:

We will get to the stage of being able to position and target dynamically over the time slices and population visits, and start collating all information and analysing the positioning it back to customers.

Over time Verma believes the AWD technology will enable organisation to become more customer-oriented and leave the tasks of optimisation and fine tuning to the platform itself:

Those kinds of activities will go away pretty much, meaning businesses will be more centred on business value creation rather than managing operations.