club catalyst. Measuring ROI of retail analytics

Big Data, fast data and the data deluge are taking the world of digital businesses, especially the customer-focused ones, by storm.

Retail is no exception as more and more customers leave digital footprints with every transaction, interaction and engagement at every retail touch-point — online, mobile, social channels, in-store and even contact centres.

Technology is enabling businesses to collect shopper data from these touch-points, analyse it and derive insights to make informed decisions, whether it is to provide a better customer experience, run marketing promotions or decide product assortment or gain tighter stock control.

However, the volume, variety and velocity of such shopper data is a great challenge and retailers are embarking on massive efforts to make sense and glean insights from it, to stay one step ahead of competition.

Retail CXOs today need a way to measure the impact of their ‘Insights’ initiatives and investments.

Let us take an example. A popular Insights initiative that many Tier 1 and Tier 2 retailers are investing in is around ‘Personalisation’ and delivering product recommendations to customers as a way to cross-sell/up-sell.

This is much easier said than done as core to this is a central 360-degree view of the customer based on the myriad footprints the customer has left in his engagement with the retail brand.

It is a task in itself to aggregate, sanitise and normalise such customer data from multiple sources such as clickstream, PoS transactions, social channels and call centre logs and create 360-degree profile views.

Emerging from such ground-level challenges and getting to build desired analytics capabilities and start deriving insights takes more time than anticipated.

What goes missing in this complexity is the need to track and measure the impact of the actionable insights that are being generated — either in terms of improved operational efficiencies or in terms of a sales uplift through better conversions.

What's measured is done Why should a retailer monitor and measure the impact of its shopper insights initiatives?

Because it can improve profitability and because what gets measured gets done! Monitoring and measuring the impact of data-driven initiatives against company-specific metrics such as gross profit, revenue and inventory carrying costs, can offer significant benefits.

A key finding of a survey of 350 retailers conducted in March 2014 by RetailWire, an industry think-tank, found that there was a direct connection between monitoring the impact of shopper insights and positive return on investment (ROI).

Companies that monitored the impact of shopper insights on their standard metrics reported positive ROI compared to ones that didn’t.

The survey also found that nearly two-thirds (64 per cent) of the survey respondent companies did not do any monitoring even though survey findings revealed that monitoring yields results. 69 per cent of those monitoring reported positive ROI, compared to only 28 per cent of those not monitoring. For better control, retailers should monitor three key areas when it comes to their investments in insights initiatives.

1. Health of the Analytics Operations: A typical Analytics Operations flows as below:

Data Aggregation -> Data Profiling -> Data Ingestion -> Processing in analytics core engine -> Outputs

Outputs can be in the form of

a. Insights that are consumed by business users: e.g. ‘People who added product A to cart but abandoned without checkout’

b. An actionable insight that is delivered to customer channels or on business applications: for example, product recommendations delivered on the website - ‘People who bought Product A also bought Product B’, ‘People who viewed Product A ultimately bought Product B’; ‘Send Push notification on 10 per cent discount for customers who have added Product A to cart but have abandoned it in the last 48 hours’

Processes at each point Monitoring the operational health is essentially about breaking down this flow and measuring the health of processes at each checkpoint — data aggregation, profiling, ingestion, processing in the core analytics engine to generate insights and the consumption of the insights/actionable by the different consumers of output.

2. Efficiency of the insights/actionable being generated: The efficiency of the insights and actionable that is generated can be measured with specific metrics that help capture the algorithm efficiency (the number of false positives, what was the model lift, what was the model accuracy), the impact on every channel (what was impact on the traffic, the average shopping time, conversion rate, effect on abandonment rate), the impact on the business users/applications (was the marketing team able to plan their campaigns better and has it resulted in better campaign click-through, was merchandising team able to forecast better and has it resulted in reduced stock-outs etc.)

3. Business Impact of the insights being delivered: This is the hard measure on the top-line or bottom-line impact created by consumption insights and application of actionable.

Whether it has resulted in a sales uplift, an increase in profit due to reduced marketing effort/due to reduction in cost of carrying inventory with better forecasting decisions.

Dashboard view A dashboard view that would help monitor and track these key performance indicators (KPIs) at the touch of a control is on the wish list of every CTO and CIO as it would help them justify the ROI on huge analytics and insights initiatives they are undertaking and build the case for more focused efforts in this area.

The key challenge we observe when it comes to monitoring such KPIs is that while for digital channels, any data is extractable with little effort, the physical stores continue to be a black box.

And the buck stops here when we talk about any attribution of investment — be it marketing efforts/campaigns/insights initiatives.

ANITHA RAJAGOPALAN, RETAIL & OMNICHANNEL CONSULTANT, HAPPIEST MINDS

COMMENT NOW