Beat the fear factor: retailers can become truly data-driven, says ThoughtSpot


By Davy Nys, general manager EMEA, ThoughtSpot

Nys: power to make informed decisions is crucial

Every retailer I meet these days insists they want to be data-driven, but the reality on the ground is very different. Depending on whose research you trust, adoption rates for business intelligence (BI) and analytics range from 21-32%. Some analyst firms actually report figures starting to decline. Given all the progress made in data analytics and visualisation technologies over the past decade, how can this be?

First let’s establish what data-driven actually means. There are various definitions out there, but essentially it’s about empowering all employees to make informed, optimal business decisions that are based on as much available, accurate data as possible. With 37 UK retailers having gone bust this year including long-established brands Jaeger and Jones the Bootmaker (Source: CRR), giving people in retail the power to make informed decisions has never been more urgent or important.

It’s not all bad. Some newer, online retailers are pioneers when it comes to mining and analysing customer data to offer effective, tailored promotions, buying recommendations and loyalty programmes. However, when I talk to more traditional retailers about data-empowering people like merchandisers, loss prevention managers and sales associates, I keep hearing the same refrain from CIOs: “Our people aren’t ready for that yet”. These, however, are the retailers that have the most to gain from being data-driven.

Deconstructing the paradox

Search-driven analytics gives answers in seconds, rather than weeks or months

It’s not until I dig deeper that I discover the CIOs’ deeper concerns. They tell me: “What if our people don’t ask the right questions?” And, “They don’t really get data” or “They don’t really get analytics”. These statements get us closer to the problem. Most experienced CIOs have multiple battle scars from trying and failing to drive adoption of BI and analytics software, some in organisations where users were fairly enterprise-tech savvy. Retail professionals are among the least likely to be skilled in writing SQL scripts, understanding how data table joins work and other esoteric data concepts.

These CIOs were basing their objections on their experiences with the ‘pipeline’ model for BI and analytics that is still dominant, but, as adoption figures suggest, starting to decline. This is the classic model where data lives in many different silos – sales, marketing, HR, finance and so forth. In this model, specialists focus on getting data ready for analysis, segmenting it into data marts for performance and then summarizing the data even further into cubes and views.

The paradox inherent in the pipeline model is that only data experts with very limited business knowledge have the technical skills to query data. Business people want answers to what they perceive to be very obvious business questions, but because of the way the data is structured, the answers can take days or even weeks to generate. Data experts in these environments typically only have the time to perform ad hoc queries for the most senior executives. Everybody else in the business has to make do with predefined ‘self-service’ reports that rarely contain answers that are specific or timely enough for most purposes. This explains why adoption rates are so low.

It’s little wonder that when I talk to experienced retail CIOs about a new search-driven approach to analytics that empowers any employee to ask questions of data directly, in real-time, without IT support, the initial reaction is extreme trepidation. They imagine massive costs, strain and chaos on IT resources. Given their experiences, who can blame them?

Can retailers overcome resistance to change?

This kind of systemic resistance to change can be difficult to overcome, but I am optimistic. One by one, I see traditional retailers like JD Sport in the UK and Bed Bath & Beyond in the US taking that leap of faith to try new approaches like search-driven analytics. As each retailer achieves success and shares their stories with industry peers, more will have the confidence to change. To those on the fence, let me dispel some common misunderstandings and objections about search-based analytics:

IT will get sidestepped and have to ‘clean up’ when it goes wrong – some CIOs assume that because a search-driven analytics approach gives so much power to business users, that it’s one of those dreaded ‘shadow’ systems that sidesteps IT altogether. Nothing could be further than the truth. These systems rely on IT to set up and maintain the data model, set access rules and provide ongoing support. Unlike with pipeline systems, however, data experts and business people no longer have to understand each other’s area of expertise to make the whole system hum.

People will get too much data access – we’re not ready for that – IT controls who can query what data, so there is no risk of this happening. In fact there is arguably less risk of this happening than with the pipeline approach. In the search-driven approach, business users are guided to ask allowable data questions. This overcomes the big fear CIOs have about people not understanding data well enough to ask the ‘right’ questions. When someone searches for a microwave oven in the books section of Amazon, Amazon doesn’t tell you you’re wrong, it suggests different departments where you can find results. Some search-driven systems are now augmented with AI that helps further guide users.

Analytics provide simple answers to simple questions

People are terrible at report-building the good news is that with search-driven analytics, people don’t have to build reports at all. Business users don’t necessarily always want reports or dashboards. Usually they just want simple answers to data questions like “How many cappuccinos and flat whites did we sell to loyalty card holders at our Notting Hill branch last weekend? Your data experts can still build dashboards for the top executives who need regular updates on the same key performance metrics.

We can’t afford the spiralling licence costs of giving analytics to everyone – fortunately it’s not just the analytics data model that’s changing. New vendors recognise that per-user licence penalise data-driven organisations by taxing adoption and are offering favourable licensing models that don’t charge per user.

But we’ll have to train people to use a whole new system – If any vendor tells you that you need to send your people on a three-day training course to use a new analytics system, show them the door. The whole point of these new systems is that searching for data answers should work in the same familiar, intuitive way as Google or Amazon searches. And intuitive means no training, instant adoption.

The bottom line is this: if retailers expect to become truly data-driven, they have to find some way to put business knowledge in the hands of every user. They will never achieve this by wishing it, limiting their efforts to customer data mining or persisting with the old pipeline-based BI and analytics tools and expecting employees to acquire technical data skills. The new search-driven analytics tools allow data experts and retail people to contribute in the best way possible to help the whole business thrive.

(A Retail Times’ sponsored article)