We’re hiring! View open roles and join our team.
Skip to main content

5 reasons why your retail store analytics projects fail

Article by Victor Hoong

Data for the sake of data is never good enough. Read more about how to build efficient data-driven store analytics capabilities to succeed at retail.

The data-renaissance in the retail industry has become somewhat of a buzzword. As retailers become aware of analytics solutions needed to understand customer demands and increase store effectiveness, the arms race continues. Stores are getting equipped with sensors, cameras, virtual reality and other technology which vendors have promised will bring a data-driven approach to brick and mortar retail. However, one main challenge arises – for the most part, in the retail business data is being generated for its own sake. Piles and piles of reports and charts are being created, seldom driving meaningful, disruptive retail innovation. How can we do better? In this article we will explore the most common reasons why these experiments fail, how we can tackle the issue from a different angle and how we can build effective data-driven store analytics capabilities.

 

1. Stop focusing on solution requirements instead of achieving business goals

There are many aspects to consider when choosing the right retail analytics strategy and vendor. What you should definitely look for is the ability to illuminate every part of the shopper funnel, and to conduct experiments that can be rapidly validated.

However, we sometimes find retailers asking for highly granular or highly accurate data without first thinking through what level of granularity or accuracy is good enough to make decisions. They then tend to abandon potentially valuable data and return to store intuition because of an error margin of 1 or 2%. Meanwhile, the accuracy provided could have already lead to statistically significant findings to improve store performance. Being focused on the perfect retail analytics science can distract us from our real goal – improving retail performance or shopper satisfaction. By looking at data as ‘vehicles’, not destinations, our mindset changes. Instead of searching for more data or more precision, we should be designing smarter tests that will still reveal robust insights into what is available.

For example, to understand the value of in-store digital assets such as touchscreens, VR and augmented reality tools, or smart mirrors in fitting rooms, some retailers want to track each step the individual shopper took. Not only is this extremely expensive to do, but it potentially also infringes on privacy rights. The mistake we have made is that we are working with a solution requirement instead of a business goal. It is as if we expect to find technology that will simply tell us with certainty what improvements we need to take without us even thinking about it. Unfortunately, it is not so easy yet.

 

Solution

Framing: We must avoid framing solution requirements and start re-framing them in terms of business goals. Going back to our earlier example, we might explain that the in-store digital touchscreen is intended to raise awareness of specific in-store products to drive an increase in sales. With this in mind, we could then test the correlation between the number of times a specific product or category is viewed on the device and the number of touches of that product on a shelf in the store, and subsequent store sales on the other. By re-framing the problem, we are able to test if the digital touchscreen is achieving the business goal, using the available data and without attempting to track detailed individual customer behaviour.

Test design: After proper framing, effort investment for good test design and preparation is required. Ensure enough time and resource capacity is available to undertake this process. We have to identify what reliable data we can surface and ensure there is enough of it to run a valid test. It’s important to identify potential test pollution factors and identify ways to control the experiment so that these do not impact the outcome of the test you are trying to perform.

 

2. Not having a standard structure and repeatable approach for data analysis

Analysts are often unleashed to churn a large variety of data without a structured framework. This leads to three problems:

  1. Confirmatory bias. We have a natural tendency to find data that supports our understanding of the world and ignore other data points. This can happen subconsciously and is not necessarily driven by any improper motive
  2. Effective time-management. Analysts without a structured process are very likely to be inconsistent with their analysis process each time that they do it. This means that they may spend a long time drilling into unfruitful areas of investigation or missing insights because they are exploring data in other places
  3. Strategic prioritisation. Not every experiment and test will have the same impact on business goals. Within constraints of resources and time, analysts should focus on initiatives with the largest potential to add value to the customer experience and sales performance

 

Solution

One of the methods proposed by Riverflex and available in the StoreDNA platform is to utilise a simple FunnelMap compass. This enables key steps in the consumer journey and the goals of the retailer to be mapped. It also helps to highlight where there is a high likelihood of having performance improvement opportunities.

 

3. Treating store analytics as retail concept innovation as opposed to a new operational capability

With the rise of start-up culture, it is getting increasingly ‘chic’ for enterprises to partner with start-ups in any manner. Many digital innovations are well catered for such an approach. Following initial set-up, they require little support or human intervention from the organisation. However, something like embedding retail sensor technology into the store is different. Such sensors are only the first step. They uncover new data sources, but in order to translate that to any kind of business value, we need the organisation to engage fully at additional levels. First, they must integrate different data sources meaningfully. Secondly, someone must interpret the data into insights. Thirdly, those insights must be translated into actions and implemented in store, and so the learning loop begins again. Any break-down in the organisational process at any point along the data-insight-action chain and the value of the whole is lost.

 

Solution

To get value out of retail store analytics, we have to see it as a capability and not an innovation. Owners must be appointed to ensure that the insights being surfaced are also being driven into store-level actions and feedback loops. New roles and skillsets are required to combine big data skills with that of visual merchandising and store design. Without organisational support in place, value cannot be surfaced from the technology investment in the longer term and we find that all time and investment into store analytics pilots simply go to waste.

 

4. Under-investing in the initial test, and not being able to scale the insights to a wider store portfolio

Over the past few years we have found that many retailers are starting their retail analytics journey in a single store and without the dedication of a focused team to support it. The focus is on a technical test instead and/or ad-hoc data point capture and is undertaken as a pilot innovation, rather than with the vision of capability build-up. The initiative is attempted in a siloed team and there is not wider corporate support. All this creates an additional friction point, where it gets very difficult to scale the initiative further.

 

Solution

We suggest our customers to choose at least three different stores belonging to the same cohort group. This allows a good balance between the initial investment required and having a set of stores for running simultaneous tests against which we can triangulate findings for greater robustness. In addition, a team with sufficient capacity and capability is needed to make sure experiments are well defined with the right KPIs, pre/post analysis, and support for implementation and rollout.

 

5. Not having the mandate or sponsorship to experiment, action and scale the insights across stores

We often say that we are actually in the business of building bridges. Retail store analytics projects are often initiated by innovative characters within the organisation. Someone who is searching for the latest ideas and is willing to invest early and experiment to get ahead of competitive. These bright innovators understand how to read data, and what are the possible solutions. However, innovators by nature work and think differently to others in the organisation. As a result they can often be isolated without the decision making power to act upon these findings. On the other hand, you have decision makers who do not believe in data science or the capabilities of today’s technology. Our job is to build a team who can bridge this gap, and foster quick flow of actionable insights to support decision makers.

 

Solution

Select a vendor which understands both data-driven decision making approaches and the change hurdle which is limiting adoption across the organisation. Identifying and engaging key stakeholders and tackling barriers and objections increase the chance of success for your change process. If we simply add additional technology layers into the organisation without changing underlying business or decision-making processes, then we are not creating any additional business value. We need to change the existing way in which people are working. Technology cannot change people. Only people change people.

 

Closing thoughts

The team at StoreDNA and Riverflex have been working together on getting retail store analytics into companies for the past four years and these challenges come up time and again. You can read about our latest implementation for Samsung’s pop up store here, and how we supported the creation of the Store of the Future for a global fashion retailer. We hope that the discussion we have shared above will help you avoid the common pitfalls in retail store analytics and make you further progress. In all cases, we would recommend that our clients spend time looking at how to build internal muscles to face-off against these challenges back into their own respective organisations.

Read more about our take on retail innovation and how we can support your retail organisation here and read more about our applications for Samsung and a global fashion retailer here. Check out our insights on IKEA’s recently opened small store format here.

 


About the author

Victor Hoong is the co-founder of Riverflex. An ex Deloitte partner, has 16+ years of experience developing digital strategies and roadmaps for major brands and retailers.

Complete The Form Below To Get Your Copy

    BLOG

    Read expert opinions and articles

    For up-to-date news, follow us on LinkedIn ↗