Of late, we have seen retailers adopt retail analytics platforms to help inform core decisions. For these systems to succeed, retailers must painstakingly align internal resources with platform features and implementation processes. That said, too often we see retailers decide on data analytics which are overly-focused on data generation. This creates a false image on which data platforms compete heavily on price, and not the core mechanics which eventually enable ROI and company-wide implementation. In this article, we investigate five imperatives that every retailer should demand from data-decision making platforms embedded in daily working routines. Spelling out company interests enables retailers to act and use time, budget, and other resources wisely to ensure proper delivery. The imperatives are outlined in order of priority.
Many retailers think of applying technology as an innovation process, with dedicated departments whose core goals are to test new technology and advise on how to scale within the company. While conceptually a good idea with a well-established purpose, these data-/decision-making projects almost never reach deep-company installation.. In theory, these departments could function well if a small subset of technology is tested and applied quickly. For example, this initiative could involve testing how consumers react to new, digital store screens, when variables for success are clear and there is no more work needed from other retail functions.
However, when testing a performance-decision platform at the fleet level, it is difficult to set up a testing environment in which success criteria is easily defined and managed. At the core, we call these projects suicide missions, as they create false images of data platforms not adding enough value. To develop a purpose and articulate it in a way that that resonates with all functions, data platforms need to be groomed at VP of Retail level, or higher. In this way, retailers can ensure that all findings will be used for a wider fleet roll-out and ensuring the impact is delivered at a satisfactory level. If we allow platforms to be contained in small environments, we will not deliver the next stage of insights, which are designed to deliver fleet-wide improvements.
When choosing the right decision-making platform, retailers must ensure its future longevity and upgradability.
The ideal working environment has the fewest different number of apps and platforms, and centralizes key elements in just a few interfaces. Thus, these platforms must immediately ingest different, already-available data sets. For example, at the minimum, POS data should be ingested and made able to compare and correlate with behavioral drivers derived from in-store cameras. Stock level data are crucial to drive real-time inventories and improve sell-thru rates. Decision-making platforms should account for all key data points and act as the ‘glue' between different functions.
Within retail organizations, there exists a built-in conflict of interest which platforms must manage very carefully. Cross-functional capabilities should enable increased levels of understanding of individual functions and cooperation.
For example, we rarely see a buyer and visual merchandiser sit in the same room, with common ground on how to increase the sales of SKUs they are buying or displaying. At its core, there is no other than intuitive logic behind those meetings, and as a result both meetings and results end up being non-productive. One of the key benefits of making decisions based on data is increasing organizational capabilities to establish a common ground for different functions to flourish. However, technology can not do this on its own. Importantly, organizational readiness for change must be enabled to truly adopt new decision-making paradigms.
There are many retail analytics platform on the market now. Most commonly, we see a deployment of a variety of wifi trackers and beacons. These are the most cost-effective ways to get to 'behavioral shopper data'. However, most of the time, these data are not accurate enough on which to base any decisions upon. Retailers should be adamant to understand base metrics such as dwell, coverage, and engagement, and importantly, platform measuring methodologies. Even with most sophisticated techniques, such as using machine learning algorithms through video are impacted by error rates. What’s critical is to be transparent about what the error rate is, what kind of bias does it have, and what kinds of decisions are acceptable to the user, understanding that the data is not 100% accurate. Also, be sure to ask your providers to deliver a full and transparent report on how they derive metrics, and what kind of error ratios they currently have. Also, we think it’s important to be present at the calibration and work together with the supplier to achieve a maximum level of accuracy for every designated environment. This will be the foundation for efficient and accurate decision-making.
As technology matures, there is the potential to reveal more and more personal, identifiable data thru cameras. It is crucial to make sure that all suppliers are using 'zero video capture' policies. This means that none of the tracking algorithms store any images or work real-time in a closed, store environments. There are many platforms which take images online for processing which could cause serious breaches of GDPR (General Data Protection Regulation) rules. Privacy protections should be embedded in data collection platforms and all parties should be cautious to enable all available protection policies.
As a retailer, make sure you answer these three questions before you procure a retail analytics platform.
Who is the user of the platform?
Many times, retailers get excited about behavioral data collection, though there is no concrete next step. The users of the platform are data teams which do not have enough authority to make changes in stores, thus limiting any opportunity to prove-out the value of the data. If the platforms themselves do not have different user interfaces which adapt to the needs of individual functions, it is important to establish a user base which can in turn inform decision-makers. In particular, feedback about what is the best course of action and what are the results of the changes made. This feedback loop is important to establish a learning organization.
What do we want from the platform?
For many situations, retailers need a quick fix to their flagship stores and demand a quick solution from data analytics vendors. While most vendors oblige in order to earn a new customer, this quick-turn process diminishes prospects of implementing long-term, platform-wide solutions with the ‘right’ people in the company. We believe that sometimes it is better to wait and address a wider-party who could benefit from the solution. In doing so, you can use the opportunity to establish data-driven decision-making as the core driver of better daily reporting and action plan implementation.
How can we proliferate learnings across the entire fleet?
On other occasions, in which the platform has been successfully installed but used in a limited fashion for a smaller territory or subset of stores, there is a danger of not enabling proliferation of knowledge and usage across the entire fleet. Typically, we advise retailers to start with fleet optimization tools for which no cameras are involved, to introduce a data-driven decision-making process on a higher level. This enables more precision on where to install camera-based shopper tracking solutions, and provides a foundation on how to scale learnings from a smaller subset of stores, to the cohort level, and then to the entire fleet.