Calibrating data for the new normal

Consumer shopping behavior has fundamentally changedprobably forever. As the world adapts to this new normal, many retailers and consumer product companies must change the way they operate. Their supply chains must be nimble; their product packaging must be versatile; their demand response must be dynamic.  

But the changes don’t stop there. The situation also forces changes in the way companies measure their business and make key decisions. Whether that is:

  • Setting up revenue and growth targets
  • Assessing store performance
  • Decisions on store closures
  • Assessing partner/employee performance
  • Design of customer surveys

To enable these changes, the data preparation and harmonization must change as well.  Insights are going to be more perishable, and it is critical to have the foundation for success to ensure your organization can deliver the right recommendation at the right time.   

 

Four paradigms of data adaption for the current times 

  1. Period flexibility.  From revenue point of view, 2020 has been a low point for most businessesYet the sudden change in the business dynamics will undoubtedly make the next year growth numbers seem ridiculous. Very soon, this year over previous year comparisons will stop being relevant. For ad hoc analyses, companies can switch the comparison year, but most automated systems will be (a) not easy to switch and (b) create chaos because every team will come up with their own logic in altering their comparison year.

    Therefore, depending on the type of business, corporations should leverage historic data and develop a hybrid model as a comparison periodThe central team, often led by the Chief Data Officer, must design/identify the best comparable periods and make those available for the larger organization to use. Since this will not disrupt the end consumption layers, the reports and dashboards will not need major change.

  2. Attribution
    • Product Attribution: Typical product taggings are based on broad classifications like sales (fast moving vs. slow moving), product categories, and customer preference. And now it is important to tag products based on the processes, too, such as manufacturing (does this involve any personnel handling), packaging (type of material used)and selling (picked up by consumer or delivered by personnel). These tags will help segment the products based on consumer perception and preferences. For example, this will help further categorize products like frozen waffles from bagels, though both might be tagged as breakfast products.
    • Store Attribution: In addition to the geographical and sales-based classifications, organizations must tag the stores based on additional lenses like the type of products that drive the overall sales, the peak sales times, and customer purchasing/consumption preferences. Asset information, such as availability of key machinery, digital menu boards, and even restroom availability, will help in monitoring store performance.
  3. Enable model dynamicity through data depth: With perishable insights comes the challenge of immediate intervention. And to enable this, the analytics set up must be based on dynamic models. Automated models serving output on demand can help course correct faster.  For example:
    • Having Demand Transferability of each store made available on the fly will help with store lifecycle monitoring and quick actions on store closures. 
    • Having automated SKU rationalization models will help with adjusting product manufacturing and placement in a more dynamic way.  
    • To enable automation of these models and having their output on the fly, performance scores at store and product levels must be modeled and fed in with the data tables. Dynamic indices can help with the scores and KPIs.
  4. Panoramic data modeling: Expanding accessibility to relevant data sets about the stores will reveal layers of truth on business performance. This does not allude to purchasing more data, rather stitching the ‘apparently’ disparate data at the most granular level. Even the existing data, when harmonized to the store level, will help with dynamic insights.
    • Incidents at stores, i.e., from what is happening inside the stores (device failures) to what is happening to the stores (violence). 
    • Government data to monitor the status of counties and localities for school closures and governmental lock downs. 
    • Social Media to get a holistic understanding of the consumer behavior and preference, and hence, the potential demand on products and brands.  
    • Panel data to monitor purchase behavior.
Changing consumer behavior is forcing businesses to change their way of operations. This needs to be followed through with changes in the way data is captured, harmonized, and leveraged to enable dynamic decisioning. Else, the dissonance between the reality and data ecosystem will impede future growth potential.