Assuring ROI for big data and machine learning based analytics in Retail.

Big Data and machine learning -based analytics solution providers often struggle with declining post-implementation satisfaction rates. Both senior management and the users experience disillusionment with the results these solutions deliver, finding it difficult to measure and justify the investment over time. There are several reasons for this phenomenon:

  • Many data science teams have not delivered results that can be measured in ROI by executives.
  • The excitement of the newly acquired visibility has led people to ignore what should be the threshold requirement: using that visibility, what actions can they take that can help them make more money?
  • They often realize that analytics aren’t really embedded into business processes at the point of where decisions are made.
  • Due to different reasons like technology or budget limitations, some implementations are using aggregated data and do not allow visibility into the most granular data available in operational systems.

In the retail industry, where margins are so thin, this issue is even more dramatic, and there’s a need to find ways to demonstrate the ROI of analytics projects. Retailers usually perform very well if they focus on a specific process or a policy, but excellent execution is expensive; it requires managerial attention and investment of resources. What if retailers could apply advanced analytics on processes which were somewhat “neglected” so far, but can generate ROI for the entire deployment?

Here are a few examples of such processes, by retail segment:

Reverse Logistics — Department stores and specialty retailers

  • In practice, up to 10% of the turnover is returned by customers, and the visibility to margin opportunities dramatically drops in the reverse logistics process compared to the visibility of sales. Do retailers give their reverse logistics the same attention they give their downstream merchandise? Is it being monitored for exceptions and trends at the SKU level?
  • Reduction of customers returns or better salvage of the returned items can be translated to hard dollar savings and used for the analytics solution’s ROI calculations.

Shrink and waste in fresh categories — Grocers

  • Do they give their fresh categories the same attention that they give their center store merchandise?
  • 25%-33% of their sales and 50% of their shrink and waste could come from these categories. Do they invest reasonable efforts into controlling them at the SKU level?
  • Shrink and waste reduction is easily measured against a baseline and can be used for the analytics solution’s ROI calculations.

Monitoring back room vendor receiving exceptions — all retail segments

  • Do regional and store managers know who are the worse vendors (and drivers) delivering to their store, by exception type and exception value?
  • Can they create a baseline of these exceptions so they can demonstrate how issues were taken care of and losses reduced compared to the baseline?
  • Mitigating the risks of delivery operations by consolidating their controls, monitoring customer claims and transporters performance, and reducing cases of employee-customer collusion etc. can deliver ROI for the analytical solution used for detection and monitoring.

Many retailers will answer at least one of the above questions with a “no”. This means that high level aggregated dashboards (which almost everyone uses today) do not replace SKU level analytics. Granular analytics (with the right AI and LM algorithms that can eliminate “noise” and false positives) can help retailers to increase profits and sales in a measurable way, thus creating a case for ROI.

However, analytics isn’t everything. Someone needs to take action and make a difference at the end of the process. In order to assure success, retailers should review the following practices that if applied will boost the ROI significantly.

  • Improve data capture and accuracy.
  • Gather information automatically.
  • Apply rules-based processing, early in the lifecycle of a return.
  • Create and track key performance indicators.
  • Share the recommended actions with other parts of the organization.

Data Driven Investor

from confusion to clarity, not insanity

Omer Matityahu

Written by

Twenty years of entrepreneurial and business experience, an expert in retail technology, retail analytics, retail operations and execution, loss prev

Data Driven Investor

from confusion to clarity, not insanity