Rethinking the store walk
Almost every client we speak to at Quorso is currently rethinking their store walks. This essential routine is a core part of monitoring any...
Inventory accuracy doesn’t need more processes or tech, it needs better use of existing data.
Inventory inaccuracy has been ~65% for the last decade and costs retailers 4-8% of sales annually, whilst also having Balance Sheet consequences. The introduction of products like RFID as well as labor heavy correction processes haven’t helped. But an exception-based, data-led approach can.
Inventory accuracy is a foundation on which all retail is based. Demand forecasting, order management, the ultimate end customer experience, all rely on accurate and timely inventory management.
Yet inventory accuracy is notoriously difficult in retail, and becoming more so with the rise of omnichannel. Independent surveys over the last decade continuously show that, especially in large complex stores, inaccuracies can be found in ~65% of stock counts.
To date, retailers have used two methods to fix inaccuracy:
A highly time consuming, banal activity for associates to perform e.g. stock counts. Processes that can also lead to greater inaccuracies due to greater manual interventions.
e.g. RFID to reduce counting or check-out errors, or tags on high-value items. However, these technologies have proven expensive, especially with fast-selling goods.
Here at Quorso, we think there is a third way, cheaper and less time-consuming than either of these methods, while leading to better outputs and thus far more productive.
There are good reasons why inventory accuracy will never be 100%. Retail is a physical activity. Damages occur, theft occurs, human errors occur. Even with corrective and preventative measures employed, inaccuracies will still occur.
An exception-based approach to inventory inaccuracy isn’t looking to achieve 100% accuracy, but to alert on:
Here’s an example of the process we used to drive exception-based inventory accuracy at a large national retailer:
The original focus was on high-volume SKUs with an appropriate value, so that any labor investment would be outweighed by the value from fixing the inventory inaccuracy. We looked at how those sales of those SKUs were expected to perform in each store, based on cohorts of stores like them.
Using data science, we looked to see where patterns suggested an exception based on sales behavior, inventory on hand and overall volatility.
This reduced the number of SKUs to check from the full Stock Count of 100s to typically 5-7 high priority SKUs, saving hours of labor per week.
Simple, clear communications to the team are essential for action. Rather than hiding the exceptions in volumes of reports, very clear alerts to the responsible team were necessary to ensure focus and confirm that action had been taken.
This clear, direct communication further reduced the labor time required.
By measuring the success of each alert and adjusting accordingly, we were able to continuously refine our logic and improve our impact. Initially, our hit rate on required adjustments was <50%, but it is now above 90%.
Unexpected benefits were also unlocked, such as releasing millions of dollars of phantom inventory.
As we go into a more complex retail environment, inventory management is only getting harder. But the solution doesn’t have to be more manual processes or large investments in e.g. RFID or computer vision. Instead it might be right under retailers’ noses – using your own data smarter.
Exception-based inventory management processes can be used to: