April 10, 2026
The company runs a large online auction platform and shopping website that connects buyers and sellers in over 190 countries. Their website has millions of sellers with 1.9 billion global listings and over 132 million active buyers.
THE CHALLENGE
The company introduced an authentication program for categories like trading cards, jewellery, watches and sneakers, to give buyers confidence that the products they received were genuine. Sellers of products which qualified for the authentication program, had to send the items to the company’s authentication hubs. There, a team of authentication experts checked each item, before confirming that it could be dispatched to the buyer.
Each hub had a fixed team of authentication experts and an additional contracted team that was deployed based on the estimated number of incoming products to be reviewed. The company did not have a robust tracking mechanism to identify when the products to be authenticated would arrive at the respective hubs. The existing Microsoft Excel based tracker did not take into account factors such as holidays, changes in shifts at the hubs, delays in receiving the products from logistics providers etc. As a result, if the number of products at the hubs on any particular day was greater than the capacity that the available authentication experts could handle, then there were unanticipated delays in processing and sending out the items.
The company needed an analytics partner to build a hub staffing optimization model that would accurately determine the number of authentication experts required, on a daily basis, for each hub.
THE SOLUTION
The Prescience team initially managed the Excel based tracking system to understand the existing process and identify the different gaps that led to staffing mismatches. Based on the forecasting requirements, the team proposed a robust forecasting approach using the Prophet model. Our team streamlined the communication processes between the different internal and external stakeholders, based on which the Model could incorporate hub holidays, shipping holidays etc. Since the majority of the products were delivered to the hubs by the top 5 logistics companies, the number of incoming products would reduce drastically if one of these companies had a holiday.
Every Friday, the mid-term staffing forecast was sent out to Category Managers and the Operations Heads at each authentication hub. In the first and third month of every quarter, a long-term forecast based on the 12-month rolling average, was also sent out.
This new forecasting system also provided users with charts on the available capacity and expected inbound items at every hub, for any time frame. Users could request for a higher number of contracted authentication experts for a particular date, with the required justification for the same. The change logs helped track the forecasted and revised count of authentic experts at each hub.
The different technologies used for this engagement included,
- SQL
- Python
- Microsoft Excel
THE IMPACT
With the weekly hub staffing forecasts from the Machine Learning Model, the Operations Heads were able to rationalize the count of authentication experts in their respective hubs, without any impact to the processing of the products to be examined. The model was able to forecast the capacity required with 95% accuracy, and this gave the Category Managers and Operations Heads end to end visibility into the expected demand and cater to it.
