Improving Customer Service with Big Data

Customer service is key for ecommerce businesses. If retailers could improve customer service while also reducing its cost, they would likely do it. Big data can help achieve this goal in the following four areas.

1. Website Is Broken

This is the most common reason to contact customer service. The retail experience could be broken because of a missing product page, a promotion code that is not working, inability to check out, a payment not getting processed, the site being down, or some other reason.

The best way to manage a broken site is to automate the prevention of it and, if necessary, automate the resolution. The retailer should analyze items such as customer service tickets, defects identified in the system, and server log files. For example, if the product pages are broken frequently, the problem might be in the validation process. This can be automated by creating a scheduled validation that will review all product pages before launching them. By analyzing all the reasons and developing automated ways to resolve them, the user experience is enhanced, shoppers are happier, and the customer service team’s load is reduced.

2. Product Returns

A product return is the second most common customer-service interaction. A return could occur for several reasons, such as a damaged product, the wrong item was shipped, an item that did not fit, the customer found the same product cheaper elsewhere, or even the customer did not like the item after it arrived and decided to make use of the free returns policy.

An analysis of the historical returns data will help determine the top reasons. They can be categorized by product, region, vendors, customer segments, or even individual customers. This will help understand questions like which products are returned the most, what is the most common reason for product returns, and whether returns are more common for certain vendors. Additionally, using a big data tool like a real-time analytics engine or a visualization tool, rules and thresholds can be created that get automatically triggered to send alerts for unusual returns activity. Once the retailer has all this information, it can optimize the product assortment, determine which vendors are working better, and assess a customer’s history. This will result in reducing the returns rate and lowering the number of tickets opened by the customer service team.

3. Fraud

Fraud is an unfortunate reality of an online retail business. It typically affects the customer service team. Common types of fraud include denying the product was delivered and purchasing with stolen credit cards. Both of these results in credit card chargebacks, which hurt retailers.

Chargeback fraud can be reduced or prevented by analyzing customer purchase patterns, confirming the purchase with the buyer before the checkout is completed — especially for high ticket items — and for new customers, analyzing system log files to identify a customer’s location and his IP address, browser, and operating system. Create alerts if anything looks different from the norm. All this can be done using big data solutions that provide a flexible framework to define rules and application of those rules to a variety of data sources. These solutions also come with connectors that pull data from log files and then analyze them in real-time. This automation will lead to reduction in fraud, fewer calls to customer service, and improved profitability.

4. Delayed Delivery

Delayed delivery of an order also results in customer service calls and emails. Most retailers have automated the delivery communication by providing the tracking information online. But there can be several other reasons for delayed delivery, such as the product being lost, getting stolen from customer’s doorstep, and delays in customs for international shipments.

Big data can only help with this issue if the reasons for delay have been documented over the years. The reasons will help refine and communicate the lead times for delivery of different products. The analysis could also identify the regions where next day shipment is not feasible, thus easing the load on customer service and improving the overall process.

Gagan Mehra
Gagan Mehra
Bio   •   RSS Feed