Returns Management System hacking Matt Clarke, @techpad Sunday, 17 February 13
Dec 15, 2014
Returns Management System hacking
Matt Clarke, @techpad
Sunday, 17 February 13
The problem...
• We promised customers we’d handle their return in 7-10 days.
• We suspected that sometimes returns didn’t get handled within this time, but our existing RMS couldn’t tell us.
• When returns handling was slow, call volume went up, so did email queries. This annoyed customer services and customers.
Sunday, 17 February 13
Step 1: Track emails
• Before we started on a major rebuild of the RMS, we added event tracking to the form used to categorise customer service emails from the site.
• We do a similar thing with complaints analysis (there’s a post on that on my blog).
Sunday, 17 February 13
Step 2: Monitor emails
• I built a dashboard to provide an overview of emails received.
• I used the API to report on increases in returns emails/proportion, which might indicate a failure to meet customer promise.
Sunday, 17 February 13
Step 3: Dig deeper
• I spent three months writing an MSc project to further investigate the problem, and proposed a solution to tackle it.
• The proposed solution used GA, among other things...
Sunday, 17 February 13
Step 4: Re-build the RMS
• We rebuilt the RMS to tackle the issues the business was facing, as well as those that impacted customers.
• We added a metrics system so we could record which returns were pending, due today, late etc, and help staff prioritise and hit KPIs.
Sunday, 17 February 13
Step 5: Plan event tracking
• I made a spreadsheet of events. There were lots...
• Why GA? Using GA would mean I could analyse and report on the data much more easily than I could if I had to write SQL queries to pull the data out of the RMS.
Sunday, 17 February 13
Step 6: Sent events in PHP-GA
• There were too many events to send using the client-side code, so I used PHP-GA, which allows you to bypass the token bucket algorithm.
• Primary keys in the events allow related events to be re-joined in the API.
Sunday, 17 February 13
Step 7: Set up reporting
• The default Google Analytics dashboards were too limited to be of use for this problem.
• So, we’re using Google Drive and the Google Analytics Core Reporting API “magic script” to create detailed reports.
Sunday, 17 February 13
We can now answer these questions on returns
• What percentage of returns are handled on-time?
• What is the average time to handle returns of different types - replacement, exchange, refund?
• How many returns are unscheduled arrivals?
• What are the return rates for different items, and why are they being returned?
• What proportion of faulty goods are non-faulty?
• How much working capital is tied up in returns?
Sunday, 17 February 13