Oracle OpenWorld 2015 - eBay runs ‘scrappy’ Knowledge experiment that yields results for customers
- Summary:
- eBay is using Oracle's Knowledge Management to help automate its customer service interactions. The short experiment has resulted in a reduction in handling times for service agents.
One of the neatest examples of this that I’ve seen this week came from eBay, which gave a short presentation about how it has used Oracle Knowledge Management to run a three month experiment that gave it some better insight into how to deal with customers via its contact centre.
Nina Patel, Director of Global Knowledge Management at eBay, took to the stage to outline that whilst eBay still carries its ‘auction site’ history with it, the company actually mostly acts as a marketplace for the sale of new goods online. She said:
Today’s eBay isn’t what it used to be. It’s not the auction site that people think of us as. Perception hasn’t really kept up with the reality. The reality is that 79% of what is sold on eBay today is new merchandise and is available for purchase immediately.
This makes us the world’s biggest shopping destination. We connect buyers and sellers globally, empowering them to create connected commerce and opportunities through connected commerce.
We have a multi-service channel strategy. We cover phone, email, chat, self-service, IVR and a number of other new initiatives. That goes across a number of customer segments, multiple markets and various transactions types that take us from cross border trading, to simple buy-sell.
And with 159 million active buyers and 800 million listings, it’s unsurprising that eBay wants to keep its customer complaints and handling times to a minimum. Patel wants the contact with the customer (buyer and seller) to be as efficient as possible, which in turn should boost customer satisfaction and bring results for eBay.
Making the right decisions
When Patel joined eBay a couple of years ago, she asked herself: what is the role of Knowledge within the organisation? She said that she believes that it is there to help eBay’s people make effective decisions with the right data, information and know-how, in place when they need it. She said:
It’s at the core of how we make decisions.
However, this comes with its challenges. Patel said that she is “constantly” being asked to provide data metrics that show that Knowledge is making an impact on the business. However, she made some progress with this year with an experiment she ran with eBay’s service agents. She said:
What is that agent or team mate, as we like to call them, going through on a day to day basis to help them make decisions?
Take ‘Mary’, for example. Mary supports a portion of our business that focuses on making sure that buy-sell transactions are processed according to all of our guidelines, policies and rules. That’s how she supports our customers, both our buyers and our sellers.
Mary picks up the phone, we have a buyer, the buyer says they have not received the item they purchased. So Mary has to start researching. What’s happening? Why has that buyer not received the item? Where should she start looking? How does she make the decision on resolving this for the buyer?
Which means that Mary has to understand position of the seller or the marketplace that the buyer purchased from. So Mary looks at a number of pieces of data, a number of pieces of information, a number of pieces of Knowledge assets. Then she has to pull all this together and make some sense of it, to start to help to make the decisions with that buyer so we can get to a resolution.
For Patel, that was the key issue. The process was not automated in the slightest, leaving it up to the service agent - Mary - to access all the information she needed, for every case, no matter how simple, in order to make a decision.
Green, yellow, red
Patel said that eBay has done a good job of getting ‘Mary’ to this point. But what it needed to do was to take it to the next level. As a result, about three months ago, eBay started an experiment that looked at exactly the problems that it was having to solve for customers.
Patel wanted to know if eBay could get better at reducing the variance amongst call agents in terms of the amount of time they spent handling calls, but also whether or not it could improve customer satisfaction as a result.
The experiment took 90 days - 30 days to design and 60 days to collect data. Patel said:
We looked at process, we looked at business rules and then the data that enabled the business rules. Then we brought all that together and made sure that we were able to measure it. So that we could start to really understand the impact and understand the opportunities for our approach and process.
Because we are a global marketplace we have a number of different policies that allow our buyers and sellers to transact with each other. We needed to understand what those policies were and the processes that enabled that policies for buyers and sellers and how our team mates supported them. We also needed to understand the business rules that drove the steps and those processes.
And then we needed to understand what are the key pieces of data to help make decisions and the key pieces of Knowledge or information to help make those decisions to get to the outcomes from a productivity perspective.
eBay started with tracking and receiving purchased items for the experiment. See the image below:
By analysing the processes, the policies, the business objects and the data, eBay was able to find examples where cases could automatically have a recommendation generated for them (green).
There were then cases that had a partly automated recommendation, plus some work from the service agent using the Knowledge centre (green/yellow).
And finally there were cases that needed the service agent to spend a lot of time with the customer, needed them to build a relationship, understand the situation and use all the information they have available for the latest interaction (green/yellow/red).
Patel said:
When you get that green use case coming through as part of the call, they can automatically get to a recommended decision. They don’t have to really think about it. We have understood the process, we have understood the business rules, we have helped the team mate get to that point. They can quickly resolve, satisfy the customer and continue to move on.
When they might get some yellow use case components, they have to go through some different flows. So the team mate (service agent) would see these various options as they went through the call type and the different options by which they could be supported to continue to resolve that call that came in.
And the results? eBay has reduced the number of ‘clicks’ by its service team by over half in some instances and it has brought the variance in team handling times right down. Patel said:
After 90 days our experiment was really a success. We measured our data using six sigma standards, so we had really strong control and test plans in place. We saw significant impact in both the handle time variation and our recontact rate.
When I say significant impact on handle time variation - we had a pool of 30 team mates and we were able to show that the team mates started to go to the mean in terms of handle time. We reduced the variation from the mean by almost 2 minutes.
Our resolution rate should impact our customer satisfaction, because the quicker we resolve the better satisfaction.
Result.