The Great Social Customer Service Race IndustryView | 2013
Lately, few topics have dominated customer service strategy conversations like social media. Once viewed exclusively as a marketing avenue, Twitter, Facebook and other social channels are increasingly used by buyers for support.
This trend presents a formidable challenge for companies that receive thousands–sometimes hundreds of thousands–of tweets per day. It’s impossible to expect them to respond to everything. But what do they respond to, how quickly, and how well?
I wanted to answer these questions and more in a five-week research project I recently conducted called “The Great Social Customer Service Race.”
Four Software Advice employees used their personal Twitter accounts to send customer service tweets to 14 leading consumer brands in seven industries. Each company received one tweet per weekday for four consecutive weeks.
During the first and third weeks, our employee participants used the brand’s Twitter name with an @ symbol. Using the @ triggers a notification to the account owner that they’ve been mentioned in a tweet. In the second and fourth weeks of the race, only the brand name was used.
The questions fell into five categories:
Urgent, or I need help right this second
Positive (“thank you!”)
A question from their FAQ page
Technical, or needs more than one interaction to solve
The evaluations in the below graphics are based on the time it took the brands to respond and the percent of total tweets that received a reply. Here’s a breakdown of what you can learn from each matchup.
Despite winning the matchup for response rate and time to respond, Coca-Cola committed a huge error when one reply came four days after the question was sent:
Taking four days to respond to a straightforward question “is basically the same as not replying at all,” says Anna Drennan, the marketing manager for social listening software Conversocial. More than half of Twitter users expect a response within two hours of tweeting a company, according to a recent Oracle report.
Drennan suggested one possible reason for the delay. The help ticket might have been routed multiple times to find the right answer.
To avoid an extended delay such as this example, companies should ensure their response policy requires agents to send a boilerplate response if they don’t know the answer right away. Something like “@customername I am looking into this for you now. I will get back with you ASAP. Sorry for the delay! (-: >AF.” Additionally, if the company uses customer service automation software, the system should be programed to alert agents if a ticket doesn't receive a response. For social media, this time limit would ideally be two hours or less.
In our credit card group, MasterCard was the clear winner. Their competitor, Visa, did not respond to any of our tweets during the race. MasterCard also earned special recognition by capitalizing on an opportunity to market a customer service interaction.
When one of our participants asked whether the credit card is accepted globally, the MasterCard team responded and re-tweeted her message:
This demonstrates to MasterCard’s 30,600 followers that they listen and respond to their customers. The post was RT’d (re-tweeted) another 12 times and personally engaged one follower.
In another instance, MasterCard used the customer service opportunity to pitch our Twitter participant a new product:
Overall, these competitors did not perform well compared to the rest of the group. Starbucks didn’t respond at all, and McDonald’s suffered from the same slow response time pitfall as Coca-Cola. In addition, it wasn’t immediately clear that the responder was from McDonald’s, and she didn’t provide a good answer to our problem.
If she really wanted to wow and delight the customer, she could have asked where our office is located and provided the nearest store’s address, phone number, and manager’s name–or gone further, and contacted the manager herself. She could have also addressed at least some of the questions, if even just to say, “Yes, some stores can do that.”
The banks were by far the best competitors in the race, for both response time and response rate. Their responses were not flawless, however. This tweet indicated substantial risk of switching brands, yet the bank’s response was robotic and did not directly address the question:
Many social listening software products can be programmed to prioritize messages that include important keywords, such as “help,” “mad,” and “switching.” Tweets containing words like these should be given more thoughtful responses.
Wells Fargo also loses points for not replying with “You’re welcome!” when we thanked them for the help. “If the customer takes the time to thank you, you should do the same,” says Lien Brusselmans, marketing manager for social software developer Engagor.
Customer service expert, best-selling author and speaker Micah Solomon told me recently that being human in your engagements with customers on Twitter is one of the most important considerations.
In contrast to the Wells Fargo example above, Home Depot’s responses were personable. Also, the agent continued the interaction past one response:
Several of the brands provided links during the race, but most led to pages with only marginally relevant information. This can be frustrating for the customer.
If you have a page or piece of content that directly addresses the customer’s query, links can be useful in two ways. First, it can save your rep time explaining something that’s already detailed on an FAQ page or in a video, blog or other webpage. Second, the customer is more inclined to share that content with others. Consider these examples:
When I started this project, I assumed that if any company was ahead in the social customer service game, it would be a major brand from this group. Instead, these brands responded to a mere 14 percent of the 280 tweets delivered during the race. Whether the issue is one of strategy or technology, brands are still far from meeting customers’ expectations on Twitter.
Shawn Campbell created the thumbnail image. All brand names and logos are trademarks or registered trademarks of their respective owners.