Twitter customer service works, and it has a blind spot: the customers who never tweet. If you answer only the people who complain where you can see them, you’re listening to the loudest few in the queue.
This year gave us a clean test of that. In April, a cable company started watching Twitter for complaints and answering them by name. By December, the man running it told John Battelle that he managed seven people, was hiring three more, and had followed up with an estimated 10,000 customers since the start of the year. He had over 5,500 followers. That’s a real result for a channel most support managers had barely heard of twelve months ago. I’m not here to knock it. I want to know what it costs the customers who aren’t in that picture.
What did the Comcast Twitter experiment get right?
It went to where the customer already was. Nobody had to find a phone number, sit through a menu, or fill in a form. A person typed a frustrated sentence into a box, and a real person with a name answered within hours.
Frank Eliason, who holds the title Director of Digital Care, told Battelle that he treats everyone the same, whether they arrive by email, in person or on Twitter. He also said he doesn’t wait for a complaint to land on his own account. He searches blogs, help forums and Twitter, and some problems get fixed in a tweet while others move to the phone or email. The point is a good one. The channel is new. The job is the oldest one in support: read what the customer wrote, work out what they want, and follow it through.
It also shows how little most companies are asking for. A fast, human reply is enough to change how a customer feels about a company that has just failed them. Nothing about that needed a new budget line or a new platform. It needed someone to read the complaint and own it.
Who is left out when support moves to Twitter?
The obvious answer is anyone who isn’t on Twitter. Techdirt made the sharper version of the point in April, in its piece on Comcast Cares. The author tweeted about a faulty HD service and had three Comcast employees contact him within a few hours. His question was whether that speed was available to ordinary customers, or only to people with an audience and a search ranking.
That’s a fair question, and every support manager should put it to their own operation. A public complaint carries a cost the company can see. A private one carries the same cost, but it shows up later, as a cancellation.
Do the arithmetic on the Comcast numbers. An estimated 10,000 customers helped in a year is a good outcome for a team of seven. It’s also a small slice of what a national cable provider’s phone and email queues handle in the same twelve months. Twitter reached the people who raised a hand in public. It couldn’t tell you about the rest.
Think about who reaches a company by which route. The customer with a blog and a following gets attention because attention is contagious. The customer on hold at 6pm, who hangs up and quietly switches provider, gets nothing, because there’s nothing to see.
Visible complaints get fixed first. Invisible ones get fixed never.
Is a tweet a service level or a favour?
This is the uncomfortable part. If a tweet gets a same-day answer and an email gets three days, you haven’t added a channel. You’ve built two service levels, and the customer with the larger audience is getting the better one.
I wrote about this chain problem earlier in the year in Internal and External SLAs. A promise to the customer is only as good as the weakest team behind it. A social team that can fix anything in an hour is a promise nobody else in the building has agreed to keep. Sooner or later it lands on someone who didn’t know about it.
The same logic applies to how you rank work. A tiered SLA matrix decides who gets what response time and why, on paper, before the pressure arrives. Audience size is a poor tier. Impact on the customer and the business is a better one. If your Twitter replies are quicker than your top tier, that’s a decision you’ve made by accident.
Before you staff a social desk, write down the answer to one question: what does a customer get when they take the quiet route? If you can’t say, that’s the first thing to fix.
How do you listen to the quiet customers?
Treat the social channel as one input, not the measure of how well you listen. It tells you what people are willing to say in public. It doesn’t tell you what the phone queue, the email backlog and the abandoned chat sessions are telling you.
Start with four numbers you probably already have: abandoned calls by time of day, reopened tickets in a sample of 50, the share of contacts that needed a second touch, and the gap in response time between your fastest and slowest channel for the same type of problem. A spike in abandoned calls at 6pm means your evening customers are giving up, and none of them will tweet about it.
These are the customers who didn’t make noise, and you won’t hear from them unless you go and look. An hour with the abandonment report tells you more about who you’re missing than a month of watching a search column.
Then check how you treat people who are already unhappy. Irate customers are easy to spot, and the ones who are just as unhappy but polite are the ones you lose. A quiet, courteous customer who has waited three days isn’t a satisfied customer. They’re a customer who has stopped expecting anything.
What would a fair channel report look like?
I’d put every channel on one page with the same four columns: contacts received, median time to first reply, share resolved without a second contact, and customers lost within 90 days of the contact. Phone, email, chat and Twitter, side by side.
The last column is the hard one, and it’s the one that matters. It’s also where the quiet customers show up. A channel with a fast first reply and a high loss rate isn’t working. It’s just friendly.
I thought speed was the whole game when I started in support. It’s the visible half. The other half is whether the problem stayed fixed, and no search column on a Twitter client will show you that.
If the numbers say Twitter customers are happier, believe them, and then ask why the phone customers aren’t getting the same treatment. That conversation is uncomfortable. It’s less uncomfortable than finding out at renewal time.
What should a support manager do with this?
Pick one person to watch Twitter and the blogs, the way Comcast did, and give them real authority to fix things. A person with a login and no authority is a press release, not a service. The practitioner community at HDI has spent years on the point that a support channel is only as good as the process behind it, and social is no exception.
Hold the social desk to the same standard as everything else. Same ticketing system. Same follow-up. Same definition of resolved. If a tweet gets answered and nobody logs it, you’ve created a channel that can’t be measured and can’t be repeated.
And keep the numbers honest. Report social volume next to phone and email volume, not instead of them. A dashboard that shows 10,000 customers helped and hides everyone else is a dashboard that flatters you.
The lesson is the one in the hotel soap story I wrote about earlier this year, in Communication and Customer Service. The guest wasn’t ignored. He was answered by four different people and listened to by none of them. A new channel doesn’t fix that. It gets you more places to answer badly, faster.
Twitter is a good place to start listening. It’s a poor place to stop.
Hutch Morzaria is a Director-level CX and Support Leadership professional with 19 years of experience building global support organizations across SaaS, Fintech, and enterprise technology. He has hired dozens of support and CX leaders across his career and holds ITIL Expert certification across V3 and V4.
