Last week my son was scheduled to take a train from Brattleboro, Vt., to Philadelphia, Pa. On the morning of his trip, he texted me this message:
I got a computerized message from Amtrak. They said there's an issue with my train. I don't know if it's late or cancelled or what's going on.
Had he checked Amtrak.com? Yes -- the story there: "Status unavailable due to service disruption."
I tuned into @amtrak on Twitter. There was no information about this incident, but the flow of recent messages about other travel delays suggested it might be useful to tweet this inquiry:
@amtrak: Got an alert about train 55, Vermonter, but no status on website "due to service disruption" -- is it just late? Cancelled?
That was at 9:53 a.m. Eastern time. While monitoring @amtrak replies, I dialed the 800-number, which (as I expected) reported high call volume and a 15-minute wait. I didn't hang out that long because I was sure the response would ultimately be the same as on the site: "Status unavailable due to service disruption."
Eleven minutes later, at 10:04 a.m., @amtrak replied:
@judell Tr 55 is originating in Springfield, MA due to a freight derailment near Essex, VT. 2 ontime buses from St. Albans to Springfield.
That's exactly what my son needed to know. Brattleboro, Vt., wasn't in the cards that day, but he could head down to Springfield, Mass., and catch his train to Philly there.
Over the next few hours I monitored Amtrak's site. The information I'd gleaned from Twitter never appeared there. Nowhere did the site recommend @amtrak, though it was the only source for information crucial to anyone planning to take the Vermonter from Brattleboro that day. (The story about the derailment didn't appear on the Burlington Free Press website until 4:27 p.m. Eastern time.)
Amtrak's use of Twitter illustrates one of the core tenets of effective online communication: Deliver every message to the widest possible audience. This isn't about appealing to a hipster demographic. Twitter is a one-to-many medium, unlike Amtrak's 800-number, which -- if you push through the pre-recorded messages -- is one-to-one. One-to-many is a scalable architecture. The answer I received was available to everyone else planning to take train 55 from Brattleboro. That's powerful leverage.
Of course, Amtrak's website is a one-to-many channel, too, but it failed in two ways. First, it never answered the question in time for travelers to adjust their plans. Second, it didn't refer travelers to @amtrak which could (and did) answer the question in time.
I had a similar experience recently with @united. Its online reservation system made an error for which I wound up being charged an extra $300. The customer service agent I reached by phone acknowledged the problem but claimed to have no authority to fix it. I filed an issue on the website to document what had happened, then escalated to Twitter. Guess what -- @united responded the same day and authorized a refund a few days later.
I suspected that my follower count on Twitter might have been a factor. I'm no celebrity, but if 5,000 people might see my complaint, would that make a difference? Maybe it did. When I scan @united replies, though, I see no obvious favoritism toward the more popular of the 35,000 Twitter accounts that @united is following. (It would be an interesting exercise in data analysis, though!) Like @amtrak, @united seems to be both remarkably responsive and far more effective than traditional customer service channels.
Also like @amtrak, @united isn't mentioned on its parent website. Why not? Maybe because it's so painful to admit that the architecture of those sites is wrong for the purpose. The conventional systems are centralized, process-heavy, and impersonal. That's the antithesis of the lightweight publish/subscribe architecture and personalization that make Twitter such an effective channel for fast-breaking news and near-real-time collaboration. You shouldn't need to know the secret handshake to use these excellent customer service operations. But for now, you do.