disclaimer: i’m no statistician or “social media expert”. just a guy and the twitter API.
there’s one thing i think we can all take as a fact: people are increasingly turning to social media to talk to brands, whether it be compliments or complaints or cries for help. as a loyal United flyer (well, no-longer-loyal, given recent changes — see my quick take on everything at the bottom of the post), i’ve watched the @United twitter account get more and more active over the past year (yay!) but also watched as @AmericanAir really engaged my fellow travelers (lucky ducks). one of my complaints to the marketing and social media team at last year’s Star Alliance MegaDO was that tweets would go ignored or take forever to get a reply to.
anecdotally, while UA has upped their twitter game, they still seem slow to reply, and they’re not as engaging as AA. since i don’t have the time to run a sentiment analysis or other lovely things i learned in grad school, i decided to write a little application that would just look at recent replies in each account and see how long it took. of course, this is not the only metric by which to judge a customer service twitter account, but it’s definitely an important one, especially when people who are already underway in their travels use twitter as their first line of communication.
to reiterate, i’m just looking at time to reply, not content or quality of tweets or outcomes.
- retrieved the most recent replies from each account, some time between 11 pm and midnight Pacific time on Saturday, November 2.
- calculated for 96 tweets from @United, 97 from @AmericanAir
- the average time it took for @United to reply to a tweet was 64.07 minutes
- the average time it took for @AmericanAir to reply to a tweet was 10.33 minutes
i don’t think i need to run a t-test to demonstrate there is a statistical difference based on the replies i analyzed (but i did, and there is).
oh, if you’re wondering about that American Airlines spike, that one i did look into because it really stuck out — the tweet that was being replied to did not @-mention @AmericanAir but used a hashtag instead. may or may not be excusable, but at least it’s an explanation. by the way, a spot check of United showed they replied (after 69 minutes) to a tweet that didn’t @-mention them or hashtag.
dismayed at the numbers from last night, i did some spot checking this morning (7 o’clock hour, Pacific) and saw that reply times improved greatly, so i got fresh numbers.
they replied twice to a tweet from two days ago regarding the devaluations (imho at two days you might as well give up or it looks like an afterthought). i took those out:
with the two outlier data points removed, the average time is 45.05 minutes, but you can see that many of the times are sub-10 minutes.
depending on time of day, it can take @United over an hour to reply to a tweet. granted late-night isn’t rush hour, and again, i’m not delving into the content of the tweets, but i can imagine that a good handful of tweets are related to flights about to begin or already underway where critical and timely information is necessary. i’d like to hope that the quick replies are to those who need them, but i can’t be sure. on the other hand, when they’re in their stride, it looks like @United can match @AmericanAir’s reply times (semi-consistently), so based on these two samples, time of day and, i’m assuming staff (specific people or staff numbers, i’m not sure) can have a large effect.
if you need immediate attention, it’s probably best to try to contact them another way — or send off a tweet and then call.
on the other hand, keep it up, American! i’m jealous of the twitter service and attention you provide to your customers. unrelated to reply times: this may be just how @United staff are trained to be, but it seems like they are so stoic. just skim the replies from @United and compare them to those from @AmericanAir. doesn’t AA look more excited and engaged? jealous. maybe United flyers just don’t tweet things @United can get excited over?
all the regular caveats go regarding sample size and data source, e.g., tweets only from two focused points in time, etc. a pretty unscientific study, but at least my (and others’) suspicions have some footing (blurred to protect the innocent, though they’re public tweets). i’d like to hope that this morning’s improvements are not anomalous and this is the kind of response times we will get throughout the day as United continues to improve (i hope) their twitter service.
update in case you’re wondering how they each replied to the blog post tweet…
p.s. in case you’re wondering how i feel about the recent devaluations, i say United is a business and they need to make business decisions. i was expecting the (relatively) merry days of MileagePlus to come to an end sooner or later and i’m not taking it personally (though i wish it was the year after my 1K). they are losing my loyalty but that shouldn’t surprise them given the type of flyer i am. so yeah. it’s a business decision on both our ends. no hard feelings!