Thanks for the feedback from my article about our 'Good or Bad?' Twitter monitoring. Several emailers are asking how and when we intend to turn from 'listening' to 'interacting'.
In my meeting with the insight team we chatted about how we might achieve this. At first glance this seems like an easy win: give a trusted group of our customer service staff a Twitter account and let them contact people who tweet about problems with Tesco, and see how we go.
Whoa! My observance of other companies doing this has met with a mixed reaction - in some cases disastrous.
The issues to consider when setting up the official company Tweeter are:
If customers are having a problem with Tesco - can we actually help? If someone tweets that a store has run out of bananas then maybe some quiet behind-the-scenes alerting to sort out the situation could be better. We think we can help but actually can't get the bananas to the store any faster so could we just end up annoying the customer?
On the other hand, if a customer has tweeted their annoyance that a home shopping delivery is getting late, it's possible to ask them to call the customer service centre so we can find out what has happened.
Is there a grey area where we have to judge whether to respond or not?
When communicating with people over Twitter, it's easy to forget that every tweet is broadcast worldwide, then indexed by the likes of Google. Even Direct Messages (DMs) are not sent in an encrypted form, yet it's easy to forget this when asking for identity such as an email address.
At least people who use Twitter generally know of their 'world broadcast' combined with indexed permanence of their tweets. I think there's a psychological boundary that allows them to accept that Tesco monitoring them is kind-of OK.
On the other side of that boundary are places like Facebook where people believe they are in control and that their messages are contained amongst their friends (which is true provided that their account's privacy controls are set correctly) - so Tesco popping up to provide help may actually create negative sentiment.
Say something verbally and the message is mixed with your facial expressions and tone of voice. Strip that context away and leave the written words, and the meaning can be easily misinterpreted. Badly. In email it's called "flaming". On Twitter it could be worse, as loads of people could re-tweet (re-broadcast or relay) your message and you cannot withdraw it. Getting the message - and its nuance - right is key - and wrong / misused / misspelled words can have a highly negative effect. Let me demonstrate:
"I went to Tesco and they had no bananas. Typical!"
Good response: "Hello this is Tesco Help - I see you can't find any bananas in your local store. Can I help you?"
Bad response: "You say you can't find bananas? Which store were you looking in? Did you look in the fruit & veg section?"
The second response is well meaning and, if said verbally over the 'phone, would probably be fine thanks to a friendly voice. But remove this nuance and the message sounds cold, rude, and accusing.
Now your solution may be to have a set of stock messages much like an email response system, but the whole point of Twitter is that dynamic 'live' effect and the opportunity to engage - not a standardised email to be contemplated offline.
The recent nefarious tweet from the official Vodafone UK account reveals the need to think about how to secure our Twitter account. I imagine a team of Tesco Help people in an office away from the general thoroughfare of staff-traffic having responsibility for output through a future Tesco Twitter account. Of course they'll change the password every day, and handover between shifts will take place where the official password will be revealed once the place had been swept for spy bugs… OK but if the Tesco brand is at stake - as Vodafone's was - it's a serious consideration!
What about someone setting up their own "Tesco Help"-sounding Twitter account name and pretending to be us, either to "phish" for personal info or generally try to trash our brand with some suitably obscene messages?
Combine this with my earlier concern that we must not accidentally let the customer reveal any personal identity - indeed directing them to call us so we can take them through the account authentication procedure. Unfortunately, a customer under duress suddenly finding a shining knight wanting to help them may cause them to reveal more than they intend to:
"My Tesco.com delivery is running late and I'm getting annoyed"
"Hello this is Tesco Help - I understand your delivery hasn't arrived - can I help you find out what has happened?"
"Hello Tesco yes please I'm at 15 Sodberry Terrace, NW3 2AB - can you check where the van is?"
…and thanks to us the customer has just broadcast to the world and its search engines where they live. Forever.
So you see, the move from 'listening' to 'interacting' is fraught with challenges - and we must think of everything before we start.