A tale of four police twitter accounts
Background
The recent riots and incidents of violent disorder in London and other cities will have far-reaching consequences for public policy, policing tactics and many other areas of life. Big shifts will occur.
I want to focus in on one, quite small, area of change: the deployment of social media in an emergency. As a first step I have looked at some statistics on the twitter accounts of four police forces. There are many caveats to this work and I really don’t want to be seen to be drawing conclusions, just raising interesting observations.
I’ve looked at the corporate twitter accounts of The Metropolitan Police, West Midlands Police, Greater Manchester Police and West Mercia Police. The first three police forces had to deal with serious incidents of disorder the last one did not. I used the service twittercounter to generate statistics for the number of followers per day and the number of tweets per day between 5th and 14th August inclusive.
Quick reminder of some key dates
- 4th August: Mark Duggan shot dead by police
- 6th August: Large protest at Tottenham Police Station followed by riots in Tottenham.
- 7th August: Riots in several parts of London
- 8th August: Riots in several parts of London, Bristol and in Nottingham
- 9th August: Riots in London, Greater Manchester, West Midlands, Merseyside continue into the morning of 10th August
- 11th August: Situation seems to calm
Metropolitan Police
The least active of the twitter accounts I looked at. In fact, as I wrote here, the twitter account made no reference to the disorder until Sunday lunchtime. Followers grew rapidly between 9th August and 11th August and more steadily to 14th August. Followers grew from roughly 27,500 followers to 40,000 by 14th August. Tweet volumes peaked on 9th August when just under 30 tweets were sent.
The account was used to broadcast information about the police position, the number of arrests and to share statements from senior officers.
There were a few dynamic updates of police actions as they happened most notably on 9th August.
“We are responding to disorder in Ealing, Clapham and Peckham. Lewisham appears to be calming down at present.”
“Rumours of violence against mosques in Ilford and Tower Hamlets are untrue. We have had no reported violence against any mosques.”
Were retweeted by over 100 people (according to twitter’s internal system).
The Met Police account didn’t engage in public with individuals (ie there were no @ messages in the timeline)
West Midlands Police
A quiet corporate account under normal circumstances the number of tweets spiked sharply on the 9th and remained at a high level for the next three days. Follower numbers also jumped between 8th and 9th and then increased up until the 12th (to stand at a little under 25,000.
This account was also used to broadcast information about the police position etc.
On the 10th the corporate account encouraged people to follow more locally-focused accounts. And retweeted this from South Birmingham Police
“At the Test Match? Fantastic event for Brum, we’re delighted to host the fixture. This event will not be overshadowed by mindless thugs.”
The account virtually never engaged with individuals in public. The bio on the account profile indicates that they won’t.
Greater Manchester Police
Tweets from this account leaped up between 10th and 11th August in response to the disorder in Manchester. Follower numbers had increased significantly in advance of this (from around 45,000 on 5th August to around 70,000 on 10th August. They jumped again between 10th and 11th August finally peaking at just under 100,000 followers. By far the most followed of the accounts I looked at.
This account was very different in tone and application. It was much less formal. There was widespread engagement with individuals in the public stream. The account provided live updates on the policing situation even though it was dynamic and kept changing. The huge increase in activity was post riots when the police began to tweet details of those charged (and convicted) of offences related to the disorder. This was controversial and the account engaged with individuals over this in the stream:
@bmckee15 we publish people’s DOBs so that there can be no mis-identification with someone of the same name
Lot of debate about publishing details – courts very clear, justice should be done publicly
On the 13th the account started to invite feedback and provided a commentary on this
“really appreciate constructive comments about consistency of tone etc as we have different tweeters. We will definitely look at this.”
West Mercia Police
West Mercia covers a large, fairly rural area (including the area I live in) and a large number of communities who border Birmingham (in North Worcestershire). West Mercia used their account to spread calm and to reassure their citizens. Tweets ramped up from a low of around 10 on 7th August to a peak of 80 on 10th August. Follower numbers saw their biggest jump between 9th and 10th August (from 2,000 to just over 4,000). Follower numbers then grew again to reach around 5,500 by 14th August.
West Mercia had a much easier job, keeping calm and trying to control rumours. The account was handled in a similar way to Greater Manchester, reasonably informal in tone and engaging with individuals in public.
On the 10th persistent rumours circulated that riots were coming to Shrewsbury.
“Again, it appears the #Shrewsbury rumour mill is going into overdrive! We have no incidents to report that would suggest rumours are true.”
An impending EDL March was also causing concern (the March was eventually banned by the Home Secretary and a static protest occurred instead).
“@graemem12 Police will be present in the Wellington and Oakengates area in preparation for any anticipated protests.http://ow.ly/620aE”
“Final tweet of the night – all remains calm across West Mercia. Thanks for your messages and support. We’ll continue to update you tomorrow.”
What does this tell us?
Well I’m wary to draw too many conclusions. I really think we could benefit from a much more comprehensive review of the use of social media across this time period. Many other social media accounts from other cat one responders and from local agencies were active and ultimately we need to be able to assess the outcomes against the actions.
We can see that there is an appetite for receiving information via this channel. In every case twitter follower numbers increased sharply across this time period.
My personal view is that the more informal and engaged approach of Greater Manchester and West Mercia should be the preferred mode. This clearly has resource and training implications. Not just for the staff actually operating the twitter accounts but for the incident commanders who need to understand the way this virtual and ephemeral world interacts with the very real physical world that they are concerned with.
Here I asked if we need new doctrines? I am happy to modify that to stating that we do need new doctrines.
And more than anything we need to learn from this experience.
Comments
7 Responses to “A tale of four police twitter accounts”
Got something to say?







Twitter
Buzz
Facebook
Flickr
FriendFeed
LinkedIn
RSS
You haven’t mentioned two key incidents with regard to the @GMPolice account – the ‘gloating about punitive sentencing’ tweet*, which was widely criticised and which they subsequently deleted, that was posted shortly before the “really appreciate constructive comments” tweet, and also the case of an arson suspect whose home address they tweeted, who was subsequently found innocent, but whose own home was burnt down in the meantime – presumably by vigilantes armed with the info from GM Police.
* “Mum-of-two, not involved in disorder, jailed for FIVE months** for accepting shorts looted from shop. There are no excuses!”
** Sentence completely overturned on appeal, by the way.
Hi Denny
Thanks for the comments and highlighting these points. I do think that responders need to learn from the experience here and improve their training and exercising.
Managing an appropriate tone is going to be difficult in situations like this. The deleted tweet you refer to is an example of getting it wrong. My understanding is that this sentence was reduced rather than overturned but I’d appreciate a definitive link on this. Not that that is germaine to this point.
The other case is more difficult I would say. My understanding is that the details of people who have been charged are normally made public. I saw a post on liberal conspirancy (link tomorrow) suggesting that it was not appropriate given the context. Subsequent results seem to suggest they were correct. On the other hand allowing the state to supress details of charges they have brought against citizens seems to be very troubling. This is an area we need to negotiate as a society
. We should definitely not just leave it up to individual police forces/officers.
Excellent analysis.
It’s probably worth noting that the more local corporate, and individual officers’, West Midlands Police accounts are very good at engaging with people.
And I hope the excellent @WMPcsidogsmithy counts…
Thanks Andy
I’m very concious that the police twitter world (let alone social media as a whole) is much more complex than this simple analysis would suggest. And clearly what matters is outcomes for people: are they safer? Do they feel safer? My hope is really to contribute a little. I do think someone should fund a fuller piece of work here though.
Hi Ben, interesting research. We are currently pulling together all our stats from that week, when we felt social media really came into its own in helping us get real-time messages out to the public and respond directly to their questions/concerns.
We’re doing a detailed analysis, but headlines at the moment are, before the disorder we had just over 3,000 Twitter followers (@nottspolice). As we began to use the account to communicate more information and the hunger for information grew, we saw our followers rise to 16,500.
On Facebook (www.facebook.com/nottspolice) our fans rose from 700 to 6500 in that week and we used this platform for getting out messages too.
We tried to respond directly to as many questions as we could, particularly if the person expressed genuine concern for their safety. We found Twitter and Facebook to be really helpful in putting to bed rumours that had no foundation.
We received well over 10,000 messages of support in that week for our officers and staff, both on Twitter, Facebook and through our force website.
I know you used Twittercounter, but the stats on there for our account are massively inaccurate, they just seem to have graphed an average.
We’re using Twoolr to analyse our accounts. What’s interesting is, at the height of the disorder, our ratio of broadcasting to conversation tweets changed – 79 of our tweets on the Wednesday were for conversation (@replies) and 66 were for broadcasting (linking back to news on our website and incident updates). Normally the broadcast element is much higher, and this is something we want to change to introduce a higher element of conversation.
The use of social media had a massive impact on our website visits – compared with the previous week total visits were up by 1058% and visits from mobile devices rose by 2700%.
I’d be pleased to share our research with you when it’s complete, but we’re certainly going to be using it to learn what could have been done better and to inform our plans for other critical incidents the force will face in future.
Hi Jo
Thanks so much for such a helpful comment.
Those headline figures look really interesting. I guess one question for all these accounts is what to do with all those followers: will they stop following, pay less attention or be newly engaged with the force’s corporate presence?
10,000 messages of support is really something.
I hadn’t come across Twoolr and I agree there are some odd things within the twittercounter stats. I guess followers especially requires the service to query the account on that day, so if it didn’t, it draws a straight line between the data points it does have.
I’m really pleased to see you were using facebook. I worry that this is being neglected by some agencies and yet for many people this is where rumours, and more sinister messages, circulated.
I guess the (not terribly surprising) top line from all of this is that social media is pretty important in emergencies.
I’d love it if you would share your detailed findings. It would be great if we could find a way to pool the learning from all this so that other responders, not just police forces could benefit.
From the very early stages Greater Manchester Police recognised the benefits that using social media could bring. At its height we had 101,000 followers and more than a million people viewed the Flickr most wanted site.
I take on board Denny’s issues as we don’t always get things right and it is about learning and moving on. in the case of the mother of two who had her sentenced reduced on appeal, we deleted the tweet that had caused the problems, and apologised. We did not delete it to hide anything but because it had caused offence.
I hope that in recognising police forces are using social media people will accept that some times we get things wrong. We have taken feedback on board, and have been grateful for the thousands of messages of support that we have received.
Best wishes
Amanda