We conducted a research to see how often someone tweeted about a page powered by CiviCRM in the past month and a half. We analysed 858 tweets by 612 users about 163 sites, some big names like oxfam or the red cross, some for tiny organisations.
These sites cover an amazingly wide range of topics: homeless, food, transgender, environment, sport, political parties, pets, public health NGOs, independant movies, gender equality, education, cancer, anti weed prohibition...
For this research, TTTP developped a robot that automatically searches for tweets containing a link that is like:
That works for the various urls generated by CiviCRM (drupal nice or normal urls and joomla). We only looked for usually public pages and obviously only for sites that were mentionned in a public tweet.
2/3rd of the tweets where about events (549), much more than donation (229), petition and profile. We didn't search for the last two at the beginning of the reasearch, we'll see how they evolve in our next report, but it's interesting to note that petitions seems to generate much more tweets per site than the other types (more on that later).
Moreover, we only analyse "out of the box" civicrm. The big ones (eg the donation for wikipedia or a campaign against lobbying we developped that generated more than 800 tweets in a month) have custom urls that are undetected by our robot.
We were not expecting to have less tweets during the week-end than the rest of the week. Unfortunately, the computer running the bot was switched off for a few days because I was busy with other issues around the 18th and as twitter doesn't save the tweets (at least the search feature doesn't work for tweets older than a few days) we couldn't fetch them in time.
The drop on the 4th of july seems to confirm that most of the user are in the US.
We will produce these reports again (we aim a monthly frequency), we'll see how it evolves.
Looking at the TLD, that's mostly .org domain names, but as well some geographical ones (ca, nz, fr, eu, de ...). However, the vast majority of the sites seems to be for US organisations.
We kept the most visible sites (more than 5 tweets) to see those using best twitter
Even so the majority of the tweets are about events, the first one is a petition whyprohibition.ca and the second for donations northcentralredcross.org. The other top performers seems to have several events promoted, and they seem to do a lot of self promotion (tweeting about their events) on the top of having their supporters tweeting/re-tweeting about them.
As we only had two sites having a petition during the period, that's hard to have a definitive conclusion, however, I am pretty certain the main reason petitions are more visibles is because that's the only part where we automatically generate a tweet button (and facebook like).
Most of the tweets for events are "home made", like
But most of the petition ones are
(this is the default text, not good and should be personalised, but obviously better than nothing)
We need to modify the event registration and donation "thank you" pages and related email templates to contain by default a tweet button and facebook like too (patch welcome).
We did try to find how often the users promoting a site where the owners of that site. Looking at the profile of the users, we tried to see if their url is the same as the url of the site they tweet about.
Out of the 848 tweets, 118 were "self promotion". Out of the 163 different sites, we only identified 49 "owners". We need to teach the bot to match subdomains, eg. being able to match sfcb.org a with www.sfcb.org and support.bcrfcure.org with www.bcrfcure.org, but
if you use civicrm and twitter and organise event, do fundraising or have a petition, please could you be sure:
As seen, having "FB like" & tweet buttons seems to help the promotion of a petition. We need to add them as well to donations and event info.
If you don't shout loudly against this improvement, we will likely add it during the code sprint after civicon london (register now). We'll monitor the result on twitter.
Beside being able to monitor part of urls in the tweets, our bot is also able to to search on hashtags or any word. If you think it could help your organisation, please contact me see if we can work together.
Moreover, our bot is kind enough to automatically create a contact for each twitter user and add her tweets as activities, so it might be a nice way of knowing better your supporters from within your own civicrm and see how they interact with you on twitter too. Again, contact me if you want to try adding social features to your civicrm install.
Last but not the least, we are developping a twitter app and a new type of activity in civicrm: "tweet" to be able to send a direct message from civicrm (needs some more hours of dev, but almost there).
I browsed the list of case studies and organisations that registered on civicrm.org, and it seems that a lot of these 163 sites haven't registered themselves or written a case study.
I am tempted to contact them (sending them an "@Reply happy to have you as a user, please register your site if not done already...")
Do you think it's too spammy do to it automatically? Do you have suggestions for a nice message?
In conclusion this research is quite exploratory and was a good opportunity to learn about using twitter api and developping an app, using R (to do most of the datamining and graphs), if you have ideas and suggestion on what we should mesure or display in the next report, please share them with me. If you have founding resources or time to code, you are even more welcome to contact me and see how we can put more social features in civicrm ;)
Update: Another visualisation of these data.