One of the brightest guiding lights in responding to a crisis is sentiment analysis. Keeping track of how your crisis is developing can be made easier if you follow and log the sentiment around the event in the digital space. Using social media well in a crisis means knowing when and where you should respond. A sentiment analysis can help you form an effective triage response system.
A basic sentiment analysis reveals much-needed information about where a particular conversation is headed and whether or not it is escalating. It can be done with a spreadsheet that logs the date, poster (username), content, and assigns a sentiment based on a five-point rating. A rating should be given to each post in an assigned column, so patterns can be seen at a glance:
++ = very positive
+ = positive
0 = neutral
- (minus) = negative
-- (double minus) = very negative
Using the above sentiment scale, put together a simple Excel spreadsheet to monitor a specific event or conversation in the social media space. As an example, let's say your company just fired its CEO. The reasons are unclear to the public, and the firestorm is starting to hit social media in blogs, online news, Twitter, and other platforms.
Note: If you have knowledge of the event before it breaks, you can have the sheet formatted ahead and be monitoring the event through your listening dashboard.
You will want a column for the date (and possibly time if it's a viral situation), the name or username of the poster/blogger, a synopsis of the content, and a column for each rating on the scale. I would assign content by topic so patterns are easier to spot at a glance. Here's an example using an Excel spreadsheet:
CEO Firing 4_1_12
Date/Time | Poster/channel username | Content | ++ | + | 0 | - | -- |
4-1-12 | mditt-Twitter | Agrees w/ firing |
| X |
|
|
|
4-1-12 | cybergeek-Twitter | Other issues -rant |
|
| X |
|
|
4-1-12 | Opus2-Small Biz Blog | Agrees w/firing |
|
|
|
| X |
4-1-12 | JackJ-Twitter | Disagrees w/firing |
|
|
|
| X |
(Because the full table didn't reproduce here, please check my blog for the whole table: http://cksyme.org/how-to-build-a-simple-sentiment-analysis-using-social-media-in-a-crisis)
It's important to understand that you are gauging the sentiment towards the company or organization, not whether the posters agree or disagree with your actions. This is where people can stray in crisis sentiment analysis. For instance, you can have posters that agree with the firing, but express a highly negative view of the company. You can also have posters that disagree with the controversial firing but are very supportive of the company anyway.
Ultimately, whether or not a poster agrees with you isn't important, their level of advocacy and support is. You may want to keep an eye on the public opinion about your actions, but the purpose of this exercise is to measure sentiment towards the organization. An angry mob gets ugly fast. Watch that "double-negative" column for escalation trends.
Also, remember that this post is about the CEO firing, so an off-subject rant, such as "cybergeek" above, would be considered neutral to the topic unless a pattern emerges. You may have trolls that use any negative situation about the organization for an opportunity to spew. Don't feed the trolls.
In addition to gauging general public sentiment, this analysis log can help you identify key influencers and critics at a glance. You may need to reach out to these individuals at some point during the event. It may also be helpful to develop a shorthand system for recording content so you can see patterns at a glance. Here's a thorough article from the Content Marketing Institute on how to identify influencers in the social space: http://blog.junta42.com/2011/02/finding-social-media-influencers-market/.
If the content stream is traveling too fast in a real-time crisis, or you do not have a dedicated team to monitor the stream, choose a ten minute recording period out of each half-hour segment for the first 24 hours of the crisis (for sentiment analysis only). Also, keep track of the number of posts per hour. As the frequency of posts die down, you can adjust your monitoring period.
Also, be aware that emerging news may fuel the fire and cause an escalation at any time. Be watching for spikes. This thumbnail sentiment analysis will help you identify patterns. With this information, you can plan when specific types of content should be shared with stakeholders. Employ consistent sentiment analysis throughout the duration of the event to check the temperature of the conversation.
This piece is excerpted from the new e-book, Listen, Engage, Respond by Chris Syme. Click here for a link to the book on Amazon.com.