Social media volumes for corporates increase manifold in times of crisis. During the recent furore surrounding Pepsi’s misjudged advert featuring Kendall Jenner, there were upwards of 62,000 tweets with the #kendall hashtag alone. Some observers estimate the crisis drove a 2000% increase in brand mentions for Pepsi in the first 24 hours after its release.
And it’s not only consumer brands that see such activity - in the week following the Grenfell fire, one affected party’s feed spiked up to an intimidating 25,391 mentions.
This level of volume poses an obvious problem for any communications team trying to quickly get a handle on what people are saying during crisis times.
While useful to gain a snapshot of activity, a dashboard overview can provide a misleading and undecipherable picture of surges, potentially including thousands of unsorted and unfiltered mentions.
Even the best monitoring software is incapable of independently contextualising the sentiment behind a specific sequence of words with the accuracy of a human being. This is a risk for companies trying to process user behaviour and sentiment, meaning that monitoring software can end up wildly over- or under-estimating the scope of a PR crisis.
Automated software can fail to accurately measure user sentiment for two main reasons:
• In the first place, most monitoring software cannot filter out irrelevant information with the dexterity of a human monitor, skewing the overall picture.
• It also struggles to identify authentic users, meaning further irrelevant content can creep into the analysis. Sometimes at a glance, the human eye can identify if a user is a bot, while as far as the software can tell, they’re just two different users employing the same basic language.
This is particularly troubling when we consider the way fake news has tended to spread across social media in recent months. As a recent MIT study points out, “Accounts that actively spread misinformation are significantly more likely to be bots”. Where these inevitably creep into even the best monitoring software’s accounting, human monitors can root them out at source and identify repeat culprits.
As Paradise Papers revelations emerge, we see several of our clients’ accounts yielding alarming spikes. Daily social media mentions of one usually quiet company during the first day’s news represent almost a hundredfold increase; total mentions jump up to over 9,000 well outside its usual range of 60-120. While we might expect such a surge to be associated with overwhelmingly negative comment, there is actually a surprisingly high incidence of neutral circulation, with major news sources jumping on a trending topic to circulate an eye-catching story.
All this goes to show that in the age of automation, good software is only ever as good as the humans taking control of it and cutting through irrelevance to present the vital information. In a make-or-break moment for brand reputation, a human monitoring service can be the difference between starting to implement an effective communications strategy and being overwhelmed by the crisis.