Anticipate crises, bad buzz, trends and campaigns on social media with AI Sentinel

In Monitoring and Social Listening, the detection of bad buzz, crises, positive buzz and any increase in mentions is essential. Whether it's for crisis monitoring, campaign follow-up, speech impact analysis or reputation analysis, it's essential to understand and detect increases in mentions of your brand, your products, your executives or those of your competitors.  It's important to anticipate these variations.  

We have previously discussed the advances in artificial intelligence that are improving monitoring and social listening tools, while at the same time facilitating the task of experts in the field.

Artificial intelligence, and in particular machine learning and deep learning, are playing a key role in the increased automation of natural language processing, the formulation of queries and expressions, the evaluation of emotions, the categorization of messages, data and publications, as well as visual listening.

Today, we continue to explore innovations with the adoption of artificial intelligence technologies dedicated to forecasting trends, viral phenomena, crises and changes in media and social media campaigns.

1. Analyzing unusual peaks with AI and ChatGPT

Digimind already offers you a virtual assistant to help you detect and analyze peak mentions.

As a watchdog or social listening analyst, communications and PR professional, it's not always easy to detect "real" spikes in mentions. With the help of AI Sense, Digimind's proprietary AI, and ChatGPT, the virtual assistant detects and explains peaks quantitatively AND qualitatively.

Artificial intelligence, thanks to machine learning, identifies unusual peaks of mentions in all the messages, publications and articles collected by your monitoring or social listening tool.

2. After observing a sudden increase in activity, the next step is to have the artificial intelligence analyze the reasons behind it. This includes examining all the possible ways in which the activity may have spread to provide an explanation for the spike. It is important to note that this explanation may involve multiple factors, as various mechanisms can contribute to the spread and virality of content.
   

To do this, artificial intelligence first provides you with quantitative indicators:

• Change in volume of mentions in % growth
• The number of mentions using a hashtag related to the topic generating the spike
• The number of mentions repeating concepts numerous times, An unusual level of interaction
• Next, it provides other, more qualitative information about the transmitters, such as the influencers and opinion leaders who share the topic, as well as the media and news sites that contribute to its dissemination

3. "Thanks to the combined power and functionality of Digimind and ChatGPT AI, you'll get a detailed analysis of peak activity. You no longer need to sift through messages or articles to grasp the essence, implications and consequences of an emerging trend, buzz or potential crisis. A summary is automatically generated for you, explaining in a few sentences the reasons for these spikes."

The different factors behind a spike - by AI Sense and ChatGPT - in Digimind Social 

• Intelligent e-mail alerts: AI Sentinel 

Digimind now offers you a new module to perfect and, above all, better anticipate sudden changes in mentions

 AI Sentinel will automatically detect any current or anticipated increase in mentions linked to your query and send you an e-mail alert, enabling you to respond rapidly to trends, crises, buzz and any emerging events.

• AI Sentinel for monitoring and social listening: how does it work? 

Configuration:

All you need to do is set your alert to the queries of your choice, choosing the "AI Sentinel" type. 

You need to select the recipients and also your time zone so that you are alerted according to your location. You can also specify the countries and languages concerned by your alert to fine-tune your notifications.

You're now ready to receive alerts as soon as an unusual spike in mentions begins to form. 

Example of alerts - Qantas

Let's say you're in charge of social listening for your competitors in the Asia-Pacific airlines sector. 

For example, you receive an alert about the Australian airline Qantas, which has announced a major overhaul of its Frequent Flyer program, offering more booking options.

(*) A mention concerns any publication on the web and social media: a post (or tweet), a comment, a blog article...

You may be thinking that, by social media standards, the volumes displayed here are relatively low: 12 mentions, 27 mentions. Well, that's to be expected. The whole point of AI Sentinel Alerts is to anticipate peaks in formation, to give you time to take any necessary action. This alert concerns the April 8 peak below, displayed as it appears at the end of the day. You have received the alert corresponding to the yellow zone here, i.e. at the start of the peak's anticipated formation.

AI Sentinel alert at the start of the peak (in yellow)

By the end of the day on April 8, the indicators for the fully-formed peak (C) were of a completely different nature: the peak reached over 1,000 mentions with more than 600 interactions on the initial tweet of the trend. We have thus gone from a 44% increase above normal in the alert to a 99% increase, i.e. a doubling of the average level of mentions.

Detection of peak mentions on Qantas and warning of the onset of peak C

This first example concerns a commercial offer that is picked up significantly by Internet users. Of course, the alert system can also warn you of negative information that could damage your brand's reputation.

Starbucks

This is the case of Starbucks, one of the brands singled out for criticism in connection with the Israel-Hamas conflict. Here, the alert concerns the call to boycott Starbucks because of its alleged support for Israel and its treatment of trade unions. Some point to the ineffectiveness of boycotting a single company.

The alert appeared at the start of the peak on April 3, when the volume of mentions of the boycott call, frequent in recent weeks, showed a 65% increase. All in all, the peak represents a sharp increase in volume of 840%, or more than 46,800 mentions.

Alert on the beginning of a spike concerning a call to boycott Starbucks

III. Applications of AI to analyze and anticipate peaks and trends

Analyzing developments and trends is essential for competitive intelligence and social media listening. Here are a few examples of how we use automatic analysis of peak mentions and their causes:

Brand and personal reputation, Crisis detection :

Understanding Reputation Drivers and Stakeholders

Peak and trend analysis can play a crucial role in understanding an organization's reputation.

Reputation Drivers :

• Identify the key factors and influential stakeholders that influence a brand's overall reputation. Peak mentions can reveal the most powerful topics that help shape this reputation.
• Analyze the dominant concepts associated with peak mentions. Which themes recur frequently and have a significant impact on brand perception?

Crisis management :

• Detecting the onset of a crisis: When a spike in negative mentions occurs, understanding the causes (e.g. a defective product, a scandal, etc.) is essential to reacting quickly.
• Monitor the spread of messages: Which media, influencers or channels amplify the crisis? What are the distribution channels?

1. Virality in Crises:
   • Identify sensitive topics that generate spikes in mentions. Which topics have the potential to go viral?
   • Understanding the causes of renewed buzz: If a crisis subsides and then resurfaces, what are the factors that reignite it?
2. Anticipating Reputation Risks :
   • Watch for recurring spikes from competitors or other industry players. These spikes could signal similar risks for your organization.
   • Analyze these potential risks and take steps to anticipate and manage them proactively.

 Early detection of misinformation and fake news:

• The spike in mentions can be a reflection of the spread of misinformation and fake news. These types of messages are highly share-generating due to their often sensational and outrageous nature, and to their characteristics: they reveal information that is supposedly hidden and censored by the mainstream media, and exploit ambiguous, imprecise or complex information to draw alternative narratives or counter-current but erroneous conclusions.

Predictive peak analysis helps to decipher and expose some of these mechanisms before the buzz builds up and drowns out the generating elements: original messages, original sources, synthesis of distorted or false information.

Campaign analysis :

• Detect the first impressions and shares of a marketing or sales action, or of an offline or online campaign.
• Examine key elements of a campaign, such as shares, hashtags and interactions.
• Automatically differentiate between the usual organic mentions of your brand or community and mentions linked to sponsored content (e.g. hashtags).
  
  

Consumer Insights :

• Detection of peaks in consumer messages about a product, service or brand statement, highlighting their feedback, criticisms, needs and wishes for improvement.
• Understanding consumer message peaks in reaction to a brand campaign, a competitor campaign or a change in the environment (political, regulatory, economic).
• Analysis of viral concepts and consumers' preferred themes around product launches or communication campaigns.
  
  

Competitive intelligence

• Analyze the topics that generate peak activity among your competitors.
• Analyze their hierarchy and visibility, and weigh their importance according to their spread and volume.
• Topics can include sponsorship, M&A, governance, new products, ethics, news and more.
  
  

For example, the combined AI of AI Sense + ChatGPT will automatically explain and synthesize diversified peaks among your competitors.

Example: Over 30 days, 4 peaks of mentions are generated for Ford with very different drivers:

Boeing

Miscellaneous / conspiracy: John Barnett, a former Boeing employee, was found dead after testifying against the company.

Corporate Governance : Boeing's CEO, David Calhoun, has announced his plans to step down at the end of 2024

Corporate Governance: David Calhoun, the outgoing CEO of Boeing who left under a cloud, is set to receive a $45 million 'golden parachute'.

Ford 

1. Product: 

1. Corrporate and CEOs: Elon Musk and Jim Farley, CEO of Ford, exchange congratulations and kind words on social media, highlighting their respective teams' hard work and success.

Flagship: On March 12, 1982, the Ford Ranger pickup was launched in North America, and continues to be a customer favorite and one of the most popular vehicles in the world. It is suggested that the Ford Ranger be upgraded with an electric powertrain, underlining its longevity and potential for ecological adaptation.

1. Sustainablilty and Environment : Jim Farley, acknowledges the ambitious and challenging nature of the EPA's final rule, emphasizing the need for public-private cooperation to meet these goals
2. Event: Lexie Alford has set a new record by being the first person to circumnavigate the globe in an electric vehicle, specifically the all-electric Ford Explorer.
3. Strategy: Tom Maoli, owner of Celebrity Motor Car Company, praises Ford's decision to delay EV launches, stating that the minerals required to build one EV could be used to build 25 hybrids.

• Innovation: The Mach-E GT 2024 is the fastest production Mustang Mach-E ever, with improved quarter-mile and 0-60 times thanks to a performance upgrade. The message also includes a playful request to test the Mustang Mach-E Rally at 5 Mile Pass in Utah, promising to be gentle with the vehicle.

Investment: stakes in crypto-currency companies.

Sports: UEFA Champions League Round of 16 at Frankfurt's Deutsche Bank Park.

Business: Deutsche Bank has severed ties with a Selfridges co-owner.

Trend detection

• Examine the creation and propagation of spontaneous popular hashtags linked to social and cultural movements, protests and contestations.
• Track the emergence of trends over weeks and months by observing the progression of concepts and hashtags.
• Compare peak consumer, lifestyle and consumption trends by period and country.
  
  

Stakeholder influence analysis

• Examine influential sources that help create spikes, such as mainstream media, alternative media, news sites and macro-influencers.
• Analyze the virality of influencers according to their speaking or campaign theme.