What is Topic Analysis?

Topic Analysis is a proprietary feature only available through Signal AI, which allows you to discover and understand the connection between organizations and topics. This means you can see how closely associated your business or your competitors are to the topics that you care most about and adjust your strategies to achieve your reputation goals.

How is it calculated?

The Topic Analysis feature calculates the Association Score by judging the likelihood of an organisation and a topic being mentioned together. We do this across every organisation and every topic within our database, allowing you to understand how closely associated your company/competitors are with the topics that you care about or identify topics that you are closely associated with but unaware of.

What questions can this new feature help answer?

In the Topic Analysis tab in the Signal AI Web App, you will find multiple new advanced visualizations that leverage multiple metrics together — Volume, Sentiment, and Association Score — helping you more accurately answer questions such as:

1. How closely is my company associated with the topics we care about (and how do our competitors stack up)?

In this example we are comparing several social media organizations—TikTok, Snap Inc., Facebook and Twitter— around their association with the topic of Consumer Insights. You can see that, while Snap Inc. has the most coverage by Volume (as indicated by the size of the dot), TikTok has more positive coverage and a slightly higher Association with the topic.

2. What industries are my competitors involved in that might give them a competitive advantage?

In this example we are looking at the most closely related topics for Facebook only. (Note: we are zoomed in to show only a small sub section of topics related to Facebook.) You can see here that they have a high association with topics that you would expect them to — Data Protection, Whistleblowers, Extended Reality, Metaverse — but they also have high association, with positive sentiment to Chatbots, not a topic you would necessarily connect with the social media giant.

3. What topics are my organization uniquely associated with which we may be able to leverage to create a distinguished narrative?

In this example we are looking at all of the topics that are uniquely associated with Snap Inc. in relation to the set of competitors we established above. We see that Snap is uniquely associated with Gen Z, while they are not at all associated with Clean Technology which their competitors are. They may use this information to guide their strategy and create a unique narrative.

4. In what areas are we winning? Conversely, what topics are we associated with that require us to engage in our risk mitigation strategy? Are there risks emerging that we weren’t previously monitoring?

In this example we are looking at the most closely associated topics related to Snap Inc. (Note: we are zoomed in to show only a small sub section of topics related to Snap.). We see here that they are ‘winning’ with their current narrative in relation to the topics of Augmented Reality and Gen Z — they have both high association and highly positive sentiment in relation to these topics.

On the other hand, we see that they have a high association and highly negative sentiment in the area of Downsizing — an area where they may want to rethink their narrative and messaging or engage in a risk mitigation strategy. In relation to Data protection and Information privacy, we see they have a relatively neutral association with slightly negative sentiment — suggesting this may be an area where they may want to reposition, or at least keep a close eye to ensure it doesn’t become a larger problem.

Unlike with more traditional tools, where you needed to already know the topics you want to measure against, Topic Analysis would raise this flag around Data protection even if it were not a risk you were already concerned about.

Why is it better than existing metrics for solving these questions?

Addressing some of the complex questions above without using Topic Analysis could result in incomplete or inaccurate answers. For example, using an increase in SoV against competitors on a key topic like sustainability without analyzing both the sentiment and association doesn’t take into account the entire topic—just the companies you’ve already defined—and could inflate a company’s understanding of its perception in this area.

While this approach can be highly effective in comparing similar-size businesses against each other, it may be unfair or inaccurate when comparing companies of differing sizes. Smaller companies will naturally get less coverage in the media. Consider a situation where an organisation ‘Cool Green Startup’ is mentioned only 10 times in the media and all its 10 mentions are on the topic of ‘Sustainability Research’. A large corporate ‘Big Oil’ may have 20 mentions in the media in relation to the same topic, so SoV (Share of Voice) will favour this corporate, even though Big Oil may have thousands of mentions in the media that are mostly about pollution.

Association Score considers all the above, making the measurement more reliable and accurate. By harnessing an Association Score, organisations can gain a more accurate understanding of the effectiveness of their campaigns and their perception in the market on key topics against their competitors and beyond.

What’s the science behind Topic Analysis?

We rely on the Normalised Google Distance (NGD), an information distance metric, to measure associations between entities and topics.

This is why the Association Score is so different from SoV:

  1. The size of the entity—The measure of Association considers how much coverage the entity has overall. If two entities have the same number of articles about a topic and one has more coverage overall, the smaller entity will have a higher association with the topic.
  2. The size of the topic – The measure of Association considers how broad or narrow the topic is. If an entity has the same number of articles about two different topics, it should be considered more closely associated with the narrower topic.
  3. Media bias – The measure of Association should consider the probability of an entity being mentioned with a topic. The media is always biased towards popular subjects (big brands, celebrities, contested topics). Therefore, if a big brand—”A”— is mentioned with a broad topic—”X”— for a certain number of times, say 10, and another small brand—“B”—also has 10 mentions with a niche topic—”Y”— we should consider that the association between B and Y is higher than the association between A and X.

Association Score is illustrated through a score between 0 and 1 — 0, meaning no connection at all, and 1, meaning very closely/always associated. An Association Score of 1 means that every document that mentions the entity also mentions the topic – complete overlap. The less overlap, the lower the score.