Artificial Intelligence offers marketers amazing new opportunities, particularly in the field of consumer intelligence.
As Jessica Liu, analyst at Forrester, says: “Marketers rely on technology to leverage massive amounts of data to make billions of daily decisions. The amount of information people need to process simply exceeds human cognitive capacity. This creates a giant opportunity gap, and that’s where AI-powered solutions come in.”
In addition to processing billions of social conversations, AI empowers marketers to spot emerging trends and explore new topics much more efficiently.
In this blog post, we will talk about how AI has enabled us to add powerful new features to the classic word cloud.
The Classic Word Cloud, a High-Level Picture of a Marketing Project
Before diving into the improvements we have just released, let’s take a look at this classic feature that’s become a staple of all social listening tools used by marketers.
Word clouds (called “most frequent word clouds” in Radarly) are a useful feature when you want to summarize a marketing project on an insight page and share it as a snapshot with your colleagues to give them a good picture of a marketing campaign or competitive benchmark.
However, when you want to explore a topic to detect emerging trends, this feature seems to be a bit basic and displays quite obvious terms, which won’t help you to spot new areas of conversations within your target audience or specific feedback on a recently launched product.
The Distinctive Word Cloud, a Deep Dive into the Specificity of a Context
What is the ‘Distinctive’ word cloud?
While in a conventional ‘most frequent’ word cloud, the terms being selected are simply the most popular terms in a set, the Distinctive word cloud highlights the terms that have undergone a significant change in popularity measured between a foreground (your context, i.e. the filters you applied) and background set (the entire project). If the term only exists in 5 documents in a 10 million document index and yet is found in 4 of the 100 documents in your context that is significant and probably very relevant to your search.
Let's look at an example to better understand the difference. So you could decide to select only negative sentiment posts to understand what upsets people about your products or services or what have been the triggering factors of a crisis.
Spot Emerging Trends
To give you a bit more perspective, let’s take another example: a brand in the cosmetics industry, offering a range of makeup products. The objective is to explore opportunities taking as an entry point the top-performing brands of the industry.
Last winter, if we were using this distinctive word cloud, we could have spotted a very interesting emerging trend about gender inclusivity:
Here we can see an emerging topic of conversations that happened around Christmas 2020, in the UK, where there was an increase in conversations around makeup for men. It would have been interesting to exploit this opportunity to promote a range of products dedicated to makeup for men and position the brand in this niche.
An industry specialist could have noted that this topic of interest was growing, and we could start a study to evaluate how this opportunity of offering makeup products for men was interesting for the company and potentially invest in it.
Feed Your Content Strategy
If you are using our next-generation audience segmentation add-on, Tribes, you know that you can deep dive into your different tribes and sub-tribes in order to understand better who they are, what are their topics of interest, their shared values and passions, so that you can talk about this and better engage your audience segments.
Using the Distinctive word cloud could help you dive deeper into what issues and themes are most important for your sub-tribes, and dedicate future content to this to address them and engage them more efficiently. A client in the games industry could find that one of its audience sub-segments, the Nostalgeek, was talking about catch-wrestling and could use this insight to fuel its content strategy, where, here, this topic would have been drowned out by most frequent terms.
The Emerging Word Cloud; Spot a Rising Trend in a Specific Period of Time
This word cloud highlights, this time, the most specific terms within a specific period of time. It’s another way to structure the data, focusing on a period. This time, the foreground is a given period of time: if the term only exists in 5 documents in a 10 million document index and yet is found in 4 of the 100 documents within the selected period of time, it’s probably very relevant to your search.
Detect Emerging Trends
This new word cloud helps marketers to uncover new trends and show what is coming in the market. This feature is time-efficient as it shortens the time (and therefore reduces cost) to detect relevant topics in a big-data environment.
Communicate at the Right Time
For instance, let's say that a sports fashion brand is sponsoring the Olympic Games in 2024 in Paris and wants to start a communication campaign when it is most appropriate for sports fans. If it’s too early, the messaging is likely to be drowned by the 2021 edition, but if it’s too late, the other sponsors could be on top of the conversations as they started to communicate too late.
Using the emerging word cloud, you can spot #Olympic2024 as it emerges.
Enhance Your Marketing Activations
Let’s say that you work for a food brand and you need to plan out a campaign on a seasonal food product. You know that your product is particularly popular historically around the winter. Using the emerging word cloud enables you to detect when it’s the most appropriate time to launch your promotional campaigns and inform your content strategy.
When AI Enhances Word Clouds
Today, you are able to choose between several types of word clouds:
- Named Entities
An AI powered ‘Named Entities’ word cloud is based on the named-entity recognition that seeks to locate and classify named entities mentioned in unstructured text into predefined categories such as people’s names, organizations, locations, medical codes, time expressions, quantities, monetary values, percentages, and so on.
To go further, our engineering team is now working hard to deliver another type of AI-based word cloud, the ‘Keyphrases’ word cloud that will allow you, using Natural Language Processing (NLP) to highlight a row of words that constitute a short sentence. It will provide you a much better experience as it will allow you to highlight a sequence of words rather than one single term so that it will be easier for you to get a context and interpret emerging signals. For example, if Keyphrase Extractor were to analyze the content of the book, Alice in Wonderland, it would find terms like “caterpillar” and “hatter”, as well as phrases such as “rabbit hole”, “March Hare”, “Cheshire Cat”, and so on.
This new feature is now available for Linkfluence customers in our Radarly product and Insights Pages.