If social media platforms were dogs, Twitter would be a Chihuahua – small and feisty and impossible to ignore. And whether your brand has a significant presence there or not, there’s a wealth of consumer insights to be gained from the conversations taking place
Compared to other social media networks, Twitter is relatively small. But what it lacks in size, it makes up for in visibility. What people say on Twitter influences narratives. If these conversations are taking place around your brand or category, you need to be on top of them.
One of the best ways to do that is through platform-specific sentiment analysis. To that end, we’ll explore a Twitter sentiment analysis here – breaking it down like so:
- What is a Twitter sentiment analysis?
- Steps for a successful Twitter sentiment analysis
Here are a few recent statistics to lend context to the discussion:
- Due to its high visibility, you might be surprised to learn Twitter is punching well above its weight. Including global social media networks and messenger/chat applications, Twitter comes in sixteenth in monthly active users at 397 million.
- Because of its easily accessible and public-facing nature, Twitter is great for amplifying conversations. Brands need a solid understanding of the platform since research shows that just a 10% lift in conversation can lead to a 3% increase in sales volume.
- The conversation rate on Twitter increases 56% when a fandom community enters the discussion.
And with that, let’s take a look at what a Twitter sentiment analysis looks like!
What is a Twitter Sentiment Analysis?
Text analytics tools use artificial intelligence (AI) to comb the web for every mention related to your search query, including social media. More specifically, these tools use natural language processing (NLP) to categorize your search results from every conceivable angle.
That means you can run an analysis on your search terms and uncover the different contexts within the conversation. For example, the word good can be positive or negative, like in the case of ‘good riddance.’ The AI assigns a sentiment value of +1 for positive statements, 0 for neutrals, and -1 for negatives. You can then get a sentiment percentage on a scale of -100 to 100 for entire conversations or subsets thereof
Since NLP uses text categorization to differentiate every component of a social media post, including words, hashtags, behaviors, objects, people, and brands, you can isolate social conversations and derive a sentiment value. For example, you can run a query on Twitter mentions of the phrase “avocado toast” and filter it for women in the UK that work in education to see how they feel about it. The point is, no matter how niche your market or query, you can dial in to understand how your target audience feels. Of course, you can keep things broad for an overall focus on emerging trends, current events, or market categories. However, having the ability to take a very targeted approach is great to have at your disposal for a precision sentiment analysis.
So, why isolate Twitter specifically? Twitter is a wide-open repository for consumer sentiment around trending topics, key opinion leaders & influencers, and brands. As such, organizations can use a Twitter sentiment analysis for a consensus on how populations feel about emerging trends, topics of interest, products, a competitor’s earned media, etc. And this is useful because Twitter is an excellent barometer of general consumer opinion
So, whether you’re studying an audience to learn how they feel about relevant topics or getting a feel for market trends, a Twitter sentiment analysis can get you on the mark. Let’s take a look!
Steps for a Successful Twitter Sentiment Analysis
Here we’ll use a sustainability analysis as an example, pulling in the Twitter conversation from the past month. We’ll talk about some of the top things to keep in mind as we go.
1. Cast a Wide Net
When first running an analysis, it’s important to keep your search query broad. That way, you don’t accidentally exclude conversational sub-topics that could be relevant to your organization’s interests. Keeping a wide-angle focus lets you capture the entire discussion. That way, you have a general idea of how the global conversation is playing out.
Here’s what that looks like using a data visualization of Twitter posts mentioning sustainability. This viewpoint gives us a glimpse into the nature of the discussion.
2. Trim the Fat
Where you go from here depends entirely on your business needs and curiosity. Once you have a broad overview, clearing away the bulk of the noise is your next goal. Your wide-angle capture helps illuminate relevant sub-topics and keywords you may not have thought of. It also reveals areas of the conversation that you don’t need in your analysis.
You can either run a new analysis using terminology you’ve uncovered in your high-level research or use the filtering tools within your social listening tools to help trim the fat. You’re not aiming for perfection here but cutting away the obvious clutter from things such as sales pitches, spam, and irrelevant or hijacked hashtags.
3. Capture Summary Metrics
As part of your Twitter sentiment analysis, you’ll want to capture the summary metrics (including a global sentiment score) around your research. Doing so is critical for tracking movement over time. For example, here we have summary metrics for our sustainability query showing mentions and post counts, net sentiment, passion intensity, and potential impressions.
We can see below that the Twitter sustainability conversation stands at a healthy 72% over the past thirty days. And potential impressions are close to a billion. As such, it’s safe to say lots of people are getting influenced by the positivity surrounding this trend.
4. Explore the Scope of the Conversation
From this point, it’s essential to explore the major and niche narratives within your topic for a deeper understanding. And since we’re out to learn how people feel, coloring our contextual clusters for sentiment helps us differentiate areas we want to explore. For instance, we can look at what’s driving negativity in food sustainability and waste. Or we can target neutral conversations in the blockchain cluster to understand fence-sitters. The sky’s the limit.
5. Fine-Tune your Analysis
If this is as deep as you want to go with your Twitter sentiment analysis, you’ll have a broad overview of volume, themes, and general sentiment summaries. However, if you want to go deeper for targeted consumer insights, you’ll want to pull out your filtering and tagging arsenal.
Say, for example, we want to explore the intersection of technology and renewable energy within the sustainability movement. In that case, isolating these discussions will help us focus in on our interests. When we’ve cleared away the segments we don’t want, we’re free for a dive into the audiences and authors speaking in the space.
6. Investigate Top Authors and Audience Attributes
Digging into the demographics and psychographics (interests, emotions) of the audience speaking within isolated conversations gives a deeper understanding of your subject matter. For instance, we find gender leaning 64% male and ages 25-34 indexing the highest in our tech and renewable energy topic. Audience interests skew highly towards green living.
Taking a peek at top authors is enlightening as well. This is where we find the top influencers and key opinion leaders at work in tech and renewables. Digging through their posts helps us find relevant, ongoing Twitter activity that’s currently pushing the narrative. Who knows – maybe you’ll stumble upon a strategic partnership in the making.
Hey Elon, I am the Industry & Commerce Minister of Telangana state in India
Will be happy to partner Tesla in working through the challenges to set shop in India/Telangana
Our state is a champion in sustainability initiatives & a top notch business destination in India https://t.co/hVpMZyjEIr
— KTR (@KTRTRS) January 14, 2022
And while we’re at it, we could grab sentiment measurements from any aspect of the conversation we’re interested in. For example, after isolating our tech and renewable energy sub-themes, we’ve placed the top locations mentioned on a bar chart colored by sentiment. In this instance, we see California coming in the most negative at 31%.
7. Set Alerts for Sentiment Metrics
Wherever your brand interests take you within your Twitter sentiment analysis, it’s vital to grab these measurements along the way for baseline metrics. This also facilitates comparison down the road.
Additionally, suppose you’ve uncovered conversations where the outcome of a situation could significantly impact your brand. In that case, you can set up sentiment alerts on your analysis to keep you up to date. Or, based on your sentiment findings, you can set your notifications of volume swells within your sub-topic. Either way, setting alerts will keep you in the loop when something you care about is going down on Twitter.
Whether you’re just keeping tabs on a general conversation or taking a deep dive into targeted market research, a Twitter sentiment analysis is an invaluable capability that keeps your brand ‘in the know.’
The beautiful thing – artificial intelligence makes this process ridiculously fast. Gather conversations surrounding emerging trends, current events, categories, audiences, and the competition. The ability to capture your Twitter data at scale and in real-time means you can quickly capture the insights you need. It all boils down to implementing an enlightened strategy before the other guys. Reach out for a demo, and we’ll help you get there.