Sentiment metrics have become an industry standard for reporting social media insights.
Sentiment metrics can be very useful to certain marketing functions, but too often, sentiment is neither informative nor actionable. Before your next report, ask yourself: Why exactly are you measuring sentiment?
Let’s back up a bit. In social media analytics, “sentiment” typically refers to a form of text analysis that measures the percentage of positive, negative, and neutral posts in an online conversation. It’s simple. It’s high-level. It’s easily digestible, and it comes standard with most tools on the market today. But it also lacks depth and context on the subject of analysis.
In that way it is similar to a standardized test score that’s meant to quantify a student’s aptitudes. Like standardized test scores, sentiment metrics are useful for broad, quantitative comparisons. Yet these tests don’t tell the whole story about a student.
The same is true of sentiment scores in a marketing use case. Sentiment can absolutely help answer a few, baseline questions, such as:
1. What should I include in my weekly social media scorecard?
Your weekly social media report requires consistent metrics that leaders can easily digest. Sentiment categories are intuitive and broadly applicable over time. The perfect fit.
2. How can I add more depth to my report?
Here’s a scenario: You’re collecting metrics (RTs, @ mentions, etc.) on a new Xbox One giveaway on Twitter. But how was the campaign received? Does an excited consumer voice emerge amongst the link sharing? A measure of sentiment is a great way to assess positive versus neutral campaign reception.
3. How can I see a summary of conversation about my product?
You don’t want to go into the weeds with data. All that you need is a high-level story. You don’t care that consumers “can’t wait” to buy your new shampoo or that they are complaining about the in-store availability. Here, context is not a priority.
4. What is the most liked and disliked brand in my industry?
Sentiment analysis uses the same categories across each subject of analysis, which lends itself to easy brand comparisons. Map sentiment trends to tactical marketing activities or other events for further clarity.
Despite these advantages, sentiment can quickly exhaust its usefulness. It’d be more difficult to answer these questions with an assessment of sentiment:
1. Why did conversation about my product spike on Wednesday?
Sentiment is a great starting point, but the answer you really want requires deeper measures than positive, negative, and neutral proportions. Sentiment will not tell you why the product was a hit or what had caused so much confusion among consumers. Advanced text analytics help you discover topics and themes in a discussion that add context to the sentiment.
2. Is my firm perceived as being innovative?
Effective messaging is particularly hard to gauge with sentiment measures. Positive, negative, and neutral measures will not reveal whether people think you are eco-friendly, cutting-edge, or lazy. Instead, you have to view the data through a topical lens.
3. Do I always need an algorithm to tell me the story?
Sentiment metrics exist to make analysis digestible. You don’t have time to read every single tweet about Coca-Cola in a day. However, if your brand only returns a couple of posts per week, it may be more efficient to read the posts yourself. As a bonus, you’ll also get 100 percent accuracy in your analysis.
4. How do I measure a macro trend or a national event?
What is sentiment about Cyber Monday? Maybe it’s 80 percent positive, or maybe it’s 80 percent negative, but does that help inspire new ideas for a shopping experience and marketing campaign? No, you want to measure custom-defined categories in line with your creative strategy, the product’s appeal, and what consumers say they need right now.
5. When can I tell my boss that our Twitter campaign is catching fire?
Is sentiment going to tell you that a hashtag is taking off? No. Sentiment is not a good indicator of when your campaign has reached an inflection point. Behavioral metrics are better at measuring the rise in consumer engagement with the campaign as it unfolds.
6. Do more people want to buy our product after we aired our TV ads?
Consumers tweeting that they “can’t wait for” or are “craving” your product are inherently positive statements. Yet positive, negative and neutral classifications don’t provide the detail you want: the intent to purchase. To measure intent to purchase, you need a different categorization scheme.
There is no one-size-fits-all measure for your social media reports.
On a final note, though sentiment surely has its place as a simple, useful measure, it will never answer the question, “Why?” What’s more, sentiment does not tell you what to do next, what to do better, or what your customers want. It measures only how they feel about a particular thing right now.
Today, organizations require a broader toolkit to use social media data to drive strategic decision making. As your social media ecosystem grows and evolves over time, you will probably lean on several different measures to provide your stakeholders with social intelligence; sentiment is just one of many and is not always the best tool.
If you don’t ask the right questions—of social media and your constituents—sentiment metrics will do you a disservice. Positive, negative, and neutral categories, without context or purpose, are just an expensive set of adjectives.
Jehan Hamedi is senior manager of strategic market development at Crimson Hexagon, where Mike Kurtz works as a senior product manager.