How can brands measure customer sentiment?
On the face of it, ‘sentiment’ is one of those subjective terms which is hard to define. How can you assign metrics to feelings?
Dig a little deeper, however, and you find that modern data science makes it surprisingly easy to measure customer sentiment.
Analysing customer sentiment is one of the very best ways to gauge public opinion about an event, a product, a campaign, and even about your brand as a whole. So, it’s important not to overlook sentiment when setting out your analysis strategies.
So, how can you measure sentiment? Why should you measure it? What does a good sentiment analysis tell a marketer?
Here, we’ll take you through everything you need to know about sentiment in marketing – from what it is through ways to measure it and how to utilise the insights you gain.
What is ‘sentiment’ in marketing terms
In marketing terms, ‘sentiment’ refers to the emotion behind customer engagement. It incorporates everything from the tone of your customers’ interactions to the context and the feeling behind each behaviour.
Every interaction – from likes and comments to click throughs and purchase behaviour – has a feeling behind it. You can determine this feeling by assessing the context and nature of the interaction.
Establishing customer sentiment will teach you valuable things about the way your campaigns are being received by your customers. It will show you the context behind their engagement in a way which many other metrics just can’t. And you can utilise those insights going forward to give your customers exactly what they want.
Why measure customer sentiment?
There are several great reasons to measure customer sentiment:
- It gives you deep insight into your audience. Sentiment analysis tells you not just what customers like, but why they like it. It gives you a far more detailed understanding of the way your audience feels about your brand and your campaigns than can be provided by other metrics. You can use those insights to tailor, personalise, and curate things like your marketing tone in ways which will resonate with your audience.
- It works in real-time. Sentiment is a highly variable thing, which can change very quickly according to circumstances. For example, when social distancing measures were introduced to combat Covid-19, content which showed people gathering in large groups rapidly lost engagement. Sentiment analysis flagged this quickly and in real-time. Marketers were able to quickly pull that kind of content and replace it with more pandemic-friendly material.
- It improves customer service. Understanding the factors which are likely to frustrate customers (and why!) helps you to be proactive in customer service measures. For example, understanding that delivery delays are likely to make customers angry enables you to put measures in place to offset that anger (emailing them with a pre-emptive apology for the delay, for example). Good customer service often makes the difference between a stinkingly bad online review and one which glows with praise.
- It’s good PR. Sentiment analysis tells you a lot about how customers perceive your brand. And the better your understanding of brand perception, the better you’ll be able to angle your PR. Knowing how customers feel about you (and why they feel that way) ensures that every message is relevant and informative, and that the topics you’re addressing are important to your audience.
- It lends insight into important metrics. There are three vital metrics which are informed by customer sentiment:
- Overall customer satisfaction
A greater understanding of customer sentiment will tell you what’s behind changes in these metrics, and help you to adjust your campaigning accordingly.
All of this, ultimately, brings you closer to your customers and improves your ROI.
How can you measure sentiment?
Broadly, there are three major indicators of positive sentiment. These are:
1) Willingness to recommend. Word of mouth (which includes things like online reviews and positive use of brand hashtags) is one of the very best ways to gain new customers. But it’s also a fantastic indicator of customer sentiment. If your customers are willing to recommend you to others, they probably think quite highly of you.
You can measure willingness to recommend by working out your Net Promoter Score (NPS).
NPS ratings are usually gained through questionnaires. By asking customers on a scale of 1 to 10 how likely they are to recommend your brand to others, you can group answers into three groups – Promoters (9s and 10s), Passives (7s and 8s), and Detractors (1s through 6s). By subtracting the percentage of Detractors from the percentage of Promoters, you arrive at your NPS figure. The higher the number, the more positive your customers feel about you.
2) In-app ratings (‘love prompts’). Most apps these days come with embedded ‘love prompts’. Periodically, users will be asked to rate their experience with the app, often using a scale of some kind (number of stars awarded, for example). This is a quick and easy way to test the waters of customer sentiment – but bear in mind that context plays a big role in in-app ratings. For example, if your rating prompt pops up at a crucial point in a game or process, this may frustrate the customer, leading to an impulse-rating that’s more negative than would otherwise be the case.
3) Direct feedback. There’s nothing like customers telling you directly how they feel. Things like customer service emails, customer support chat logs and so on can all be valuable in assessing overall customer sentiment.
In order to most accurately measure sentiment based on feedback, we recommend that you find a way to assign scores to that feedback. For example, you could categorise customer conversation by tag (“Complaint”, “Query”, “Praise” etc). Or use feedback forms using metrics which easily translate into data (the 1 to 10 system mentioned above, for example).
Remember – even when your feedback is negative, it’s good for you. It helps you to gain a greater understanding of customer sentiment, which in turn helps you to boost performance across every metric.
The three methods we mentioned above broadly measure positive and/or negative sentiment towards your brand.
You can drill down deeper into the whys and hows of these broad sentiments by assigning tags and values to specific feelings revealed through feedback. Often, however, you’ll get a clearer and more immediate picture of sentiment in real-time by monitoring engagement on your social media.
Social media engagement is a great way to take a sentiment pulse-check. It can tell you how your brand is perceived moment to moment – which is vital if you’re to capitalise on surges in engagement, or to avert crises before they have a chance to damage your brand.
When monitoring your social platforms for sentiment, consider the following:
1) Comment velocity. How fast is conversation moving around your campaign? If comments are coming thick and fast, that’s usually an indication that something has triggered your audience – either for better or for worse.
Keeping an eye on comment velocity is an excellent way to establish when something’s changed in your customers’ sentiments – but on its own it doesn’t tell you much about what’s got them all fired up. To learn more, you need to dig into other things, like…
2) Comment tone. It’s easy for a human marketer to tell whether comments are positive, negative, or neutral. A quick glance at a comment section that’s taking off is usually enough to let you know what’s going on.
However, you can take this further by turning the tone of your comments into data. There are various ways to do this, including assigning values to ‘positive’, ‘negative’, and ‘neutral’, and/or setting up your analytics suite to pick out certain keywords (for example “Love”, “Hate” etc). This will show you trends and patterns in customer sentiment at a glance.
3) Reaction tone. Platforms like Twitter offer a simple ‘like’. As a general rule, the per-follower number of ‘likes’ your campaign posts get are an indicator of how well your posts are going down with your audience.
Things get a bit more interesting when it comes to Facebook. Facebook’s range of possible reactions is much more nuanced, incorporating both positive and negative emotion. Facebook’s analytics will let you sort responses according to reaction, giving you a picture of the specific sentiments your post is inspiring.
For example, a lot of ‘likes’ indicates generally positive feeling. ‘Love’ the same – with perhaps an added element of “Aaaw, that’s adorable!” But be careful. Everything is context-dependent, and the way reactions are used differs from demographic to demographic.
A lot of ‘Angry’ reactions could mean that people are cross with your post – or that they are cross with what you’re posting about. Charity posts about global injustices often experience many ‘angry’ reactions – but anger about these injustices is precisely the sentiment they want to inspire.
Similarly, a lot of ‘Haha’ reactions could mean that people think your campaign content is hilarious – or that it’s stupid. Are they laughing with you, or at you? Comparing reactions with the tone of the comments will give you a clearer picture.
4) Shares and mentions. Shares and mentions work in a similar way to your NPS (see above). However, social media usually gives you a fuller picture of the context around your shares and mentions. Consider, for example:
- The tone of your shares and mentions.
- The frequency of your shares and mentions (are they clustered around a particular campaign or event?)
- The volume of shares and mentions (how does it compare to your brand average?)
How can you use insights drawn from customer sentiment analysis?
Sentiment analysis is useful because it gives you a detailed picture not only about what is and is not working – but why this is happening.
For example, simple engagement stats may not be too useful for a restaurant brand struggling to get customers through the door. Negative engagement looks much the same as positive engagement when you’re collecting basic, non sentiment-based stats.
Conduct a sentiment analysis, however, and everything becomes much clearer.
Assessing the tone of comments and reviews might reveal recurrent themes. “Long wait”, for example, or “Poor atmosphere”.
Studying trends in post reactions could reveal that your marketing is inspiring the wrong sentiments – anger where you were aiming for laughter, or sadness where you wanted love.
By comparing the tone and reaction your brand is inspiring with particular themes, events, post types, and more, you can develop insights about the feelings you’re evoking – and why. You can then use these insights to:
- Improve service based on what your customers are telling you. Sentiment analysis can reveal specific ways in which you’re going wrong (or right!). Use what your customers are telling you, and give them what they want.
- Target and personalise. You may find that sentiment differs across segments. Use insights drawn from segment-based sentiment analysis to intensively personalise and target your content.
- Track sentiment over time. Your NPS score will fluctuate over time. Tracking sentiment alongside NPS can give you insight into why this is happening (and what you can do to re-enthuse your customers during NPS lulls)
Combining sentiment and data
Modern data science is bringing us closer to our customers than ever before. Through the magic of modern digital analytics, we can turn even ephemeral things like sentiment into data.
And we can use that data to give us insights into what our customers are feeling on a scale that, just a few decades ago, would never have been possible.
Sentiment analysis is here to stay, and it will help you build the kind of loyal, engaged, and customer-led relationships that every brand needs.
To crack open that data gap, we need to start conducting in-depth semantic analysis of our content. Only then can we begin to truly understand why some content resonates and some doesn’t.
The datasine Team
98% of marketers agree that personalisation is of huge benefit when developing customer relationships. But there’s a very fine line between ‘personalised’ and ‘creepy’.