An interview with Patrick Charlton: sentiment analysis is finding its feet

Patrick Charlton is the director and co-founder of Buzz Radar, which is an industry leader in capturing and visualising real-time data. What follows is a sentiment analysis focused interview with Patrick for our book, The Conversation Engine.

Why do you think more businesses are looking to analyse social media using sentiment analysis?

By and large, sentiment analysis a strong key performance indicator (KPI) for businesses to understand what their audience is saying about them and get a wide-ranging picture of how well they communicate with consumers. It is, however, not without its complications. When used badly, sentiment analysis can be highly misleading.

What regulations do you think need to be in place to allow businesses to use sentiment analysis while protecting consumers?

I think technology is already providing all the regulation required — the rules of engagement are fairly straightforward. If it’s publicly posted information such as a tweet, then it’s fine to analyse the post’s sentiment.

If it’s in a walled garden environment, like Facebook, then the only way to analyse the sentiment is via the people who work at Facebook, who anonymise and aggregate the data before producing results.

Most organisations are largely looking for barometers and predictions, so sentiment analysis is used to look at many posts rather than at individual posts — humans are much better at judging smaller data sets.

Who do you think should drive the development of these regulations? Should it be businesses, a trade body or should it be at a government level?

The current data protection laws and technology providers are doing a good job at present. I can’t think of a single way sentiment analysis is being used in a way that could infringe on someone’s privacy that they couldn’t stop easily with a quick tweak of the data privacy settings on their social media profile.

Can you give any examples of sentiment analysis in use, and how this helped a business to be successful?

Individuals inside an organisation’s management are sometimes not aware that there is an issue with negative sentiment around their brand. By visualising data to highlight the levels of negative sentiment and how they tie to certain events, we can help them understand the true impact of key decisions outside of traditional hard KPIs, like sales numbers.

Customer service is also a great example. Using sentiment to quickly sort through large volumes of posts to find the most negative around a subject, and then alerting a support team to address complaints in order of severity, can be uniquely useful in allowing organisations to triage customer issues effectively.

How can sentiment analysis help with brand engagement?

Simply, sentiment analysis shows that you’re listening. It’s well established that identifying negative sentiment issues, addressing underlying reasons and communicating with the sources is a remarkably successful way of taking key detractors and turning them into engaged brand advocates. It also demonstrates to the wider audience that you are present and proactive. This, in turn, can only promote further organic engagement.

On the flip side, understanding what factors are generating positive sentiment around a brand is useful in navigating the topics your audience likes to engage with. Creating content around these themes and highlighting positive posts generated by the audience are proven tactics that motivate brand engagement.

How can sentiment analysis help with competitor analysis?

Sentiment is only really useful when it’s looked at in context rather than as an absolute. Monitoring competitor sentiment helps provide a valuable benchmark for your own sentiment, while also giving you great data for competitor Strengths, Weaknesses Opportunities, and Threat (SWOT) analysis.

What would you say to someone who is apprehensive to use data from sentiment analysis?

We would ask them why they have misgivings and address those directly. Sentiment used correctly can be an extremely effective indicator when used alongside other metrics, and in context. It can be misrepresentative when mishandled.

How can companies that have never considered using sentiment analysis begin to integrate it into their practices?

A simple checklist is a great idea to assess the value of investing time, effort and resource into sentiment analysis. There is a real risk with any KPI of investing energy in an area without a clear understanding of how it will move your organisation forward.

A quick look at the data to assess how effective sentiment can be for your brand is important, as it can depend on the kind of conversations your organisation and industry has with its audience.

For example, conversations in the airline industry show much higher negative sentiment than those in fashion retail. That is because people have a high rate of mentioning airlines when there is a problem with their journey compared to fashion retail, where there is a culture of sharing posts about purchases.

In other instances, there isn’t enough data to analyse or, due to its nature, there isn’t enough emotive conversation to obtain accurate analysis, with most mentions being neutral.

I would really try to understand what you want to use sentiment analysis for, what insights you’re looking for and how are you going to act on them. What levers are they going to drive you to pull inside the business, and what is the expected outcome?

For example, are you going to use sentiment as a core KPI to measure success with your social content strategy or overall brand health, or are you using it to identify specific customer service issues? This will help you choose the right sentiment analysis tools for the job.

When looking for sentiment analysis tools, it’s good to keep in mind the languages you need to analyse and the context those tools provide. That way, you can see beyond a simplistic percentage of positive or negative.

Additionally, you need to consider how you are sharing and consuming sentiment analysis inside the organisation. The biggest – and most common – challenge we see is valuable insights like this being kept in silo inside reports and not acted on.

What do you think the future holds for sentiment analysis on social media?

With the advent of readily available AI, specifically machine learning, we are starting to see a rapid improvement in accuracy. Leading brands are now training sentiment platforms to understand the way their customers speak and to understand the nuances of the language specific to their industry, as well as picking up on tone of voice. This is something we are already working on, alongside building intelligent alerts that notify stakeholders to key events that can then be acted on in the moment.

Finally, we are starting to see brands take a more mature approach to sentiment analysis. They are neither seeing it as a panacea to insights or as an inaccurate, unnecessary expense.

Sentiment analysis is starting to find its place in the marketing mix alongside other KPIs, and it is backed up by context that can be found by looking at posts that drive sentiment, rather than raw numbers alone.

NB: This interview is only an extract from The Conversation Engine. To read the full book, download a free copy here: