Businesses feel the need to reconnect with their customers through long-lasting and impactful relationships. The good news is that we can do that with sentiment analysis. Sanjay D’Humières, CEO of RTCX, gave us more details about his work.

Sentiment analysis is a process whereby we analyze how people feel about the theme or the brand through social media interactions.

It can also be a piece of content that’s also offline, which is not from the digital sphere. It’s all about trying to understand whether people are happy about something, whether people are unhappy or whether people just don’t have an opinion about an issue.

Therefore, there are three categories of sentiment: positivity, negativity, and neutrality.

How about those who don’t use social media?

How people consume media content in France is different from how they do it in the United Kingdom. To do sentiment analysis you need a community panel. We also have to get the opinion of people who don’t comment, who don’t use social media, but have an opinion. That opinion matters too.

We have to look for a kind of consensus between the online and offline. It’s two different types of profiles, but then we put that data, that feedback together, and then we can feed it to the client. And through this, our solution is also able to detect the future expectations of audiences. We can analyze what they want to see and what they expect from a program.

Which sector can benefit the most?

We developed for the media sector. Then, we quickly realized that other sectors would also benefit from this analysis method. So we’re very happy to say that we’ve extended the art solution to the customer service sector, which is also sentiment-based. We are also working towards offering this solution to the public, and local government authorities in the United Kingdom and France who need to work with the local population.

So our solution is all about listening and also helping our clients to be open-minded with whoever gives them the feedback. So we are data-driven.

Are there any industries that would not be suitable for this kind of analysis?

Not really. Anybody wants to find out what other people say about them and therefore would benefit from this solution. However, our approach is slightly different to our competitors, and that is we do not have one solution at one price that the client can either take or leave where there to solve the client’s problem.

We will go speak to the client and assess what their problems are and offer a tailored solution. And therefore the solution might not be the same from one client to another. And that’s very important to bear in mind.

What is your work based on?

Our solution is part sentiment analysis. Another part is panel, research. So we’re a hybrid between sentiment analysis and market research. And the reason why we’re a hybrid is because we feel that there aren’t any solutions at the moment that do both, that offer both methodologies in one solution.

It’s either one, and therefore companies have to buy two solutions, which tend to be more expensive. The process that we undertake is very simple. We have the analyzer (our tool that captures all social interactions online, whether it be from blogs, forums, and social media), and all those captured interactions are then classed category according to their sentiment.

So our sense score is an algorithm and a sentiment scale that goes from plus 10 to minus 10. And based on what we decide from the message, then the machine will give it a score. For example, between plus four and minus four, we consider the message neutral.

Market research

Now, if the message is not understood by the machine, the machine will send us the interaction and will ask us to grade the interaction. And that’s where at this point we have an artificial intelligence that will understand what the talking points are and will create a questionnaire.

And we have about 50% response rates within the first hour after the questionnaire is sent. So we have very high response rates and that means we can also provide data in pretty much real time. And we can also send about nearly 400 to 500 surveys a day if we wanted to. Automating the market research, the analysis process allows us to go and get data in a faster manner and in a more widespread manner across the panels, but without compromising.

How long does it take for a project in order to be successful?

I would say that it takes about maybe three months for us to set up a project for it to be fully operational. Technically speaking, our solution is designed so it can be quickly and efficiently replicated. However, it does take some time to collect data that can then be compared over a certain amount of time. It depends what data you’re looking for.

It takes a relatively short amount of time, but then if you collect data, you need to know what you’re going to do with it. You need to know how you’re going to compare it to other data sets. And if you want to compare over a longer period of time, then obviously the project will need to be deployed for longer.

The more data you collect, the better the insights are going to be. What I’m saying is that it depends on the needs of the client because data is very much linked to strategy and to business decisions.


We’re working with the national school of statistics. We’re working with BBC research and development on this project, and hopefully it will be launched. But we’re taking our time because developing a truly, truly clear and solid methodology is taking more time than what we initially envisaged.

Therefore, we’re not going to stop. Until we have our methodology clearly under our belt and clearly set up, but that should come before end of 2022.

You can watch the whole discussion below:

Feel free to contact us if you need more details about sentiment analysis, content marketing, branded media and more!

Here are other articles that may interest you about:
Key Performance Indicators to Track in a Content Marketing Strategy;
How to Improve Your Content Marketing Strategy;
Common Mistakes in Content Marketing. 5 Things You Should Avoid