“The key to social media analysis is being able to rank information in the most relevant manner”

Alexander Polonsky

Interview with Alexander Polonsky, Co-founder and VP R&D of Bloom Social Analytics, a social media analytics company in France.

Hi Alexander, what is your background and what is included in your current role at Bloom?

My background is in Neuroscience. I have a Phd in Neuroscience and a Master’s in Applied Mathematics, and I have been working in information and analysis management for the past 20 years.

At Bloom, I research. More specifically, I research algorithm design, in order to design algorithms to analyse the social media data that we collect. The process of design is a scientific process in which we define hypotheses about how we would go about solving business problems. We then create those models and options in order to test them, before refining and iterating based on the test results we receive from testing. We do this until we reach a satisfactory performance.

What differentiates Bloom from other social media analytics companies?

Our positioning is very different and we’re unlike other social media analysis companies in the sense that we focus on the quality and relevance of analysis over the speed of analysis and the spread of source coverage. We cover fewer sources but do so in far greater depth. We cover only Twitter, Facebook, Instagram, Youtube, Weibo, WeChat and TikTok.

Gaining access to data is challenging, but in addition there are challenges to analysis. When trying to access data via search queries there can be all sorts of problems, particularly if the intention is to statistically analyse the results of the search. Algorithmic issues must be solved, such as detecting the noise, completing the query, and detecting content that is relevant but not easily tracked with keywords. Bloom is really the only company to address these issues adequately, using powerful automated query modelling assistance. Our classification approach to data allows us to go beyond keywords.

There are two key points we stress in terms of analysis of data. One is the quality of our mathematical analysis. There are many companies that analyse social media data very superficially, and the majority of companies commit grave mathematical errors in their analysis by treating different sources as though they were the same, or using a trend calculation that is not really a trend, or performing sentiment analysis very poorly.

The other is that we analyse data in a way that goes beyond our competitors’ abilities. We analyse positions, not just sentiment. We also analyse emotions, and we have developed unique models of information importance. The key to social media analysis is being able to rank information in the most relevant manner, and counting different types of engagements in the same manner is a huge oversimplification.

What are your greatest challenges ahead at Bloom, when it comes to serving your customers and developing your offer?

Data access remains a major challenge for all players in the space because platforms are not predisposed to sharing. On one hand, this is understandable. On the other hand, it’s short-sighted. Twitter has shown it can be a major part of a business model, and importantly, if companies feel that they cannot see what is really happening on a social media platform from the outside and verify information for themselves, it becomes harder and harder to trust platforms that are charging for advertisements.

So far Bloom has not focused on real-time analysis. We prefer to take our time and analyse thoroughly to provide deep insights and strategic recommendations. But more and more our clients are asking us for real-time updates. We will now focus on developing this side of our offering.

We have invested heavily into our investigation capacities but we also want to develop our alerting capacities.

When it comes to the actual data behind the analytics you do, what kind of data or media not currently used can be interesting in the future?

Access to some of the platforms is getting harder and so the best practice is to diversify and increase source coverage. We are continuously integrating new sources, but when integrating we do so in a way that goes far further than most competitors. Being able to access data on a platform can have many different levels – what exactly a company is able to access is key to its abilities.

You have recently launched a partnership with Dassault Systèmes. How will that partnership affect Bloom moving forward?

The partnership opens up a whole new sector for us: product design. We’ve already worked a little in this area for previous clients but the Dassault Systèmes partnership is a strategic alliance to launch a new offer because the main clients of Dassault Systèmes are those designing and managing different products, for example, in the transportation industry. We will be feeding these clients with strategic insights about market trends, consumer behaviour and similar information.

How do you think the social media intelligence industry will change in the next 5-10 years, and what are the greatest challenges ahead?

Challenges to data access will certainly remain but we’re seeing new social media platforms spring up all the time, and not all of them are bought by the majors right away. This means there is a high diversity of players. Keeping up with them all will not be easy but may be easier than gaining information from some of the traditional major platforms.

Another challenge is proving the new use cases that companies are presenting. Social media analysis has addressed only part of the large number use cases it could work for. While new use cases are still to be proven, this trend of moving the industry beyond marketing and communications will continue. Our partnership with Dassault Systèmes is an example of that.

I hope that the industry will mature. At the moment it is somewhat of a Wild West and grand claims are made that are not then backed up. Clients do not generally have sufficient expertise to be able to judge the relevance of one solution over another, so hopefully we will see some standards emerge. We’re still far from that at the moment.

Can you please give an example of some surprising findings from your analytics that made a significant impact for a customer?

We worked for one company in the bottled water market. They had been putting a lot of effort into the recycling of their bottles, but we helped them realise that the consumers did not trust recycling. Indeed, consumers really wanted the re-use of plastics and the abolishing of single-use plastic. This difference struck at the core of the strategy for this large company.

We worked for another client in the prepared food industry where many companies are looking to migrate from plastic to aluminium packaging. We discovered that there is a growing weak signal of alarm about the health risks of aluminium packaging. In that sense, they may have been moving from the rock to the hard place.

By Peter Appleby