The French of connected TV: TvTweet


After JoshFire JoshFire In this detailed panorama, which will last several months on the French ecosystem of connected TV, let's move on to a small startup that is just starting to emerge and that is working on a current topic in "Social TV": Twitter. In this category, we have applications that use Twitter conversation feeds to improve the user experience, which we'll see later, and others like TvTweet that allow us to analyze them.

Twitter and TV

The use of Twitter has seen rapid and recent development in France. There are an estimated 5.5 million active Twitter accounts in France, with 2 million accounts having at least one Tweet on a TV programme. In the UK, 40% of Tweeter conversations are TV-related. In 2011, a record was broken in France with the NRJ Music Awards in January 2011, which generated 70K tweets. The 2012 edition progressed moderately to 90K tweets. But the final presidential debate generated 360K tweets. And the final of "The Voice" on TF1 generated 204K.

With an average of 25 million viewers per day, this gives a penetration rate of about 8%, which is quite respectable. Knowing that a good proportion of Tweeters users are "listeners" who do not broadcast Tweets. There are few or no data sources but we can at least estimate that for a "transmitter" there is between 1/2 and one "listener", which would bring the audience share of Twitter to more than 10% of viewers, which is far from negligible.

Twitter and to some extent Facebook have become rather good news for TV channels. Why is that? Because they enhance the attractiveness of their live programmes, especially high-audience reality TV shows, political debates and sports. On the other hand, this does not seem to have too much impact on the consumption of TV series. Twitter is a generic platform that could, however, accentuate the horizontalization of uses, for example by reinforcing the role of multi-channel social programme guides to the detriment of channels' attempts to integrate "proprietary" social feeds into their own content and applications.

Why not enjoy unlimited reading of UP'? Subscribe from €1.90 per week.

What triggers viewer engagement on Twitter? First of all, there is a strong appetite for the young and hip audience segments, the overall audience level of the show, the clever themes and, most importantly, the role of the host in promoting new uses. In terms of audience, a reality TV show like "La belle et ses princes charmants" that airs on M6 and W9 generates more tweets on W9 because the audience is younger and more engaged. Same thing with NRJ12's "Les anges de la télé-réalité", a show that puts us at the bottom of the intellectual league but which beats records for audience ratings and interactions on Twitter. In terms of cleavages, Laurent Ruquier scored very well with heckling debates, as with Marine Le Pen or Audrey Pulvar. As for the role of the host, the case of Nikos Aliagas is interesting since it brings together more than half of the most tweeted programmes in recent months, including "The Voice".

You can learn more about what the channels are doing with Twitter in France and in the USA in the excellent article "The Twitter Channel". The TV stations are chasing twittoes... " by Pascal Lechevallier on ZDNet. This one is an in-depth zoom on one of the solutions he evokes.

By the way, let's talk about Facebook, which does not have the same presence in "social TV". The reason is simple: the information flow in the leader of social networks is not readable by everyone, but only by your friends. Only the "pages" most often associated with stars and brands are publicly accessible. As a result, even though there may be conversations about live TV programs, they cannot be aggregated by third parties or made public. With Facebook, the "broadcast" of messages is limited to friends, while with Twitter, anyone can follow anyone, unless they are explicitly blocked.Facebook can, however, play a role in recommending content, but this is a different story and involves a different engagement of users. Conversely, conversations on Twitter are by default public. Only DM (Direct Messages) are not.

equipetvtweetThe TvTweet team

TVTweet was created by Thomas Landspurg (opposite), who lives in Bordeaux and spends half of his time in Paris as it should be because that is where his clients in the French market are located. He has an entrepreneurial background with a CTO role behind him. In particular, he played this role in the Webwag startup that publishes mobile solutions for telecom and media operators. Prior to that, he had developed Java-based downloadable game technologies with a business model of catalog sales and billing via telecom operators. Good times for the latter! The solution had been sold to Orange, Vodaphone and SFR. Thomas is accompanied by a second person for business development. But it remains a very small team.

The products

What's the offer from TvTweet? It is available in three forms:

A website providing statistical data rather intended for media professionals, allowing to analyze the flow of Tweets by country, by channel and by program and with a very interesting comparative logic.

APIs allowing access to data visible in the previous application. They are provided to developers of third-party applications, such as online Program Guides. This is a form of open data in social TV.

An iPhone/iPad mobile application broadcast under the Digiwiz brand, aimed at the general public and presenting the most Tweets-generating programs. It is mainly a demonstrator of TvTweet's API platform.

To fight against disinformation and to favour analyses that decipher the news, join the circle of UP' subscribers.

TvTweet APIs

These APIs thus constitute a kind of "open data" offer of Twitter conversations around television programs. They provide access to both a stock of data and real-time statistics of Tweets covering currently 43 channels, mostly free and broadcast on DTT, of TV in the UK, France and Belgium. The service is based on hashtags of the programmes of these channels and on the content of the country's Electronic Program Guide (EPG).

A popular show can generate up to 3000 Tweets per minute. The service integrates a filtering function of this stream to limit the number of Tweets broadcast by the API operators. It is a configurable filter for application developers at the level of the filtering rate of the original flow of tweets. You can request to retrieve 2%, 5%, 10% or more tweets. The filtering is based on semantic techniques. It relies on hashtags and combinations of associated terms.

These statistics allow the creation of second screen applications such as "social" TV program guides as well as the integration of Twitter data into more "btob" tools for professionals.

Under the hood, TvTweet accesses data by directly accessing the Twitter service and without going through the data "wholesalers" that distribute what Twitter calls "firehose".

The interface to the TvTweet service is provided in the form of JSON APIs. It is a way for a web client application written in JavaScript to query the TvTweet server and easily retrieve the answer as locally formattable data to, for example, create graphs. The system has been tested with a load of up to 100 tweets/second.

Data is stored and managed by an open source database. MongoDB MongoDBThis is a so-called "non- SQL" (non-relational) format that directly stores data in the structured JSON format. Heavy processing batches are queued using the message manager. RabbitMQ RabbitMQ. All of this is hosted on several dedicated servers at

You can react to this kind of offer by saying that it's a piece of software that's easy to reproduce. But developing and assembling all these bricks requires a certain amount of know-how. There are many integrated bricks. And the articulation of the offer is the result of knowledge of customer needs acquired in the field and through practical experimentation.

TvTweet's analytics site

The destination site TvTweet TvTweet which is for the moment freely available, allows easy access to a whole bunch of statistics:

- Number of users tweeting per show and per channel.

- Commitment level per issue that measures the number of Tweets per person.

- Comparative performances of several episodes of the same series of programs.

- Identification of influencers: those Twitter users who are the most retweeted and/or have the largest number of followers.

- The flow of tweets per show about a person. Anyone or influencer.

- Trending topics are the hottest topics on Twitter.

- Audience share on all the tweets on TV. It's a kind of "social ratings".

- A tabbed time view that allows to detect tweets flows by time slots.

- A stopwatch in the dashboard that gives the temperature of the Twitter feeds in real time.

Let's go for a little tour of the interface:

analyticstvtweetLet's start with one show in particular, here on TF1. We see the stream of tweets spread out over time with classic sequencing in a TV show and a double speed bump area of interest. The "climax" of a show is positioned towards the end to preserve the audience right to the end, especially through the advertising tunnels. The show was quite engaging with 3.24 tweets per twitter. We still don't know how to measure the number of passive followers of a hashtag!

Still on a show, you can retrieve all the hashtags used and the main contributors in number of tweets :




crosswordsinfluencerIt is completed by a list of influencers and the most retweeted tweets:

Another example is the time flow of emissions for one day on 21 May 2012. There are two traffic peaks:

At the end of the evening with Yves Calvi's political show "Mots Croisés". It happens to publish the best Tweets on the screen since this last edition, going beyond " C'est dans l'air " that we have already examined recently which aggregates SMS, web messages, tweets and HbbTV messages without specifying their origin. By "editorializing" the tweets in the show, the channel makes the explicit choice to encourage viewers to express themselves through this channel in order to give a participatory form to the show.

At the end of the afternoon, there is a peak in traffic with NRJ12's "Les Anges de la TV réalité". As its name suggests, this is a reality TV show that alternates on-set discussion (with Loana) and scenes shot on location (in Los Angeles and Hawaii). All with former participants in reality TV games. A kind of meta-reality TV. Intellectually, you can put your brain in the closet and even in the locked basement! This traffic spike illustrates the "format war" between TV channels. This one was created by "The Big Team"the company of animators Jérémy Michalak - who worked with Laurent Ruquier - and Zuméo (Fun TV, M6). The show has been sold to 8 countries. This format is much more targeted than "The Voice", which comes from the English language. Shine Productions and also broke records for viewer engagement on Twitter. Two things are new with this program aired on NRJ12: the program's time slot and very explicit youth targeting. Everything to generate traffic on Twitter!


The next screen shows the statistics summarized by program with the programs that generated the most tweets in the last seven days. Not surprisingly, it shows that reality and talk shows generate the most tweets. On the other hand, TF1's seem a little less engaging in terms of the number of tweets per station. Perhaps due to a higher average age of viewers than channels such as W9, M6 and NRJ.


Finally this last screen which illustrates the ability to follow all this in near real time, with in passing a small speedometer that indicates the twitter pressure.


There are probably a lot of dashboards that we would like to have, particularly to compare the level of commitment by channel, all programs combined, and spread out over time. These are features that will appear either in this application or in those that will use the TvTweet API.

Finally, the tool will be customizable by customers to add hashtags to track and to generate PDF reports that are easy to distribute.

In the case of both application and APIs, TvTweet will be marketed as an online service with a monthly fee. It will be of interest to TV channels, program producers, advertisers and media planners as well as the press that follows television news.

Competition from TvTweet

The market for social network monitoring tools has become very dense in the last two years, including measurement tools related to Twitter. On the other hand, fewer people link Twitter feeds to TV programming because they are more of a niche market.

Amongst others, we can quote:

- Mesagraph which proposes APIs to access activity data in social networks and especially in Twitter. The company also offers its "front-end" application to exploit them. It is now positioning itself in the TV market because it is the most promising one to exploit these data. We will talk about it in this series of articles.

- Devantlatélé, which is the opposite of the previous one, offers a web application without APIs, oriented towards the general public, allowing to consult the traffic on Twitter on all the programs and to identify in real time the most popular programs. I had discovered this company at the web2day in Nantes in 2010.

- BlueFinLabs, a startup based in Cambridge, near Boston, which is based on work from MIT in semantic text processing and offers an analytics solution similar to that of TvTweet, but which covers "social networks", i.e. Twitter and Facebook. It covers the American market and both TV content and TV advertising. Its CEO, Deb Roy, was also involved in the project during LeWeb 2011. Just to disgust us, the startup created in 2008 raised $20.4m. But hey, is it well used? We have to wait a few years to find out.

- Trenderrr is a second American startup, based in New York, which offers as TvTweet both an application and APIs to use its data in its own application. It also covers Facebook and Twitter as well as the social TV applications Miso and GetGlue. But they offer analytics with geographical segmentation (how retrieved?) as well as the comparative performance of different characters / actors / speakers of TV programs.

- SocialGuide is another American company that offers both a web/mobile program guide with bits of social in it and reminds, in a nicer way, of Devantlatele, and an analytics application for professionals that produces dashboards similar to those of TvTweet. The company came to Spain through the synergy with the offer of the North American Hispanic channels.

- SecondSync, a UK startup focused on this country and on measuring Twitter feeds. The application looks rudimentary.

As is often the case, the solutions of American competitors have not necessarily worked better, but on the other hand, their interface is cleaner. The graphic design often comes later with the first funding... with the first financing.

TvTweet customers

TvTweet is still very young but the founder already has a few clients, including a large French company that uses the TvTweet API for its SocialTV applications. NRJ12 also exploits TvCheck reports, as well as consulting companies in the field of social TV. It has in mind all the publishers of content or programme guide software that need to integrate a dose of "social", especially to promote catch-up TV.

In short, we have here a small agile structure that has created a very generic software solution and a real services platform via its APIs. It's a mix of cloud and big data in the air and the reports provided meet a real need of TV professionals. The challenges? In B2B, the market is quite small and such a structure will only be able to reach a critical size by developing internationally. On the other hand, the btoc market could eventually develop locally, but it is expensive to approach it, with uncertain results because it is another business, more in the register of a front office.

It is also ironic to remember that this startup, like thousands of others, has based its strategy on an American company whose sustainability and business model has yet to be validated. Innovation that depends on innovation that depends on an uncertain market. This is the common lot of digital television!

And two...

The next one will ba-aa... I don't know yet!

{Jacuzzi on}

Inline Feedbacks
View all comments
Previous article

The French of connected TV

Next article

Cyber Scams alert!

Latest articles from Media Archive



Already registered? I'm connecting

Inscrivez-vous et lisez three articles for free. Recevez aussi notre newsletter pour être informé des dernières infos publiées.

→ Register for free to continue reading.



You have received 3 free articles to discover UP'.

Enjoy unlimited access to our content!

From $1.99 per week only.