” Free the culture! Starting from these conflicting sanctions, Fnac runs a massive communication campaign that is biased (and therefore more effective) at various levels. First, the contemporary inequality between instruments and humans – and especially booksellers vs. algorithms. The first fly in the ear: The seller Fnac is not a bookseller, but a consultant. It says a lot, but in music, you accept everything.
This does not mean that he or she is incompetent. And so the instrumental creatures who are being told to practice artificial intelligence, to attract sympathy, do not lack courage – but we will return to it. Improve employee image, why not: There is no doubt that CGT Fnac will appreciate the pay rise. That’s it.
Since then, Magic Messages has put a decent cover on Fnac Darty’s financial reality: 26% of overall turnover was generated online in 2021, with 30% probability. Advertising and public denial of the importance of algorithms in the economic model is therefore ridiculous.
Because the real issue may be less to promote Fnac sales consultants, as far as cultural offerings are concerned, than to create underground powerful digital tools. Pierre Framoux, co-founder of the Babelio site, came back with us on this com operation, or how the bladder supply lantern was suddenly replaced.
ActuaLitté: You run a social network that offers readers book reviews, literary quizzes and more. How do you worry about algorithm questions?
Pierre Framex: Babelio was founded 15 years ago, offering an online reading management tool, with the idea that by providing readers with detailed information about their interests, we can offer specific and relevant automated referral services. This was a time when Netflix was more enlightened by the quality of its advice than its original production, and we were on our side to keep reading data for the development of such so-called collaborative filtering tools. If you like this book, other similar readers recommend … ”
Without questioning this initial principle, we have expanded our perspective on the social and human dimension of the network of readers; The value of the service lies not only in quantitative data, but also in the contributions of Internet users: more than 30,000 reviews from enthusiastic readers each month enrich the site and make it possible to make many humane recommendations.
Read: What human rights fears from algorithms
Today we offer booksellers and libraries the opportunity to enrich their sites with this dual knowledge: an algorithm that promotes content discovery and human recommendation. We believe that these two dimensions are complementary and combine more frequently than it seems.
Fnac claims that algorithms trap us in a bubble – adopts the idea Filter bubbles : As a result of the information collected, the Internet users find themselves trapped in a vicious cycle. What do you think?
Pierre Framex: The campaign runs efficiently with the exception of online platforms like Amazon and it is a fair game. Outside of advertising discourse, however, it seems important to better understand the challenges of these automated tools. An algorithm is a series of methods of solving a problem, but once we say it, we do not qualify at all for the nature of the purpose! The goal of Youtube’s algorithm is to keep you on the platform for the longest time, Spotify’s goal is to discover enough music to keep you subscribed, what will happen to the book?
Our advice engine must be trade-off: personalize recommendations according to readers, on the one hand, confirm diversity, and surprise, on the other. For a bookstore or a library that uses our services, the challenge must be to borrow or promote sales, but if we reduce the response to a single genre of highly anticipated books, we will have no added value.
With more than a million titles of data from different genres, one can easily reproduce the brutality of unexpected discoveries and explode filter bubbles that limit recommendations based on past tastes. Today Babelio’s tool recommends me Mimesis Investigating Erich Auerbach, a mountaineering manga and shale gas comics, I don’t feel too limited.
The antagonism between the algorithm and the bookseller, now the proverb, is the amount of disconnecting the machine from the people. How is this possible?
Pierre Framex: The historical starting point of these recommendation tools is to offer books that have been appreciated by readers who have shared your taste. The tool must inject data, which is only the sum of the reader’s assessments. Without humans, so without information about the taste of reading, the machine has nothing to eat. So these must be DatesToo many people: Who read the book, on what date, were you satisfied? What else is the reader reading, how has it qualified quantitatively and qualitatively, and so on.
Although data encoding is done with computer bits, we are dealing with a social, cultural and psychological truth.
What is the specialty of the algorithm in the context of the book?
Pierre Framex: On a platform like Youtube, Tiktok or Spotify, Internet users consume hundreds of content and feedback each month (by which the system knows instantly whether the recommendation is relevant). Internet users consult or zap and communicate directly about their preferences.
Read: When robots steal advice and book summaries
In the case of books, we live a long time. Although 60,000 books are published each year, a “great reader” will read 20 per month. It is clear that in this context, he needs reassurance. Recommended tools are part of the puzzle that will not be taken at face value. It is unlikely that a reader came runningFinding lost timeUsing only one algorithm.
The reader will actually consult a series of signals to validate these suggestions: metadata (book cover, summary), social appreciation (reader’s note, reader’s review) or even reading the first pages. And if he is tempted by the advice, he will have to wait a few more days to find out if he likes the book.
In the world of books, algorithms are much more needed than anywhere else to appeal to people to become truly useful.(Comments collected on June 20, 2022)
Amazon, Fnac and Books: Episode 2
That being said, communication shouldn’t be overemphasized: it often has a small memory – at least it doesn’t encourage consumers to keep their minds sharp. Thus, the shift to stores and advisers tends to restore Fnac’s image, as opposed to its first competitor, the infamous member of GAFAM, Amazon.
A recent Senate report noted that the two companies are waging a fierce battle over book sales. Fnac will sell 42 million books in 2020, of which 36% are via the Internet (delivery, 27%, clicks and collections, 9%) – or 15 million copies. In contrast, we lend 10% market share or 400 400 million of publications to Amazon. In short “Amazon will sell about 40 million books“, The Senate concludes. And the two operators will then weigh 80% of online sales, selling 55 million headlines.
In 2019, Obsoco produced for the French bookstore syndicate – not really a fan of Fnac or an ardent defender of Amazon – a study on customers. 91% of online shoppers frequently visit Amazon and 64% choose Fnac.com. The well-established and proven DuPont also ate up market share: in 2013, 48% of respondents who went to independent bookstores bought a book on the Internet, up from 55% in 2019. Data increased 51% in 2013 and 64% in 2019%, to non-customers of the aforementioned bookstore.
Reprinting Le Tendre’s map from the side of independent bookstores and nice people, is tantamount to opposing Americans without too much subtlety. And let people forget that since 2011 and the arrival of Alexander Bompard, Fnac has made a point of developing a so-called “omnichannel” strategy. In other words, the combination of store and internet – a message that was hit repeatedly and well at home after the exit of the former CEO.
Algorithms and people
The current CEO, Enrique Martinez, reiterated this in 2018: “The goal is to bring together the latest independent vendors and expand the regional footprint to be ubiquitous. Moreover, when you open a Fnac or a Darty there, online sales in an area double the average.. The internet is therefore integral to the store. And who says the internet is called algorithm.
And to do that, there is no shortage of investment: in 2015, not to go too far back, Camille Burland, head of the brand’s CRM division, indicated: “At Fnac, we started collecting data very quickly through our membership program and our multi-channel store. And a few years ago, we merged our different databases.A
Read: “The Amazon world is not where I want my children to grow up“
All this with lots of targeting, customer base and data analysis. An intervention that took place in one of the most obvious settings: Big Data Paris 2015 Symposium. And to strengthen its web skills, Fnac has introduced its partner, TinyClues, a machine learning expert. We smile now to read:Speaking of culture, it’s better to have a heart than a processor.So, what about a multi-core processor?
Most recently, in 2019, the brand’s web and mobile director Francois Ozan returned to this development. “For many years now, we have been practicing machine learning and betting on personalization. In 2018, we decided to internalize our algorithm. We have created our own platform hosted in Google Cloud. But isn’t data collection at the heart of GAFAM’s strategy? Fnac Darty can’t ignore it.
My AI is not fresh?
And in detail:These are the artificial intelligence algorithms of self-learning. We don’t define rules like “X happens, you do Y”, but we do care to help the machine learn. To give an example: our AI must understand that with the purchase of a specific typology, the customer can buy some related products within a few weeks. A customer who is buying volume 1 of a book is likely to buy volume 2 in a few weeks, etc.. A
Here again, large gaps become difficult to maintain. Especially when the CEO, in early 2022, gave it a famous lead by highlighting its agreements and developments with Google. The perspective was twofold: on the one hand, promoting the use of artificial intelligence between parties – a shame? On the other hand, for the better, repeatedly, the online shopping journey. “The performance of our site has been enhanced with a more immersive, efficient web experience powered by artificial intelligence.“, He concluded.
And we can only assume that in the current web competition, if Fnac does not strengthen its tools, customers will go to Amazon to see what color the grass is. From there, pretending to be a small book seller, the recommending heart is at hand …
Moreover, the algorithm or advisor, Fnac, had a very bad press when, in September 2021, photos of the store showed POS with a largely conspiratorial work … as well as its sales site requested similar recommendations. Reviews that Amazon has suffered, on a larger scale. Gluops?
Photo Credit: ActuaLitté, CC BY SA 2.0; Fnac