Article

06.03.2018

The banking profession through the prism of big data

In an article praising the smart use of data to redesign the banking model,Edouard d'Archimbaud of BNP Paribas says data is worth its weight in gold.

Fifteen years from now, will robots be able to exactly replicate a human brain? One thing seems certain: in 10 years, we will no longer be able to distinguish a biological voice from a robot's voice on the telephone, says Edouard d'Archimbaud, head of the artificial intelligence (AI) laboratory of the BNP Paribas Corporate and Institutional Banking Division. 

In years to come, the phenomenon is expected to have a major impact on the financial industry and create millions of opportunities, making it the greatest ever technological revolution. "Imagine, for instance, that we can extract a mountain of data from conversations between traders: despite the complexity, the machine would be able to pick up certain details of the job. Responses to requests for prices could be automated. And people and robots could collaborate to offer new services", says Edouard d'Archimbaud, one of the top 20 data scientists in global finance.

An algorithm can describe and comment on an image

Every day, 1.5 billion images are posted on Facebook. Today, almost all information is in digital format. This vast amount of interpretable data (pictures, descriptions, comments, etc.) is beginning to be readable by more and more sophisticated artificial intelligence systems. This means that the photos and captions we post are turning into a large educational tool for Facebook's robots.

Greater access everywhere 

Open source is working in favour of artificial intelligence, and the power of computers is skyrocketing. Data remains the key asset, however, dictating that some companies can be involved in AI, whereas others cannot.

Just as a child learns by observation, artificial intelligence learns by observing data. The more it receives, the more its performance is enhanced. "The bank, for example, mixes external data (such as transcripts of European Parliament speeches) with its own rich, free sources of internal data in order to create volumes of "trade" conversations that are not replicated anywhere else", says Edouard d'Archimbaud. This is how robots manage to master every nuance of our financial jargon."

How would banks be particularly affected?

Firstly because of their data "culture", in the sense that banking is an information processing industry. The statisticians of the past and today's data geeks have comparable profiles", d'Archimbaud continues.

Secondly, banks are especially affected because they invest an enormous amount in improving their operational efficiency. "We want to protect our customers' interests even better still, which is why we recently developed a tool for reviewing documents. One of our robots can cover 150 pages in 15 seconds. Another example is the tool that provides our internal and external translations: it enables us to manage information effectively in the many countries where the BNP Paribas Group operates. Robots have grasped some of the specifics of what we do, and the results are really exciting", he adds.

While robotisation enables greater quantities, AI, which is systematic and rigorous, enables improvements in quality. It excels at tasks which are difficult, repetitive and costly: it refines the accuracy of the analyses, to which human intelligence adds the finishing touch. AI is therefore transforming bank staff into "enhanced employees". It should be understood that AI adds to their capabilities. Without the assistance of robots, certain tasks simply could not be carried out by the required deadlines. The bank is also building new-generation "user interfaces" based on a range of components: webchat assistants, chatbots, search engines, and the ability to convert voice to text."

Altering the banker's value chain

The bank possesses a prized asset in its data – even if its use is obviously restricted by the bank's code of ethics. "In the same way as a diamond, the richness of which is derived from its multiple facets and complex structure, data releases its value when it is shaped and contextualised. Once combined, merged and cross-referenced, data can broaden the perspectives from which we "grasp" the customer's activity. New dimensions appear that once again create a "centricity" revolving around the customer. Once separate entities, data media such as e-mail and the internet now speak to each other!

"We no longer simply manage transactions; mostly, we draw out meaning from information that we blend together. It's a bit like the automotive industry: the core of the profession used to be manufacturing engines, but today, it has become a business that automates processes and information."

How can we accommodate this new phenomenon?

Do not be afraid that a machine will steal your job. But if your manager knows nothing about technology, it might be prudent to find employment elsewhere.

When was the last time you brought out a new product? Facebook changes its products all the time. And when did you last sign up a high-tech entity, or invest in a technology start-up?

If you have fears, confront them. Fall back on the fact that people and machines operate in tandem. Machines will learn effectively from us providing we always correct mistakes. And conversely, machines are able to pick up human imperfections and help us advance.

Robots work at high speed in a systematic manner free from constraints including tiredness, boredom, and limited time. They help us reassert the value of our roles, release our creativity and give more of ourselves to human relationships.

Article

22.10.2016

Digital transformation: your action plan

33% of IT decision-makers don't have a clearly defined strategy for digital transformation, a survey reveals. Why is this and how can this situation be rectified effectively?

In the fast-moving consumer goods market, technology is now radically reshaping competitive dynamics in the marketplace, for both consumers and distributors. This has irrevocably changed how people buy things. In the results of its "Fast-Moving Consumer Goods" survey, Progress reveals the current paradox: digital transformation is crucial, but implementation is slower than desired.

"60% [of IT decision-makers] admit that their organization is still largely in denial about the need to transform digitally."

Where exactly does the problem lie?

More than half of IT decision-makers see this process as something daunting that will take a long time. 66% feel that their marketing and IT teams are not in alignment to deliver on the project. 64% find it difficult to keep up with the ever-changing digital landscape. When it comes down to it, two departments clash on the distribution of tasks and budgets: IT versus Marketing. The good news? 96% of companies have plans to act within the year.

The key elements of your action plan

1. Begin with an inventory

Achieving digital transformation involves customer satisfaction, the ability to alter the focus of the business and rollout of a flexible platform, but you shouldn't throw out the baby with the bathwater. The transformation should be initiated after taking stock of the company's assets, in the form of an inventory.

Then you should visualize possibilities for change to consider the future of your business, no longer as a traditional firm, but as an ambitious digital hub.

2. Inspiration to think big

The founders of Google say that they have always sought to reject a traditional management approach by adopting two basic principles, which apply perfectly to digital transformation: focus on satisfying users and hire "smart creatives". This can be achieved through the LEAN method.

3. The will and the time

The company has to recognize that digital change is permanent in order to be relevant. It requires the creation of new experiences and the definition of new paradigms that can overthrow traditional models but are ultimately beneficial: 41% of CIOs notice an increase in market share post-transformation, with 37% of employees becoming more motivated as a corollary.

4. The vision and the tools

In order to improve and optimize the customer experience, you have to work on the speed, responsiveness, security and standardization of distribution channels. Mobile devices are widely favoured (62% according to the survey) for analytics, data connectivity, e-commerce, content and the Cloud.

A full copy of the report can be downloaded for free from the publisher's website.

Source: Progress

Article

18.11.2016

Big data: six questions to ask yourself before getting started

Big data is a new class of assets that companies must embrace, develop, protect and make work for them during their transformation into a digital enterprise. We have put together some points to help guide your strategy.

Is there a course in big data?

Most universities around the world have come to understand the importance of big data. More and more, they are using analysis, both in research and to improve the lives of students on campus and help guide them; however, there is little in terms of training on this topic. Nevertheless, some establishments have recently started to offer their own diplomas and programmes to train the next generation of data scientists.

Do I need to provide training for my staff?

Yes. However, it's difficult to send your IT teams back to the school room in order to train or bring them up to speed. Nevertheless, various training courses have been organised around the country by specialist service providers. A two-day training course already teaches its students about the specific issues surrounding big data and the potential technical solutions.

Do I need to hire a data scientist?

Not necessarily. Some figures: last year, there were 4.4 million jobs in this sector, of which only 40% were filled. Not everyone has the budget for a data scientist. You can instead call on an independent consultant to pave the way and get your company up and running with big data.

What main techniques are required?

Techniques such as machine learning and data mining are essential for those working with big data. They help you tackle tasks that are difficult or even impossible to complete using more classic algorithms. The art of Data Visualisation enables you to communicate discoveries from data analysis.

What keyword should I take away?

Hadoop! In the same way as Microsoft Office is known for productivity and Apache is synonymous with the internet, apps are the key in the world of Big Data. Hadoop should be the cornerstone of your strategy. Without such expertise, it is impossible to master big data. This open-source software framework is designed for distributed data storage. It is highly scalable and resistant to failures. Its role is to process and analyse new and old data silos to extract significant knowledge from them that can be used in a company's strategy. Your experts will have to become familiar with its components: ‘Spark’, ‘Hive’, ‘Pig’, ‘MapReduce’ and ‘HBase’.

Is big data relevant for SMEs?

Certainly, in particular for marketing: big data enables companies to sort data in order to gain a clear profile of its customers. Segmentation can be used to optimise campaigns. Analysis also allows you to  really observe how customers behave. SMEs don't have the same budget as a large group, and so they must primarily focus on data which is both crucial and can be exploited to reap the greatest reward: creating a stronger link with their customers.

Article

14.12.2016

Tant qu’il y aura des data…

Le graal des big data? Créer une expérience client sans précédent. Mais pourquoi une start-up sans passé réussit à être affective là où ses ainées croulant sous les data rêvent de proximité ? Quelle est l’alchimie gagnante ?

Stocker et traiter ses données numériques, ce n’est pas nouveau. Le datamining non plus. Mais avec les objets connectés et les usages mobiles, les données déferlent littéralement. SMS, chats, photos, vidéos, requêtes à un moteur de recherche, clics sur le net, demandes d’itinéraires sur google maps ou autres, paiements en ligne, contacts client par chatbots ou messagerie, renouvellement automatique des commandes à partir d’un frigo intelligent… des données, nous en produisons sans cesse sans même nous en rendre compte ! Même lorsque nous acceptons la géolocalisation ou que nous nous connectons à une borne wifi...

En 2020, le volume des données devrait être multiplié par 50. Une voiture connectée, par exemple fournit, en une heure de temps, des millions de données utiles à l’automobile, mais aussi aux assureurs aussi ou à l’e-commerce. Et les enjeux ne sont pas moins prometteurs qu’ajuster sa stratégie, personnaliser un service, prendre de meilleures décisions, détecter des tendances, établir des prédictions… 

Il y a toujours eu des statisticiens pour interpréter les chiffres du passé afin d’améliorer le futur, mais aujourd’hui les ‘data scientists’ sont des geeks. Des cursus universitaires voient le jour et l’explosion des données adopte un rythme quasi insoutenable pour que nos connaissances puissent suivre. Seules des machines sont encore à même de gérer de tels flux de données. Les techniques d’apprentissages automatiques (‘machine learning’) permettent de faire mieux et plus rapidement. Un standard pour une utilisation correcte de l’intelligence artificielle serait en cours à l’initiative de noms comme Google, Facebook, Amazon, IBM et Microsoft. Pour Nicolas Méric, fondateur et PDG de la start-up DreamQuark, acteur de deep learning appliqué à la santé et l'assurance, de telles technologies dopent les capacités humaines mais elles ne sont pas vouées à pouvoir s’en passer.

Qui est concerné ?

Aucun secteur n’échappe vraiment au besoin de récolter ses données afin de les faire fructifier en transformant son environnement. Mais disons que certains se montrent plus pressés – ou opportunistes - que d’autres. Les télécoms, le transport, les fournisseurs de gaz, eau, électricité, émergent : la SNCF mais aussi le fabricant de produits de beauté Nuxe épient tous les canaux en ligne en quête de verbatim client pour mieux le connaître. L’ascensoriste ThyssenKrupp, qui veut chouchouter ses cabines et surtout leurs utilisateurs, récolte moult paramètres sur celles-ci afin de parfaire la maintenance et d’anticiper les pannes désagréables.

Les responsables des Big Data en entreprise sont face à trois défis principaux, rassemblés sous la règle dite des ‘3V’: pouvoir gérer de gros Volumes, tenir compte de l’infinie Variété des informations, et parvenir à gérer la Vitesse à laquelle elles sont générées. Les banques n‘y échappent pas. Ces entreprises qui ont d’ailleurs beaucoup à y gagner puisqu’elles disposent de tonnes d’informations transactionnelles sur leur clientèle et créent des processus en tout genre, sont mises au défi : celui de se servir d’un tel trésor pour tester elles aussi de nouveaux services à valeur ajoutée dans un délai le plus court possible.

Momentum

Jean-François Vanderschrick est Head of Marketing Analytics & Research chez BNP Paribas Fortis : « Ce qui me fascine, c’est moins la multitude des données disponibles et des objets connectés que tout ce que la technologie permet désormais d’en tirer. Pas un jour ne se passe sans que je ne sois surpris par quelque chose de neuf. JP Morgan détecte des tendances en achetant les photos de l’occupation des parkings des supermarchés. La Chine développe la reconnaissance faciale pour adapter le lay-out de ses interfaces à l’expression de ses clients. Vous pouvez suivre à la trace votre paire de chaussettes made in USA de son expédition jusqu’au moment où elle franchit le seuil de votre domicile… Tout cela fait partie de notre quotidien au moment même où une banque manifeste ses intentions de s’adapter à la phase de vie que traverse son client – celui qu’elle suit depuis qu’il est actif – pour lui offrir juste ce qui lui est utile. »

Chez BNP Paribas Fortis, le management data franchit récemment un nouveau pas avec la nomination d’un Chief Data Officer membre du Comité Exécutif, Jo Couture. Ce qui signifie aussi des renforts humains, de nouveaux outils analytiques et de nouvelles capacités.

Jean-François Vanderschrick : « Les data analytics doivent nous permettre d’améliorer l’expérience client, ainsi que de garder les coûts sous contrôle et in fine, cela conduit généralement à une plus grande efficacité. »

Selon lui, la courbe d’adoption entame à peine sa phase exponentielle.

Le timing est aussi important que le service lui-même

Les données servent une multitude de domaines : excellence opérationnelle, marketing, détection des fraudes, risque crédit… Les entreprises comprennent désormais qu’elles doivent transformer leurs données en connaissances et en services et bon nombre d’entre elles ont tout pour y parvenir. Toutefois, il convient de ne pas se laisser noyer par la masse d’informations. Le plus compliqué - et source de frustration - est sans doute de pouvoir accéder aux données et de parvenir à les qualifier. Les aspects de compliance ont naturellement tendance à freiner les développements. Réduire le data to market reste cependant un défi majeur car souvent, le timing de la mise sur le marché s’avère bien trop long. Il s’agit aussi d’offrir un service en temps réel, comme c’est le cas chez Monoprix qui analyse le processus de traitement de 200 000 commandes quotidiennes de ses 800 magasins pour intervenir directement sur sa chaîne d’approvisionnement, un processus critique pour l’enseigne française.

« C’est une délicate alchimie à produire entre les tests (la maquette du service est souvent très chouette, mais encore faut-il réussir la généralisation), la mesure du risque et la ‘prioritarisation’ des objectifs », soutient Jean-François Vanderschrick.

Eduquer l’algorithme

Pour peu que l’on dispose des données et de la technologie, et qu’il y ait des enjeux financiers liés, l’imagination reste notre seule limite pour libérer la valeur des données. A côté de projets conséquents et complexes, des quick wins relativement simples sont ici aussi tout à fait possibles et souhaitables, notamment pour permettre aux directions opérationnelles de l’entreprise d’effectuer des analyses élémentaires à partir de grands volumes de données. « Aujourd’hui, une variété d’informations qui semblent peut-être anodines peuvent nous éclairer et servir de déclencheur d’actions : un client qui commence à travailler avec la concurrence, qui place des lignes de crédit ailleurs, ou emprunte un montant particulièrement important, traite avec un autre pays… autant d’informations qui commercialement parlant, méritent toute notre attention et qui sont jugées utiles dans 70 % des cas » ajoute le responsable de BNP Paribas Fortis. Analyser le modèle transactionnel d’un client permettrait de prendre de meilleures décisions de crédit. Il est possible d’améliorer de manière conséquente la pertinence des décisions, comparé à ce que nous pourrions faire sans modèle, prétend Jean-François Vanderschrick qui ajoute encore :

« Grâce au machine learning, nous éduquons l’algorithme à fournir des réponses de plus en plus pertinentes.» 

Si ‘Big is better’, est-ce accessible aux petites ?

Grâce au Cloud (espace en ligne), les PME disposent désormais des capacités de stockage - auxquelles s’associe la puissance de calcul nécessaire pour exploiter les données. C’est un des enjeux majeurs des Big Data. Le second est de savoir comment les traiter. Des logiciels de gestion d’entreprise usant de la technologie cloud, style CRM, outil de suivi des commandes ou des coûts de production, traçabilité des fournisseurs, rendent les big data accessibles aux petites et moyennes entreprises. Seule condition : rassembler toutes ses données au même endroit. La différence entre les corporates et les PME se jouera sur le long terme. Mais les PME pour qui un super statisticien serait impayable, peuvent toujours acquérir des études ciblées et enrichir leurs données par des bases externes…

(Sources : BNP Paribas Fortis, Les Echos, Transparency Market Research, IDC, Ernst & Young, CXP, Data Business)
Article

27.12.2016

These 4 giants from Silicon Valley want to seduce your IT management

Already champions in everyday life, Google, Facebook, Slack and LinkedIn are adopting innovative and complementary approaches to convert companies. What strategies are they implementing in order to convince you?

Google: the value of data intelligence

Google is adopting an approach which goes beyond communication tools and suites of productivity apps/services. The company has largely transformed its business divisions so that they can exploit cloud infrastructures, big data, analytics and machine learning as a matter of priority. Two competitors are blocking it along the way: Amazon and Microsoft, but for different reasons. Developers have been using Amazon Web Services for a long time, which gives it a history of trust. Microsoft (Cloud, Office) also has a historical presence in IT departments around the world. In this approach, linked to the processing of sensitive data, Google still needs to evangelise: a company is not as easily convinced as a consumer, particularly when it comes to strategic or confidential data. Its weapon: the power of its artificial intelligence tools to process data silos.

Facebook: introducing WorkPlace, naturally

After more than a year of development with partner companies such as Danone, Starbucks, Royal Bank of Scotland and Booking.com, Facebook officially launched WorkPlace last October. This Facebook spin-off enables organisations to create an internal social network - completely private and secure - within an interface familiar to all employees in their everyday life, introducing head-on competition for already widespread tools such as Chatter (Salesforce) or Yammer (Microsoft). Unlike free Facebook, WorkPlace is billed monthly depending on the number of users: $3 for the first 1,000, $2 for the next 9,000 $1 for over 10,000 users.

Slack: real-time collaboration becomes mainstream

Despite the introduction of Microsoft Teams on its turf, Slack remains confident in its strategy of creating tools that allow greater communication and productivity within companies.

"We find this offensive both flattering as well as intimidating, given Microsoft's means, but we think there is sufficient space in the market for several players", declared April Underwood, VP of Slack at the beginning of November.

A market that Slack has largely contributed to opening and driving, by introducing the concept of real-time collaboration. Its weapon? Agility, despite its still limited size and its proven and copied tools. Result: 4 million active users everyday and constant growth.

LinkedIn: from B2B marketing for... Microsoft

Microsoft Closes Acquisition of LinkedIn at the beginning of December. The transaction, which runs into billions of euros, has been followed closely by the European Commission. Despite a strong position in the business, mainly at a human resources level, LinkedIn needs 25 billion euros from Microsoft to pursue its offensive in the domain of professional tools, in a hugely competitive climate. For Microsoft, the acquisition will enable the company to reach B2B marketing targets such as recruitment agencies, head-hunters and businesses. To explain the synergy sought in simple terms, the CEO of Microsoft, Satya Nadella, gives the example of a meeting where everyone present sees their LinkedIn profile, linked to their invitation.

Discover More

Contact
Close

Contact

Complaints

We would like you to answer a few questions. This will help us answer your request faster and in a more appropriate manner. Thank you in advance.

Is your company/organisation client at BNP Paribas Fortis?

My organisation is being served by a Relationship Manager :

Your message

Thank you

Your message has been sent.

We will respond as soon as possible.

Back to the current page›
Top