The 2018 edition of the Digital Analytics Forum (DAF) brought more than 350 analytics pros together on the 8th of November at the Pavillon Vendôme in Paris. Each year, it’s an energising time for AT Internet’s teams, customers and partners: participants discover the latest Analytics Suite features, of course, but they also come to be inspired, to gain new perspectives on their daily work as analysts or marketers, and to share experiences with their industry peers. And as always, we like to make the day as enjoyable as it is useful! The event took place in a premium venue perfectly suited for our special 10-year anniversary edition. It was the perfect backdrop for rich exchanges and tasting of Grand Cru Classé wines – the DAF’s trademark since it first began. Below we look back at the 6 biggest highlights from the day. 


For ethical analytics: Mathieu Llorens’ keynote 

Internet has changed our lives, for the better, but sometimes for the worst. The World Wide Web, as it was imagined based on a principle of decentralisation, is today being cannibalised by a handful of platforms which have built their success on tapping into the reward pathways in our reptilian brain: distraction (notification), social interaction (social networks), fear and anger (algorithmic filtering of shared content) … so many techniques meant to create addiction and harness users’ attention. The consequences of these models are beyond problematic; not only do they generate major cognitive issues and widespread depression amongst the youth, but they also severely impede the democratic process by making a game out of fake news and populist factions which play on fear and anger. Armed with figures from recent studies, AT Internet CEO Mathieu Llorens described this phenomenon and emphasized the data industry’s responsibility to integrate an ethics dimension into its projects. It’s no longer about asking what technology can do, but rather what it should do. 

This is why AT Internet offers a three-fold commitment which aligns the interests of users, companies and analytics tools:  

  • Data and its processing must be understandable for everyone (our customers, Internet users, organisations) in order to guarantee transparency and lawfulness, and also to ensure that created value goes back to customers and Internet users – and not the service providers. 
  • Data must be reliable… It seems so obvious, yet it’s still far from being reality in the digital analytics world, particularly when the tool used is provided by a corporation whose primary interest is misaligned with that of its customers. 
  • Data must serve the user – and not the other way around.  

“Machine Learning in Marketing”: the DAA’s Jim Sterne’s presentation 

We were honoured to welcome Jim Sterne, founder of the Digital Analytics Association and one of the industry’s leading experts, at this year’s DAF, where he delivered a riveting talk on applying AI and machine learning to marketing strategies.  

To start, Jim first cleared up the mystique surrounding the term “machine learning”. Forget what you’ve seen on TV, he says: AI cannot simply be equated with the latest trendy science fiction film. And then, what is machine learning good for? A few examples: robotic assistants in your favourite store, augmented reality that’s already available on our smartphones, and chatbots. As a brand, you can use this technology to extend your customers’ brand experience, and to attract new customers – but be warned that it won’t necessarily work for everyone. Here are his tips for getting your machine learning projects off to a strong start: 

  • Set a clearly defined objective: and be specific! “Boost turnover”, “reduce costs” and “grow customer satisfaction” are all rather vague goals that encompass many ever-changing factors. On the other hand, “increase the number of clicks on my ad banner” is a more precise goal, which is therefore much more compatible for machine learning.  
  • Be ready… to be disappointed: like us humans, the machine requires some non-negligible learning time. Your first tests will probably not be too satisfying, and you should definitely warn your management of this before starting out on a machine learning project. But rest assured, after a difficult start (be patient while the machine is learning!) the progress will be rapid and dazzling, before eventually plateauing. 
  • You must be able to fully trust the data you’re injecting: clean, reliable and consistent data (just what we happen to provide). A solid machine learning project requires a significant volume of data in order to work: so be prepared to spend a good chunk of time regularly verifying your data streams. 
  • Be human: Machines are fast and accurate, but stupid. And on the contrary, we humans are rather slow, not always accurate, but intelligent. Reason, common sense, emotion and empathy are all human traits that machines will never have. But together with machines, we can accomplish amazing things! 

A talk that’s definitely worth (re)watching: 

Data science at work in the Analytics Suite 2 

Take Explorer, the central module of the Analytics Suite 2: thanks to machine learning, you can now use the “data science” feature to detect anomalies in traffic patterns and even predict how traffic will change over the next 7 days.  


Explorer can now detect anomalies in your traffic, indicated by the circular symbols along the line graph. The confidence interval is clearly defined and indicates the expected values based on your traffic history. 


Another application of machine learning in the Analytics Suite 2: the RFM (Recency, Frequency, Monetary) analysis available in the Sales Insights module. E-retailers love it, as it categorises logged-in visitors according to their purchase behaviours. The advantage here is that this categorisation is intelligent and automatic, thanks to data science modelling done on massive volumes of data. That’s what we call actionable value – and that’s just the beginning!  


New in the Analytics Suite 2

So many new features were announced at the DAF 2018 that it would be impossible to mention everything here (you can find all the details on our website), but we wanted to highlight those that had the biggest impact during the event. 

Sales Insights, our e-commerce tool: KPIs dedicated to retailers, tracking across the entire purchase process (adds to cart, shipping, payment, transaction confirmation), integration with PrestaShop CMS and Lengow product feed manager: you’re ready to optimise your conversion rate! 

Navigation  provides the tools you need to understand your visitors’ pathways: analyse entry pages, get a funnel view of a particular page sequence, and view a sunburst graph for a synoptic yet comprehensive view of navigation steps taken on your site or app. Navigation is currently in beta but will soon be open to everyone!  


CDDC, or Custom Domain Data Collectionby using a custom domain name to collect data, you can work around the negative effects of adblockers on your traffic data (as adblockers can cause up to 30% of traffic data to be lost!). Learn more about the impact of adblockers on web analytics tools in this article. 


GDPR: Taking stock 6 months after legislation came into effect 


“With great power comes great responsibility”: as the processor of your analytics data, we felt it was critical to take stock of the situation 6 months after the new European data protection regulation has taken effect. Armand Heslot, privacy expert at the CNIL (France’s national data protection authority) took the stage to remind us all of our GDPR-related obligations as data processors and controllers. Along the way, he dispelled certain myths about the GDPR and the ePrivacy directive… For those of you who haven’t yet sorted out the protection of your users’ data, the first enforcement penalties are being distributed (see the Information Commissioner’s Office current actions for an idea of fines being doled out). Our DPO Nicolas Boudillon then presented the various privacy-related features in the Analytics Suite as well as the purpose of data collection for our customers and on our own websites. 


Your data, unsiloed: the AT Connect programme 

What good is your quality data if you cannot use it the way you want? Our API is always available , but we also offer an ecosystem of partner solutions with which we’ve built bridges to make things easier for you. The DAF was a great opportunity to meet some of these partners, experts in adtech (Kamp’n), product feed management (Lengow), SEO (OnCrawl) and data visualisation (Reeport).  


Pavillon Vendôme, the perfect setting for the DAF’s 10th year 

Situated on one of Paris’ most renowned squares and belonging to Potel & Chabot (luxury caterer!), the Pavillon Vendôme was the ideal venue for welcoming our customers and partners for this 10th edition. Foodies and wine lovers were duly sated. And for fans of Money Heist or Inside Man (depending on your tastes), the conference room was a converted bank vault whose massive door and mechanisms made quite an impressive entryway. We’re serious about security! 

Until next year’s DAF, with a little luck and (artificial?) intelligence, we’ll finally figure out the answer to Jim Sterne’s challenge: chihuahua or muffin? 




Alexis joined AT Internet in early 2014 after obtaining his master’s degree in digital communication and community management. Passionate about social networks and online advertising, he is also in charge of marketing automation at AT Internet. A proper geek, he strives to stay on the cutting edge of all innovations related to social networks and new technologies.

Comments are closed.