SuperWeek 2022
It is over. Five days of analytics in a remote Hungarian hotel (this time we moved to Saliris Spa Resort, because the former place is under renovation).
It was fun. It was informative. It felt a bit like a college reunion. This was my third time and I feel gratitude that I’ve had a chance to fit into this brilliant pack. Although Robert Petković pointed out that a group of people that enjoy themselves by talking about stuff no one else understands, isolated in a remote building resembles a mental institution.
Elephant in the room
This year’s main topic with no doubt was the establishment of a government of privacy. Through the GDPR a couple of years ago we realized that something needed to change. Through the latest progress we get the assurance that it is inevitable. At the same time we can see quite a sound guidance on how it should and should not be achieved. The do’s are obvious. Having full control over the smallest pieces of information that we collect (ideally via Server-side Tag Management solutions — Server-side GTM presented by Simo Ahava or Jentis presented by Thomas Tauchner) with proper consent and appropriate care is mandatory. But still this is not enough. The legislation and the culture varies and if the same level of protection is not guaranteed, we must swallow a bitter pill and use services that might not be the most convenient or use setups that may introduce additional complexity and unpleasant consequences. The latest cases (did someone say that Google Analytics is illegal in the EU now? as well as TCF by IAB Europe?) were discussed and we all appreciated the deep knowledge and corrections during the discussions from Aurélie Pols.
But there is a risk. If the EU says that US companies can’t have EU citizens’ data then why would US companies share the US citizens’ data with EU companies? It makes sense, because in a way privacy as a concept is based on one’s isolation. To protect ourselves from the attacks of the hostile environment, we hide who we really are by creating an individual safe space surrounded with obstacles. But this approach always introduces some collateral damage. The system also protects villains and introduces unnecessary difficulties for some of the favorable deeds.
So maybe there is an opportunity to introduce more radical change and forsake the old broken system. It is always refreshing to hear about the latest ideas Kris Ewald has to share with us about the way to decentralize the ecosystem and give power back to individuals to decide what should happen with their personal data and what benefits they get in return.
Purpose is the king
I have been thinking about a word that sums this brilliantly and it is the Purpose. We have to be absolutely clear about the purpose for which we ask the customer for his data as well as we need a deep understanding of what activities are there to perform in order to fulfill that purpose and which tools are suitable for such a task. And it won’t be easy to juggle with the purposes at will like we used to do not that long ago. The thing is that by the rules of the purpose limitation, we should be always able to separate the processing to a single purpose no matter what and that may not be that easy with solutions that are multi-purpose by design.
The king is crowned by Actionability
As analysts — usually introverted nerds — we get easily seduced by complex issues and try to solve them just because it is challenging. Or because we feel the opportunity to be the first pioneers to solve the newest shiniest puzzle. It won’t help much if we focus just on the latest buzz around Web3, metaverse or newest proposals that try to be the magic bullet to solve the shift in privacy perception (Did someone say “Flopics”? — I am looking at you, Doug).
If it is obvious that we will have less data, we should make sure that those are the most important, reliable and actionable. And when we are big enough as well, acquiring machine learning skills to fill some of those blanks will surely not hurt (as mentioned by Doug Hall in his opening presentation about the 18 months roadmap for the analysts). But at the same time machine learning skills by itself doesn’t help us to mature just because we can do more complex stuff. Maturing still depends on our ability to apply those skills when they are appropriate and solve business problems with them.
And when we talk about maturity, when it comes to actionability, it is usually not an issue of an isolated person in analytics. It is a challenge for the whole organization and that challenge might be addressed only with the right culture (as pointed out by Michael Helbling). If our organization doesn’t have it now, it may seem impossible to achieve. It requires strategic thinking (focusing on the process, promoting the results), determination (start with yourself), consistency (feedback loop full of reviews) and a constant self-improvement (finding that calm perseverance of a poker player).
“None of us are as dumb as all of us.”
What else?
And that brings us to the second theme. I would summarize it like:
“Hey guys, we do magic stuff with the data, but if we want to get recognition, we need to work on our communication skills as well.”
We need to focus on delivering the expected value (as Ilya Chukhlyaev presented from the study, that according to managers we are rarely able to meet those expectations). And this goes hand in hand with the ability to understand the businesses we are trying to support with our insights from data. The numbers in our tables are not isolated. They usually represent just a portion of the real life situations and customer journeys. As Steen Rasmussen pointed out, the context is the king here. It may seem easy to walk 78 kilometers in 2 months. Unless you realize that those 78 kilometers are an expedition to Mount Everest for which just a marginal share of people is capable of and tens of people die trying annually. And we need to try harder to understand the context in which we collect the data and do our analyses.
Up your game
We also got advice and inspiration on how to acquire or improve our skill sets. Astrid Illum shared practical tip on improving the customer experience by using knowledge gates (using triggers like a specific copy or imagery to address specific target audience), Ashit Kumar from Spotify shared their progress with scaling the experimentation culture, Matteo Zambon and Roberto Guiotto showed their tools and skills with Google Tag Manager, Yehoshua Coren gave a ton of tips on SQL in BigQuery, Tim Wilson shared his tips on how to think about attribution, Krista Seiden shared her almost-insider knowledge about Google Analytics 4, Robert Petković showed us how to perform onstage and David Vallejo demonstrated his deep technical knowledge of the Chrome Developer tools to charge our debugging skills.
And last but not least if we want to improve we should be pretty certain (in the limit) with the basics that underline most of the work we do. Therefore Matt Gershoff shared his deep understanding about the concepts behind the inference and uncertainty and we all got #mattgershoffed once again.
Let’s summarize
This was my third time at SuperWeek and I enjoyed the whole event a lot more thanks to the fact that I managed to utilize some of the venue benefits — I usually got up early and went for a run, visited the saunas twice and went swimming after dinner.
This enabled me to process the presented information with ease in comparison to my previous experience when I spent all the time in the follow-up discussions till the late evening by a glass of beer and that was usually pretty overwhelming.
And as usual I tried to spice up my attendance with some fun as well.
You can find some more of my buzz contributions on my Twitter: https://twitter.com/cataLuc
