The real treasure of Superweek is the people
Five days packed with talks — around 90% of which were top quality — would be more than enough on their own. But the venue on a hilltop in the middle of nowhere naturally draws attendees to the lobby bar or spa, sparking conversations about everything from analytics to working with clients to travel tips for Australia. Your brain runs at full capacity practically nonstop, and even though I approached my first Superweek as a winter holiday, I would have needed an extra week just to process everything and get some real rest.
These informal conversations are ultimately the greatest takeaway of the whole week. Compared to other major events in the field, such as Marketing Festival, Superweek is intimate. Attendance is just over a hundred minds, and thanks to those evenings it is easy to meet and talk with most of them. The large shared tables in the restaurant meant that even an introverted wallflower like me ended up having breakfast with the creators of Google Tag Manager, lunch with the creator of Da Vinci Tools, and dinner with the hosts of the Digital Analytics Power Hour podcast.
Legal vs. Ethical
At Proof & Reason, I primarily approach digital analytics from a technical standpoint. So I was surprised to find that the more philosophical topics around privacy and data handling resonated with me the most. Lukáš Čech from EtNetera Activate wrote more about this topic.
For ethical topics, I recommend studying the Contract for the Web. The idea of regular users potentially being able to monetise their own data in the future is explored by, for example, Tim Berners-Lee’s Solid project.
The significance of the topic was symbolically underscored by the Avast affair. Midway through the week, it emerged that the company had been collecting data on its users’ internet behaviour and selling it via Jumpshot. The whole case and Avast’s response became one of the main conversation topics.
3 things I am taking away about ethics in data
- Just because something is technically possible does not mean we must or should do it. Just because something is permitted (or not prohibited) by law does not mean it is ethical, moral, or acceptable. And there is no single right path. For example, the European GDPR and California’s CCPA are based on similar principles but are not 100% mutually compatible.
- Personal data has greater value than ordinary users realise. Perhaps we will soon get some of that value back — for example through the Solid project mentioned above.
- User privacy is increasingly being taken on by browsers. Cookie banners may soon become a thing of the past.
Back to the maths textbooks
Julien Coquet kicked off the week with a rather “encouraging” talk titled You’ll never be a data scientist, my son. Fortunately, it was not as apocalyptic as it sounds. It just requires getting back to mathematics (which might sting a little, as the subtitle hints: “or maybe you will, but it’s gonna be tough”).
While the community previously buzzed about the “technically-oriented marketer” and everyone was learning JavaScript and Python to deploy measurement and automate routine tasks, today it is becoming clear that an analyst simply cannot do without mathematics.
In a way, it makes sense. Among today’s buzzwords are A/B testing and machine learning. There are tools that do most of the work for you, easy to click through — but without a solid grasp of the fundamentals, it is easy to misinterpret data and in the data-driven labyrinth take a turn that leads not to treasure but into a deadly trap.
That said, you do not need to brush up on matrices and differential calculus right away. An analyst can add enormous value by calculating the ROI of A/B testing like Ton Wesseling or correctly deciding whether to try “multi-armed bandits” instead of an A/B test, like Matt Gershoff.
3 tips for building a relationship with mathematics
- Go back to secondary school or university textbooks. Surprisingly, “for dummies” titles work well in English. You can download the Mathematics for Machine Learning textbook for free (and legally).
- Learn the difference between statistical Type I and Type II errors. And think about them when a tool claims to be 99% certain that variant B beat variant A in a test.
- Only death and taxes are certain. So when reporting, work with uncertainty and express estimates as ranges rather than “precise” numbers.
Events and users at the centre of things
But it would not be Superweek without something more technical.
One of the announced novelties was server-side tag firing in Google Tag Manager. The aim is (among other things) to speed up the website for users. Serving tags from a server also means that ad blockers and similar tools can be bypassed to some extent. On the other hand, this places higher demands on marketers’ integrity and thorough management of user consent — just because something can be done does not mean it should be.
Google Analytics introduced the App + Web tool with a new measurement philosophy in mid-2019. It swaps the data model based on sessions and pageviews for one based on users and events — the same principle that powers Firebase and most other analytics tools for mobile analytics. Although A+W is still in beta, it has come a long way since summer and it makes sense to deploy it as secondary measurement for an alternative view of the data. Details can be found in Krista Seiden’s setup guide.
Another advantage is native raw data import “for free” into Google BigQuery. This means data is no longer fixed and aggregated as dictated by the Google Analytics interface. You can report on it however you like. It should be noted, though, that while the data import itself is free, BigQuery is only free up to a point, and actively working with larger data volumes will cost something.
3 things to keep in mind with App + Web
- For now, it cannot do many things. So deploy it as secondary measurement.
- On the other hand, it can do a lot of things regular Google Analytics cannot. For example, you can track external link clicks without needing JavaScript skills or Google Tag Manager.
- Even though it is currently free, working with data in BigQuery can incur some costs. For Google it is therefore an elegant model for reasonably monetising Analytics for websites that do not need paid GA360.
Think about people
You can start by visualising and presenting your data in ways that reduce the cognitive load on the recipients. For example, following Tom Wilson’s tips.
Even better is recognising that most people are not interested in data — they are interested in insights and results. Analysis itself is rarely the end goal; in most cases it is a means to an end. I was surprised how many people found this a revelation — even though Avinash Kaushik has been rallying against reporting monkeys for quite some time now.
Part of the talks and informal conversations also revolved around customer-centricity — not only in the relationship between companies and their customers, but also between agencies and clients. Even though Proof & Reason is very client-oriented and our design process is based on thorough questioning, it is good to occasionally be reminded how to ask better questions, understand client needs, and run an agency in general.
3 actionable tips
- In dashboards, maximise the data-to-noise ratio. A simple trick: if you were running out of printer ink, what could you remove from the chart?
- Most customers do not care about the drill bit’s specs — they care about a hole in the wall that is easy to make. Data and charts should not be the sole output of a real analyst’s work.
- Make sure you and your client are speaking the same language, solving the same problem, and working towards similar goals. That requires knowing how to ask good questions.
See you next year
Superweek was an incredibly intense experience. For European analysts, there is probably no better opportunity to absorb and share expertise. It reliably works as a motivator and a trip outside your comfort zone. And nothing helps with networking quite like running away from sparks flying off a large bonfire.
Conference highlights can be followed on Twitter under the hashtag #SPWK. All presentations were compiled by Marek Lecián. The gallery is on Facebook and the first videos will soon appear on YouTube — in the meantime, you can draw inspiration from previous editions.
Article by: Jan Kadlec, data analyst
