Frederik Werner is a Senior Digital Analytics Specialist at DHL and formerly led digital analytics, optimization, and personalization at RTL Disney
He is the author behind Full Stack Analyst and has quickly been recognized as one of the top bloggers on Adobe Analytics and the latest trends in web analytics.
F.W.: Over the course of those 10 years I learned to be proud of different types of achievements. I still remember the early days, where I was most proud of a clever, sophisticated analytics implementation on a website, often with “new and fancy tools” like tag management systems. I think this is quite common for people who started out with a development background, where technical excellence is a priority.
Once I became more experienced and confident on how to collect data, my focus shifted and I became increasingly proud every time I assembled a large-scale analysis for my stakeholders. I enjoyed feeling needed and complimented on the results of my work, which I spent great amounts of time on optimizing. But thanks to my amazing mentors at that time I learned why it can become a problem if stakeholders are dependent on me spoon-feeding them insights. That allowed me to not only learn how to master a system but also master enabling others to do the same.
Today my perspective is different again. Instead of being proud of what I can achieve on my own, I feel accomplished the most if I can provide value to the people around me. I am not trying to do everything on my own anymore but tend to look at the bigger picture: How can I build the perfect, scalable analytics system for my company? What is the best way to ensure analytics will be a priority during product development? How can I enable my stakeholders to be data-driven and self-sufficient without my direct involvement? Every time I succeed in building a solid business process or see how well my stakeholders are using my systems without me, I feel most proud today.
F.W.: I like to start with an exploration phase, where I talk to senior management and try to understand the current long-term strategy of my company. It is also important to understand where the business is standing in terms of analytics maturity, as this will influence management buy-in and stakeholder commitment. Exploring all of this helps me in architecting a solution that is both feature-rich and sustainable for my team.
Depending on the previous outcomes, there are some natural restrictions implied by team size, technical complexity, and available budget. It is not unusual to find a discrepancy between the level of ambition and budgetary commitment, which is why I emphasize transparency and education towards management when setting the stage for my later work.
Once I aligned with management and know where my company is headed and what resources I will have, I can finally start building the tool stack. Depending on the corporate strategy and analytics maturity I next define which level of analytical depth and self-service I need to support. If digital is not the first business priority, I will probably spend the most time on simple reporting, so a solution like Google Analytics or Matomo might be sufficient. If digital is part of the corporate strategy, I need to look at bigger solutions like Adobe Analytics that allow me to be truly data-driven and user-centric in a scalable way. Depending on the need (or desire) for self-service, tools like Microsoft’s Power BI might be a good addition to support low-involvement ways to interact with data.
F.W.: At my current company Super RTL, we enjoy a remarkably high level of liberty in regard to how we do our business, which is good, and receive a huge demand for insights from the business. But since we are only taking our very first steps in terms of analytics maturity, there is still a lot of work to do in terms of educating and managing our stakeholders towards the potential of being data-driven and self-serviced.
As a consequence, we need to be very careful with what I can commit my team to, since this phase requires a lot of communication, proactive analysis, and training towards our stakeholders. While it is important to create value and increase visibility, we need to be careful not to over-promise and disappoint our internal customers. Although our systems and processes are in place and running, we are still constrained the most by team size.
F.W.: Good analytics starts with good data. Adobe is constantly extending the feature set of their products, which also means customers need to stay current to get the most value. For example, some years ago we would need to use tracking plugins to get certain parameters that are built into analytics today. This means you can, and should, update your implementation to gather some more useful information in your variables. Also, with Global- and Virtual Report Suites, it has become more important than ever to maintain a consistent tracking standard across all websites and apps. I recommend having implementation audits by an external agency every few years to make sure data is still being collected in the best way possible.
With healthy and uniform data, you should then stress the importance of user identification to your product teams. Logged in users are still the easiest and most precise way to analyze cross-device and cross-product journeys. Tools like Adobe’s Customer Journey Analytics are perfectly tailored to take your analysis to the next level and generate even more value from a healthy relationship with your users.
Once all those prerequisites are met, there is virtually no limit on what you can get out of your data with Customer Journey Analytics and Adobe’s Experience Platform. Some questions I personally find quite interesting: Which users are most valuable to your business depending on where they start their journey? How do your web- or app marketing campaigns influence other products? Should you aim at making your users use either two websites, two apps, or one of each? Those questions already demonstrate the company-wide advantages you can generate with cross-device analysis tools like Customer Journey Analytics.
F.W.: That is one of the things I love about Adobe Analytics so much, since it lets me build those metrics end-to-end within the frontend without any complicated process behind the scenes. I think I would cluster my favorites into three different classes of metrics:
First, I want my stakeholders to be able to judge whether the current development of their KPIs is normal or not. For example, calculating the deviation from a mean for the last 30 days is quite helpful in detecting outliers in data that should be investigated.
Second, we can model how the future might look like using predictive analytics. This allows my business to answer the question if they are going to reach their goals if the current trend continues. Linear Regression Analysis is a great way to create a corridor of expected values and comes with a measurement of probability.
Third, our product teams should be creating lots and lots of A/B tests on their products. While tools like Adobe Target have some form of built-in significance tests, adding custom calculations can really help to bring certainty to testing results.
F.W.: On a high level, digital analytics can be described as collecting behavioral data from our customers to allow our business to makes educated decisions and empower them to behave differently based on insight. In psychological terms, we analysts try to extend the behavioral repertoire of our stakeholders to allow them to act in an optimal way. I like to stress the involvement of real humans across the value chain to avoid over-objectifying the way we run our businesses.
To be more concrete, this mindset starts with product development and the way our business tries to bring value to our users. Human-centered analytics can bring value here already, for example by estimating the size of a target audience for a new feature. Product teams should ask their analysts questions like “we are evaluating if we should build a feature for commuting customers, so how many users do we have that use our apps from more than two locations per day?” to decide on their feature priorities and project incremental revenue.
When analyzing digital products, we need to translate questions like “do users like our new feature?” into behavioral patterns that we are able to analyze. For example, the aforementioned question could be translated to “how many of our loyal users have used this feature more than once?” or “how many users use this feature at least once per day for every day of the week?” which can be answered using analytics tools.
A good analyst should be able to turn those rather broad questions into concrete patterns and analytical approaches by being empathetic toward the customer. As a tip, always imagine how users are behaving in a certain situation or how they would behave if they had a certain mindset. Try to focus on a broader behavioral pattern (like “use a feature”) instead of microtransactions (“clicked a button”). And also, think about the imperfections in your approach and try to estimate or analyze how common your proposed patterns really are.
F.W.: One of the most uncomfortable facts I had to bring to my managers is what I like to explain as the difference between a one-man-band and a symphonic orchestra. Sure, both can provide some musical entertainment, but the experience is not even remotely similar! At the same time, going from a one-man-band to a duo is a huge step already. Analytics is one of the areas where real commitment in terms of manpower can drive change unlike any other field of activity simply because of allowing more time and a more nuanced approach to analysis.
A follow-up question is then: Once you actually have hired your analytics team, what are they actually supposed to do all day? If you want to transform your business towards being digital-first and data-driven, you cannot bury your analytics team in stupendous reporting tasks. Instead, they need to enable your business to make data-informed decisions at scale, requiring self-service, which demands for education and mentorship of business stakeholders. That in turn changes what you should be looking for when hiring your analysts. Remember, there is little use of having a data-driven analytics team when the rest of your business does not care about data or does not want to actually change their behavior based on new insights from data.
F.W.: The telco business is incredibly competitive. Pricing actions from your competitors may have quite some impact on your day-to-day sales performance. The same goes for price comparison sites and aggregators. If you manage to build a transparent monitoring system to keep the others in check, your sales team will love you. As far as onsite analytics go, take a close look at how long your sales funnel is. Are there steps that could be left out before the purchase? Should upselling happen before or after the purchase? Experimentation can go a long way there. Also focus your data quality efforts on those key interactions so you avoid misreporting and losing trust.
For media companies, product and content analytics is most important. Depending on your business model, retention or engagement might be the top priority from a product perspective. Try to look at content in the same way and cluster it depending on its function, defining up-front if it should attract new users, engage or convert existing users, or reactivate inactive users. It is a real challenge to enable your business to be self-serviced with a vast number of contents and features, so take your time standardizing the way you analyze both. A shared view on how performance should be evaluated will go a long way.
F.W.: With all the recent developments towards GDPR, COPPA, ITP, etc., privacy and data protection will be a top priority. That should not surprise anyone.
But what will that mean for digital products? Is staying anonymous worth sacrificing products that provide genuinely interesting content to us? I do not think so. Instead, earning user’s trust will become imperative to be able to conduct a sustainable business practice. A clear value proposition from a trusted brand will still convince users to give businesses information about them. At the same time, this will challenge businesses to rebrand themselves as trustworthy with data. For analysts, this will finally mean the end of the “track everything”-approach and respecting our user’s preferences. But we might profit from a move towards high authentication rates and login-based products. We need to focus on the datapoints most important to our business and be ambassadors of a human-centered product design approach.
One thing I have mixed feelings about are the recent efforts from browser and OS vendors to limit tracking on a technical level. While this certainly has value for non-trusted brands, it just means more effort and a worse experience for companies you do actually trust. Browser limitations mixed with a move towards logins means a horrible user experience for your favorite websites, since you would need to re-login dozens of times. While this trend is caused by companies exploiting well-intended features, I have high hopes for some positive change once brands start working towards truly user-centered experiences and regain user’s trust. Once we learn to communicate the value we offer through analytics and personalization, we will see a surge of possibilities with analytics
Another very exciting trend is the move towards a more sophisticated analytical system that is still accessible and pleasant to use by business users. Especially Adobe has made some significant effort to build an ecosystem that allows for simple reporting use cases just as well as sophisticated deep dives, which were traditionally exclusively reserved for CRM or Data Science teams. It is an amazing achievement to open up the superior analytics capabilities of Adobe Analytics to any data we could possibly dream of analyzing. Literally, every user of Adobe’s Customer Journey Analytics can build a segment of people who used both call center and website to analyze how their retail conversion rate is. I hope more companies will follow this path of democratizing analytical capabilities and thereby making it accessible to the whole company. Concepts like the citizen analyst or citizen data-scientist will finally come to life with systems like Adobe’s Experience Platform.