OWOX BI Research on the State of Digital Analytics: Interview with Tim Wilson
The questions for this interview were prepared by Mariia Bocheva, OWOX BI Business Development Executive, after she met Tim Wilson at the Superweek conference. All the photos were kindly presented by Bánóczy Zoltán, Founder of Superweek – European Analytics Summit.
It was a great pleasure to get such deep insights from Tim Wilson (follow him on Twitter), an active analyst with tons of experience and a positive view on life.
Tim has worked with the many dimensions of marketing and customer data for over 15 years. He ran the business intelligence department for a $500 million high-tech business-to-business company and has consulted with multiple top 50 internet retailers on how to apply digital analytics to their business. Tim is a marketer-friendly data geek. And now he’s Senior Analytics Director at Search Discovery.
Search Discovery is a business intelligence and analytics company that empowers organizations to use data to improve business performance.
Table of contents
- Skills & resources for analysts
- Companies and the "holy cow" of data
- Problems and challenges
- OWOX BI bottom line
Skills & resources for analysts
What hard skills are most important for analysts today? Does an analyst have to know SQL, Python, and R and build dashboards in the most common visualization tools like Data Studio, Tableau, and QlikView?
This is one of those pretty hotly debated topics of late. Let’s start by saying analysts have to know spreadsheets — Google Sheets and Excel — really well. Even if that’s not where they’re doing most of their work, they’ll almost always be getting and sending out information using spreadsheets on a fairly regular basis.
And if an analyst isn’t very comfortable with VLOOKUP and pivot tables, then they’re going to struggle with any other platform where they’re doing analysis.
From there, comfort with a BI/visualization tool (I’d add Power BI to the list in your question) is increasingly a requirement, although there are still plenty of companies that have yet to invest in one. I’m a big fan of Google Data Studio for that reason — even if the company hasn’t made the plunge into a paid platform, analysts need to know how to automate some base visualization and exploration of their key data in a way that can be shared with their stakeholders.
SQL/Python/R are where things get tricky.
It’s becoming fashionable to insist that analysts must have skills there, but I’m, honestly, not so sure.
I see plenty of analysts whose days are quite full doing valuable work without having ascended the very steep learning curve required to gain fluency with those languages. But there is a limit to what can be done with Excel and BI platforms..
For an analyst or an organization that has high volumes of granular data they want to put to use in models, SQL and Python or R quickly become a must-have.
And I would add a third class of hard skills: statistics. This means going beyond just summary statistics like mean, median, and standard deviation, and truly understanding how and where regression and correlation — and p-values and R-squared — can be put to use. This is as much about actually gaining a deeper understanding of the nature of data as it is about actually deploying those statistical methods.
What soft skills should a good analyst have?
There are many! Effective communication skills are at the top of the list, and that includes everything from one-to-one interactions with stakeholders — listening to and understanding their needs in an active manner — to applying best practices in data visualization to being able to present the results of an analysis in a way that’s clear and comprehensible.
A good analyst also is curious (about the business, about the data, about the world), skeptical (about what the data shows — any time there’s a surprise, he/she assumes it’s a data issue and digs in to confirm), and has a high degree of perseverance (to overcome data challenges and even people/process challenges within the organization). These may be more "character traits" than "skills," but they are all pretty important.
Do you think miscommunication between analysts and marketing teams is common? If yes, do you have any recommendations on how to overcome it?
Absolutely! Analysts and marketers speak different languages. In my experience, it really needs to fall to the analyst to "speak marketing." That comes down to doing a few things:
It can be tempting to start problem-solving (What data can I pull to answer this question?) prematurely and not actually probe for the underlying business problem the marketer is looking to address.
- Not speaking analyst
Some education of the marketer is fine — what a metric means, what the limitations are in the data — but I’ve seen analysts slip into deep analytics terminology wayyyyy too quickly when it’s not necessary.
That can leave the marketer confused or, worse, feeling uncomfortable or put down (I don’t know what the analyst is talking about. Am I supposed to? He/she seems to think I should!).
- Never, never, NEVER telling yourself that stakeholders are stupid
That’s a fatal mistake — deciding that a relationship challenge is a matter of intelligence rather than an issue of communication.
What professional resources or events can you recommend for analysts?
The top resource I can recommend for analysts, regardless of where they are, is the Measure Slack team. It’s free, and it’s got thousands of analysts constantly engaging and sharing with each other. Every few weeks, there’s a discussion thread about What are good resources to learn X? or What are good conferences to go to for Y? so it’s even a resource for identifying resources!
Separately, there are some great conferences: MeasureCamp is a free unconference that gets put on around the world (and if there isn’t one near you, then you could start one!). Marketing Analytics Summit and Digital Analytics Hub are great conferences in the US, and Superweek is a fantastic option in Europe.
The Digital Analytics Association is continuing to grow the resources that it offers to its members, so that’s a great resource with all sorts of great material (and even a mentoring program). There are a number of MOOCs like Coursera and edX that offer any number of online courses.
Finally, if I went through this entire interview and didn’t plug the Digital Analytics Power Hour podcast, I would be a terrible marketer! But there are numerous analytics, machine learning, and AI podcasts out there — too many (and too many that I don’t manage to listen to regularly) — for me to try to make recommendations.
30 handpicked Google Data Studio dashboards for marketersDownload now
Companies and the "holy cow" of data
What knowledge are analysts and marketing specialists missing in order to make companies data-driven?
Business context and a deep understanding of uncertainty. The former, I believe, as I constantly see both analysts and marketers jump into the data wanting to "find answers" and "generate insights" without having clearly formulated a business question for which the answer will lead to a likely action.
It’s the tyranny of expecting the data to "provide answers" when there hasn’t been a clearly formulated question.
The latter is a bit trickier to articulate. But going all the way back to John Wannamaker noting in the early 1900s that "Half the money I spend on advertising is wasted; I just don’t know which half," marketers and analysts have been expecting data to provide "the truth" rather than learning how to operate in a probabilistic world.
While we may know "the truth" about the past, like How much traffic came to my site last month? (and even that is not "true" — there is a lot of messiness about the data collection there that even makes that number just an estimate), effectively putting that data to use to impact the future means developing a deep understanding of how "true" a picture of the world the data is painting, and then operating accordingly.
What are the most important things analysts need to do at different stages of business maturity (startup, SMB, SME, enterprise)?
I’ve never actually considered that question. It’s an interesting one! Startups, I think, tend to be more strapped for resources and, generally, more able to take on some data risk.
So startups often — out of necessity — need to be scrappy and cobble together free tools with limited data governance and some potentially fragile (or even manual!) integrations. And that’s a fine way to operate, as long as there’s a recognition that, with growth, there will need to be a maturing of the systems.
That generally means that data platforms will need to get overhauled and even replaced entirely.
On the other end of the spectrum — enterprises — governance becomes a really big deal, as there are more people and processes impacting and relying on the data. And, typically, there are more resources to devote to analytics, so some of those can (or should!) be responsible for governance and processes — minimizing the risk of a failure that impacts a large number of roles and, potentially, even the near-term top line of the business.
What difficulties do you see when it comes to implementing analytics and how would you assess the overall development of the market?
If we limit "implementing" to the implementation of data collection platforms, there are a lot of challenges:
- Single-page apps (SPAs) are very much in vogue, and those introduce tagging challenges.
- The macro-level consumer shift to mobile — and the increased likelihood that they are engaging with brands across devices — has made cross-device tracking more important for many organizations, and there’s no straightforward way to accomplish that (it depends on the nature of the company and the site).
- Privacy regulations like GDPR and (in the US) CCPA have added additional restrictions and considerations to the data collection world.
- The blocking of tracking and deletion of cookies is increasingly shifting from ad blockers installed on a tiny fraction of users’ browsers to being standard defaults within the browsers themselves, which certainly can muck with the data!
What’s the biggest mistake an analyst can make? Can you share some of your analytical mistakes?
The biggest make, I think, is getting so excited about fiddling around with the data that the What am I trying to achieve here? question gets lost.
I’ve had that happen more times than I like to admit — I’ve started with a perfectly valid business question, then got sucked into the mechanics of the underlying data and its complexity and my ability to conquer that complexity such that, by the end of the project, I had done some pretty interesting things with the data itself but I hadn’t delivered something that was really moving the business forward.
What do you think is the future of marketing analytics? What trends do you see coming and what’s in high demand?
Despite my earlier thoughts about needing to really focus on the basics before diving into data science and machine learning, I think we will continue to see more and more companies putting machine learning to effective use.
The biggest reason for this is that, increasingly, organizations have access to raw data about their customers and prospects — behavioral, observational, and demographic.
10 years ago, for instance, digital analytics platforms primarily provided aggregate data. There could be multiple dimensions and metrics in a report, but even so, the data available wasn’t at a session level or person level.
That’s been shifting as the major players in the digital analytics market have started making that hit-level data available, and any number of newer martech platforms have that level of detail available from the get-go. And that’s the data that’s needed to really get value from machine learning techniques.
Problems and challenges
What problems do you see on the market today?
A big problem I see is that there are an enormous number of venture-backed startups that are, essentially, promising the market that they have a magical AI wand for sale. In many (most) cases, that is marketing hype.
But because technology is higher margin and inherently more scalable than people or processes, that’s where the marketing dollars in the industry go, which means "the market" gets bombarded with a message that, if an organization just buys and implements the right technology, they will suddenly become a data-mature, AI-powered organization.
Reality just does not work that way, so many, many organizations are writing large checks for technology while starving themselves when it comes to analytics staff and internal investment in analytics process improvements.
What analytical challenges do you have right now?
I still have to continually educate my clients about the importance of getting the basics in place: clearly established KPIs, data collection that is reliable and robust, automation of repetitive tasks, and analyses that are driven by clearly articulated hypotheses. That’s a challenge that really hasn’t changed over the past 15 years.
A more recent challenge is the excitement around data science, machine learning, and AI. Part of the challenge is actually ramping up the tools that are used in those areas, and part of the challenge is keeping clients from wanting to chase these shiny new objects when their highest value would come from focusing on the basics.
OWOX BI bottom line
Thanks for so much great advice, Tim! We hope that more talented analysts will follow it and that the whole analytics market will become more educated, enthusiastic, and professional.
Let’s keep in touch! Subscribe to our blog to read the next interviews from our research on the state of digital analytics and adapt the experience of top analysts to your business.
What would you ask Tim about? Let us know in the comments below!