Why doesn’t anybody high-five the analyst? Hard but true answers from Steen Rasmussen
Mariia Bocheva had a great interview with Steen Rasmussen, one of the most talented and influential web analytics evangelists we know. We also got a recording of the talk he gave at the Analyze! conference where he covered the details that help us form a coherent picture of our research.
We’ve decided to break from our traditional interview style and turn this into essential reading that will help analysts come up with new ideas.
In this article:
Table of contents
- Core problems of all analysts
- Existing challenges you may not have even noticed
- Existing challenges you may not have even noticed
- Where is the best place to hide a dead body in analytics?
- For homework: Why data is not modern oil
- To sum up
First, a few words about Steen Rasmussen: He’s currently the Web Analytics Evangelist, Director of, and a Board Member at IIH Nordic. He can also rightfully claim to be one of the framers of modern analytics. IIH Nordic is one of those companies with a totally unique data ecosystem and an experimentation-driven culture, where everything is designed to promote meaningful and efficient work. For instance, they have such a level of organization that they work only four days a week while handling the workload of a five-day work week.
Sounds like a dream, right? Let’s find out what Steen has to share with us.
Core problems of all analysts
Caring for skills, machines, and tools more than for understanding why you were hired
Who told you that to be a good analyst you have to be stuffed up with skills and tools? Or that being an analyst is a safe job where you do reports and then relax until the next time reports are needed?
Having a clear understanding of why you were hired and what value you bring to the company is essential for your success. An analyst works to provide insights for businesses in order to accomplish two main goals:
- Decrease costs while increasing profits
- Acquire new customers and retain existing customers
This is the primary aim of an analyst’s work, and each task performed and report built should start from the goals described in the slide above.
What [the] analyst can say from this point of view is that we can decrease the cost of acquisition, we’re able [to] shorten the time of acquisition, we’re able to increase profits on the individual customers.
And this is not the language we normally speak when we talk about data. We talk about bounce rates, pageviews, and sessions. At the end of the day, those questions make all the difference [as to] where we are actually having an impact.
Change the narrative into, “we’ve increased the average profit rate of a customer by 25%” and you’ll get the high-five.
Analysis tools, technologies, and approaches are only the instruments. And while they influence the quality of the service provided, they can’t guarantee a deep understanding of business needs – that’s up to the analyst. The tools you choose won’t necessarily be the Google tools stack, but a mix of tools that fit your organization’s flavor.
Difficulties in communication between analysts and others
Yes, every analyst works hard as hell to move their company forward. But often, no one commends them for this hard work. The main reason why that happens is that coworkers see the results of an analyst’s work like this:
That looks pretty great, but what are we supposed to do with these things?
Have you heard similar questions? This happens when people don’t perceive the value the analyst delivers.
I started doing analytics back in 2000 – yes, I’m that old… things have changed in the sense that [the] complexity of data has increased. And one of the challenges that we live under is that the complexity of data is increasing every time we’re doing analysis. And it’s actually one of the reasons why we’re not so good at communicating — because we know that we’re not really certain of this data we’re presenting.
Your coworkers aren’t mean or immune to analytics; they’re just constantly needing to make the right decisions under the pressure of a market that’s no longer a friendly place.
And here’s another side of the coin: you’ll experience the Dunning-Kruger effect or Survivorship bias even at the most data-developed companies, because that’s people’s nature. They don’t believe analytics is so hard, and they make conclusions based on the information they have at hand.
Or sometimes an analyst just puts too much focus on numbers, not the insights or what-to-do advice. But stressing data that will never be 100% perfect – combined with the uncertainty inherent in business decisions – won't make anything better.
Analysts should try their best to:
- Communicate what they know in understandable terms.
- Develop empathy for the people who will work with their numbers.
- Never fall into the data abyss without a hypothesis with which to pull themselves back out. Always remember why you perform your analyses.
- Translate the results of analysis into solutions for people. Make people understand how far data can take them. Educate your coworkers on how to apply data analysis in their work.
- Cultivate healthy self-criticism. Get your ideas to the desk and don’t be afraid to fail.
These are some of the biggest mistakes an analyst can make:
Not checking the rough copy
Without double-checking the numbers in their results, analysts can lose so much credibility. You should always check with other systems and conduct a plain reality check – can your results exist in reality? Can there actually be 89 pageviews on average within three minutes? When you doubt yourself and are strict on yourself, you’re much more than a specialist – you’re a reliable specialist.
Avoiding giving advice or answering additional questions to avoid the risk of looking stupid
Analysts can be afraid to share any additional numbers. Be bold about sharing numbers; it’s a great skill. Even if you’ve already made some mistakes, even if it was mentioned, and even if your ideas were questioned, don’t neglect your creative ideas while mechanically doing what was asked. Stay curious, creative, and fearless.
30 handpicked Google Data Studio dashboards for marketersDownload now
Existing challenges you may not have even noticed
Data we care and care not about
Analysts build reports based on the data they possess at the moment, and sometimes these reports describe obvious facts tediously. So what should you do? Are you a bad analyst for doing that?
Always remember that part of the data you possess might be wrong or corrupted, a much bigger volume of data is unavailable because you didn’t gather it, and there’s even more data that you can’t even imagine yet.
One of the analyst’s tasks is to push the limits of available data to answer questions for management and colleagues. Not trying to find something that sounds like the answer in what you already have, but cultivating the company’s data. Not trying to find support for your hypothesis while ignoring the countervailing evidence. Try to cover all you can reach, and remember that even after that, there will be more data to collect.
How can you reach more data than you have already? One approach you may use to enrich your marketing data is to consider each advertising campaign not only as a source of leads but as a source of data. You may launch a whole campaign just to get data, just as each A/B testing campaign is launched to find out what works better.
Speaking of A/B testing
A/B testing is a great tool, and it was valuable for a long time. But times have changed, and the logic of A/B testing was established when the facts on the ground were different.
Every time we decide that A is better than B or B is better than A, we are excluding some visitors who preferred the other version… So we’re actually limiting our business scope every time we do it.
From this point of view, we see that A/B testing is no longer the best tool for data activation. What’s fascinating is that now we can offer variant A to those who liked A, and offer B for those who chose another variant. This is one of the possibilities presented by the modern speed of data collection and activation. Today, you don't need to choose; you can just satisfy as many clients as possible.
So don’t get stuck in the either/or mentality; change your approach to applying data insights. Google Optimize and other personalization tools will help you with that.
Let’s see what other threats can lie in wait for an analyst in the middle of the work day.
Existing challenges you may not have even noticed
Management of self-driving marketing machines
The best way to understand this point is to model a segment partition for a typical online store.
In the slide above, people who have already bought from you are along the green line. There are also those who will never buy from you — those are on the red line. A typical advertising budget distribution won’t exclude any of these segments because:
- Marketing team will keep trying to get the red line audience interested. These people may click on the advertising and waste your budget, but they still won’t buy anything.
- Marketing will try to get back those who have already purchased from you by offering them a discount and losing profit.
This happens because it’s how marketing works, and everybody is used to it. But if you modify your automated system and focus your advertising on the middle group, investing all the resources you have to cultivate this kind of audience, you’ll get a true revenue increase even in the most difficult industries. Thus, the best advice for marketers and analysts involved in decisions about distributing the advertising budget is that you have to understand your data correctly.
This moment is a chance to go from washing data to orchestrating it. Automated marketing machines are still just tools managed by smart analysts.
Ethics of marketing and analytics
Ethics is not an aspect of an analyst’s work that’s discussed widely. But while promoting the interests of businesses, it’s important to stay ethical in your work. Remember what happened with Cambridge Analytica a few years ago? End clients really do care where you take their data and what you do with it. Your company may pay a huge price if you collect and store data without care, leave borrowed data unprotected, or use it to manipulate people without permission.
Cookies, privacy policies, and checkboxes are only the first step in maintaining the relationship between customers and websites that use personal data. Analysts may find these limitations unfair because they lead to fragmented data. And sometimes, experiments with personal data of customers may be technically interesting but questionable from an ethical point of view.
Also, there’s no safe place for children on the web. Kids use smartphones more than adults and post a lot of deeply personal data online. This is a new ethical challenge that marketers and analysts have to meet. Can we sell online to children who are under the guardianship of their parents? Starting at what age can we show advertising to children without harming their rights?
Where is the best place to hide a dead body in analytics?
The single case when you don’t need analytics is when you won’t use it. If you really don’t want to use the results of your analytics, don’t waste your money on it.
How to spend money on analytics is a whole different story, but remember that marketing analytics is an investment, and it must increase ROI in order to be worthwhile.
As for me, I’m lazy and use the easiest pack in the world (in the image below) because I’ve been working with Google tools [for the] last a lot of years. But the interesting thing is that the cloud in the middle starts to fill more and more… But for now, it’s simple to say that on the one side we have tools to gather data, on the other side we have stuff to consolidate data, and on the other, the tools where we activate data… What I see we’ll be doing in the future is actually using our skills to activate the data side.
Complete your toolbox considering the profit each tool can bring, not its popularity or how cool it is. And remember that each tool has its limits and was created to help you, not to replace you.
More advice for analysts:
- Stop comparing your conversion rate with the average. This isn’t the best basis for growth. What you should compare is your own current conversion rate to your conversion rate from the previous month. Set the benchmark against yourself, surpass yourself each time, and stable growth will not be a dream but a reality.
- Stop leaning on average metrics. Because averages hide really important information… and a body. Sometimes, people are lazy and think linearly, and we’re really glad when we see numbers that satisfy us. But when we get deeper, we confess to ourselves that the conversion event (for example, a Google AdWords campaign click) may not lead to a purchase – or at least not to a purchase of the exact product we advertised. What if people buy an iPhone case instead of a Macbook Pro? The margin is different, isn’t it? So even if a company still shows good ROAS, check this metric from the inside and find out how everything happened.
Advice for businesses:
- Stop assuming that if you hire one digital analyst, you’ve built up the whole data science department. You need to build a team with a wide range of skills: hire a statistician, a business analyst/interpreter, a tech-skilled data scientist. These specialists are your main investment in analytics, not the toolbox you buy!
- If you’re a small business, gather data about yourself, know how similar startups work and the market benchmarks, try to get broader data, build your own data set, and define the world around you.
- If you’re in a big market, look out for competitors and move from within to the outer world. There’s so much to do to overcome the competition!
For homework: Why data is not modern oil
For a long time, analysts were learning how to gather data and were obsessed with it as with goldmines or oil wells.
It’s been the ambition of analysts, saying the one who has the most data when he dies wins. But the problem is that data is not very durable. Data can’t be an oil for modern industries; it’s like any good with a certain expiry date. Data is more like meat than oil. We can keep it for a while, but if we don’t use it when it’s boiled… the data becomes more and more worthless over time.
The biggest value of data is at the moment the event is happening. By the time data is prepared for analysis, the results are delivered, and decisions are made and actions taken, it might be too late. The world has already changed, and new events have happened.
Thus, we must remember that our data exists in the context of the world around us.
To sum up
Thanks, Steen, for such a deep interview and a great talk at the Analyze! conference!
We hope you enjoyed this read and got a new vision of analytics and the analyst’s role in the modern market. At OWOX, Steen inspired us to stand for anthropocentric standards in analytics where tools matter less than specialists. We believe that the analysts themselves are the most important part of analytics — not the powerful tools they use. So let’s learn from each other to make analytics the best discipline ever!
Subscribe to our blog to read the next interviews as part of our research into the state of digital analytics and adapt the experience of top analysts to your business.