4 important skills of great web analysts
The data analysts and web analysts of today are not just required to do reporting and relay data in simple charts and tables. True web analysis is about answering business critical questions and come up with intelligent answers. It’s about delivering real value based on insights derived from data. And it has a lot to do with recommending specific actions or to qualify discussions and planning in marketing and sales. I’ve thought about the most important skills of great web analysts for some time. I’m not done thinking, but this post contains some of what I’ve come up with so far.
Compared to web analysts 10 years ago, it’s now an entirely different skill set that’s required. The days where a web analyst was just the person who could navigate the reports in Google Analytics are gone. But what makes a great web analytics today? Well, a lot of different skills do - but in my mind, these are some of the general requirements:
Identify causal factors
The great web analyst must be able to identify what causes an observation or a result. Take for instance a look at the ice cream sales in an entire city over a year, and another look at the number of swimming pool accidents. Chances are that they will correlate. But it would be wrong to conclude that more sales of ice cream lead to more accidents in swimming pools. This type of correlation is called a spurious relationship in statistics. And web analysts must be able to avoid those.
For instance, it would not necessarily be correct to say that conversion rate on mobile devices is lower than on desktop simply because it’s mobile traffic and because mobile users have different intentions. What are the traffic sources? What are the landing pages? Do we have technical issues with certain device brands?
In a recent project I worked on, the client did indeed have a lower conversion rate on mobile. That’s not uncommon at all. But a deeper look at their data showed that conversion rate would vary greatly between different campaigns. It then turned out that some specific products converted great (even better than on desktop) compared to other products. So their low mobile conversion rate was not caused by the traffic just being mobile, but was caused by their campaign mix.
What I’m saying is that web analysts must be able to identify causes; do proper regression analysis if you will. Otherwise, it’s far too easy to end up making hasty conclusions, and incorrect recommendations.
No report - standard or custom - is ever enough. There’s always another secondary dimension to apply. There’s always one more drill down to do. The good web analyst will question all types of trends and patterns, and will not assume that a trend or observation is either positive or negative (and will also know when an observation actually is a trend). And the great web analyst not only asks questions related to dimensions and metrics, but uses his/her knowledge of the company and business. Google Analytics is not the place to come up with questions. Far from it. In fact, most questions must be asked from a business or human perspective.
The ability to understand and analyse quantitative observations and findings in a business context is crucial. That’s just about the only way to know what data is in fact important for the business. Notice how the title of this paragraph isn’t “Always find answers”? Every analysis has to begin with a hypothesis; a business question. Something to validate, examine further, prove or disprove. The great web analyst knows how to ask the right question before beginning to look for answers.
But while questions can be asked for an eternity, sometimes enough is enough. At some point there are just so many drill downs that all the segmenting narrows down the data to a minor and insignificant subset of data. That’s micro-analysis, and even if an answer is found, it might not provide any business value, simply because solving the problem for a single user won’t affect the bottom line.
Provide actionable recommendations
With each analysis, there’s an objective, an hypothesis to prove or disprove. There’s a set of dimension and metrics to describe and examine that objective in relation to a target or goal. There may be a timeline, a chart showing the month-on-month status of a set of metrics. Which together makes up a report that almost anyone with access to your account could do.
The actual analysis though, is not complete before there is one or more specific actions or recommendations. The great analyst knows how to translate patterns, trends, metrics etc. into something actionable. And the great analysis tells people the estimated effect of following a recommendation. You don’t want people to read your mail or pretty PDF and then just delete it. You don’t want them to think “Oh, okay. That’s interesting.” You want them to take action and yell out “Why the heck haven’t anyone done that yet?!” Right. Now.
The actions and recommendations should be do-able and measurable. There’s no point (often) in suggesting solutions, that are too difficult or too costly to act upon or to implement. That’s not the same as saying that recommendations should be easy to follow, but they have to be do-able. And the best way to understand what can be done is to work closely with the people that make the decisions. In my experience, some of the best web analysts out there are the ones who didn’t start out as analysts. They have worked with sales and/or marketing and have practical insights into the decision process. This makes them capable of relating data and recommendations to actual business recommendations.
Visualise and present data
A web analyst must be able to tell a story. In front of people. Being able to understand data, relationships between data sets, analyse those data and identify possible recommendations is great. But in order to explain how you got there, it’s just as important to be able to explain your process and findings. That’s also why it’s so important that all steps in any given analysis are documented. Then it’s much easier to update or repeat an analysis, but it will also make it easier to tell the story to non-analysts.
Now, presenting data is not about moving data from an Excel sheet to a chart and into a Powerpoint. Any chart, table or standalone metric should be instantly recognisable in terms of what it’s showing. My personal rule of thumb is that if it needs an explanation or a paragraphs, it’s not good enough. I often turn to sites such as Flowingdata or Information is Beautiful - just to get inspired. But I’d always recommend reading up on data visualisation theory. In any case, presenting data in the right manner to the right audience is a corner stone when it comes to conveying data analysis and the derived interpretations, conclusions and recommendations.
I’d love to hear about your thoughts on the role and required skills of today’s great web analysts. If you’re a web analyst yourself, work in sales or marketing or simply take an interest in digital marketing and analytics, do comment or send me a message.