You Totally Missed the First Analytics Cohort

By Jim Sterne on Sep 28, 2020 in Articles by Jim Sterne
This is a recap of the first Analytics Cohorts' private conversation about Digital Analytics

You Totally Missed the First Analytics Cohort

No Fear of Missing Out

Unless, of course, that  you were there and didn't miss it - in which case you know who you are - but I can't share who was there because the Chatham House Rule is in play.

 I can, however, tell you what you missed.

But first:

What is Analytics Cohorts?

Analytics Cohorts is a small, private conversation between consenting digital analysts. It's a safe space for talking about people, process, and sometimes technology. Careers, corporate relationships, fears, aspirations... it's a free-range exchange.

What did we discuss this time? Lots!

What Does a Digital Analyst Do?

  1. Take requirements from groups that don't really know what they're asking for ...
  2. Convert them into something that's actually doable on the technical side ...
  3. Generate something that is actually useful on the business side ...


All you have to do is:

Collect All the Data!

Understand the business question
Collect all the data (extract)
Beat it into conformity (transform)
Get it into a common location (load)
Slice and dice until it reveals something different (analyze)
Deliver new-found insights to the business (storytelling)
Then, watch in despair as they:

Do not understand
Question the data
Question your methods
Shoot the messenger
Ignore the advice

Why does this happen over and over?

Because the business side does not understand how data works. AND THAT IS OUR FAULT.

That's right - it's up to you to teach data literacy across all business units. Only then can you become the trusted advisor we all aspire to be like Adam GrecoMichele KissMary OwusuAndy Crestodina and Mai AlOwaish.





Why Don't They Get It?

From the start, businesspeople were told that the Internet was, "the most measurable medium ever." And it is! It's just that ALL the rest have always been so very bad at measurement. Example?

This Article is all about how we used to measure how many people were watching TV shows and seeing movies - and how much worse it is today.

That "most measurable" bar is a very, very low bar.

We cannot measure everything automagically and that has to be explained carefully. Just like Santa Claus, the fact that doctors can't cure everything, and politicians' promises... let's not even go there.

But for heaven sake do not try to expound upon the underlying technical architecture with all of its quirks and quillets. That will immediately generate the Blank Stare.

Not only will they not understand, they will be mad at you for making them feel dumb. Nobody likes to feel dumb.

You're better off turning the conversation toward their business problems:

 "You know, I think you would be much better at explaining your business issues to me than I could ever be at explaining how events or variables are passed to a tag manager via a data layer and how triggers can be set up based on the values of those variables to determine if said values should be passed through to other tags."

Then casually point to one of those colorful posters on your wall.


Credit Nicolas Flament in The GPlates Portal: Cloud-Based Interactive 3D Visualization of Global Geophysical and Geological Data in a Web Browser

Witness the Blank Stare working in your favor.

"If I have to have another conversation about the datalayer I'm actually going to start drinking a bottle of wine right on the Zoom call." 

Don't Tell Them Their Data is a Mess

You inherited a Gordian Knot of data, didn't you? Of course you did. We all did. Left over tags, missing tags, changes variable definitions, broken pipelines - it's a mess.

But remember that the businesspeople who rely on your analysis don't care and don't need to care about why the date is a mess. They need to respect you as the subject matter expert in answering the business question they have with the data you have available.

You need to train them out of solutioning. What's that? It's the business thought process that starts, "I want to get a handle on my advertising and compare LinkedIn with Facebook. So, maybe I can compare how many people came to our website before we ran ads and during the ad runs...." Then they send in a request for the number and source of all unique visitors for the last 17 months.

"The minute some muggle comes to you with a metric, you'll drive yourself insane trying to explain to them why the data won't answer their question, why you can't deliver the data they're asking for. What they hear is that you can't answer their question and you're working against them."

"It's like going into your doctor and saying, I need to know how many white blood cells I have. The doctor has to wonder what websites you've been reading. 'Tell me where it hurts'."

You have to teach people to treat you as a subject matter expert in the database that you have been given. Then, you are a partner and they are incentivized to realize the data is a mess and invest in fixing it.

The Tyranny of the Ad-Hoc Question

In 1975 Mihály Csíkszentmihályi defined the Flow State - being in the zone - being hyper-focused. It's when you have all the data up off the hard drive and into your random-access brain and you can see the whole problem and all the logical connections. You are in a state of intellectual creativity that is exhilarating, intoxicating, and one of the main reasons you love analytics.

And then somebody walks into your office with a "quick question" and...

The Tyranny Of The Ad Hoc Question

... your concentration flies right out the window.

  • Just a quick question.
  • Can you tell me how many people saw this and did that?
  • Can you just add another column in this report? 

No such thing as a quick question.

Artist Analogy

              Saw and sand the wood to nail together for the frame
              Stretch the canvas
              Gesso the canvas
              Mix the paints
              Prepare the brushes
              Improve the lighting

Most people think what you do is just that very last bit and how hard can it be to just answer this one question??

One Analytics Cohort member said they have banned ad hoc questions. They are not allowed to even use the word "ad hoc" and instead they perform "bespoke analyses." They consider every ad hoc question that comes in is a sign that the reports and dashboards aren't doing their job.



If you are in a small department, how do you handle juggle all of the different balls in the air? How do you prioritize?

I gave the Cohort members the homework assignment to think about a question for the next meeting. Well, Monika Mesnage proved herself to be an overachiever. Not only did she think about the question, she wrote this amazing email about it, and then she agreed to be named in public!

"Please go ahead and name me, sure! And it's totally fine to publish as you wish. I'm proud to be a part of Analytics Cohorts so seeing my name in your post will make me super happy :)

OK Monika - you're welcome!    ;-)

Here's what Monika had to say:

How do I handle juggling? I plan, prioritise (yes, she's from the UK) and track my progress. And - incredibly important - I do forgive myself if I haven't delivered on something (for reasons other than laziness).

 For me, prioritisation is easy: client work comes first. All else, second, and I do it in order of my own liking of the task. Planning takes time and a lot of energy to stick to if you have conflicting priorities. You will also always de-prioritise your own development, which should not be the case. Learning to plan & stick to it is something I am still working on. Below are my tactics I have developed over the years.


Short-term (1-2 days) planning if I have an insane amount of tasks: Eisenhower Decision Matrix (figure 3, I have not read the full article). I cannot stick to using this all the time but that also works for some people. Only works for prioritising conflicting tasks.

For week-to-week: post-it notes. Now I also have Jira, but still rely on post-its. Generally, I have a look at my tickets (or post-its) + calendar. If I have a busy week, I don't plan anything extra, just focus on delivery. If I have fewer tasks, I plan to deliver some additional projects I want to work on. Post-its have a great quality: if you have too many, you gather them and put all non-essential stuff in the bin (trash can). At some point you'll learn to be merciless. 


For big tasks (measurement framework design, I am looking at you!) that I need to do and have been putting off for too long: Pomodoro. I break the task into smaller chunks and only plan in one piece. This worked like magic when I started thinking about big tasks in 20-minute pieces. (Yes it did take 2-3 hours to split my monolith task to smaller chunks the first time but it was SO worth it!.)

For longer term objectives, I follow more or less this process:

1. I think about what I want to deliver in the next year (this is just in my mind as an objective)
2. Plan tasks which I can realistically deliver in the next 3 months
3. Write those down and plan them in (sometimes I book out time in the calendar) 

For example, for the next 3 months my plan is to: 

Write 3 technical blogs
Finalise GA audit process (GA settings + GA data quality)
Start developing a framework for 'best practice' datalayer + GA use
Finish automating an internal business process
Create videos for basic GTM implementation  

In my opinion this is already quite a lot on top of all the meetings and client deliverables, but I think I can do it. I will also not kick myself too hard if I don't - it is a lot, and dependent on the work which comes in.

The important part to also consider is the value to the business. When I was in-house, my main rule was (sadly - I was young back then!)  'who shouts louder'. Only then turn to what is important for the business but often ignored. Anything with an immediate ROI was first, and of course anything impacting the bottom line (customers can't check out!) was a priority. 

I still believe that planning is the best resolution to this: planning will mean that if something needs to be urgently slotted in, at least you know what you are delaying delivery of something else. If the task you are delaying will provide more value, it's an easy way to push a stakeholder back.

Brilliant. And so good, having just read Avinash's recent post on Ruthless prioritization

What's Fun About Your Job?

  • Trying to find insights. That Ah-HAA! moment is really exciting.
  • It's the Sherlock Holmes thing.
  • It's never it's never routine. It's never boring. There's always something new. I like that challenge of it.
  • It's very intellectually stimulating.
  • There's a real tangible value to it.
  • You can really cause change and make people's lives better.
  • I really enjoy learning to do new things.
  • Solving a problem or getting to the point where technical you built is working - it feels good.
  • When I feel like my work has an impact.
  • I love that I have some knowledge that can help clients achieve some of their goals.
  • Solving puzzles.
  • I think digital analytics isn't hard, but it's intricate. It's a 40,000 piece jigsaw puzzle, and it's entirely white. And three dimensional.

I Love My Job shirt on Amazon

Analytics Centeralization

Inside the company, everybody is their own business. Everybody needs their own data points. The idea of centralizing all this makes sense, but there's a conflict between having control of their own data and being subjected to some sort of imposed structure that not everybody wants to follow.

Everybody wants to do their own thing with their site or their region or their locale or their agencies and trying to get everybody the same something-that's-useful just doesn't make sense.

"My approach was to introduce the concept of Reconcile. We have 25 analytics teams in the organization not counting the data scientists. And I said, Okay, we will reconcile our numbers with your numbers. We will explain why Adobe differs and it's really easy to do with digital because you just pull up the website and bring out a debugger. We can say, this is how we count this. How do you count it? Then we can understand why it's different. And we put the two numbers side by side. We agree on a goal and how we are going to get to the goal, but the goal ain't changing. That's not negotiable. But now, getting there is collaborative. We have an overarching view of the business and your individual/departmental world has its own view and that's totally valid. But without an overarching view, it's chaos and we can't celebrate your success to the C-Suite because we've no common way of talking about it.


How Do I Get Them to Care?

"It seems like the leadership in many organizations isn't invested in web analytics. I have only seen a few businesses that have clear digital strategy that are testing around proper hypothesis and actually have KPIs or even one web KPI. I came into one organization as a consultant, and I see that there are no KPIs. The leadership just doesn't care about them. How can I help such an organization? I am not coming in at that C-Suite level. And if they don't care, they wouldn't want to really listen to me anyway. How can I help a business understand the importance of KPIs?"

"Well, first of all, welcome to digital analytics! It's this is it. This is the gig. I think sooner or later, you're going to see more and more CEOs and officers and people who understand that data is a assets that can drive revenue. It's coming. Hang on in there. I tend to approach these things in two ways. First, you can try and find an advocate find someone who is competent to ask questions, and make them shine. Then everybody wants to be them.

 "And now you've incentivized people to be like, I want to be like you. You got C-Suite time. You got more money, more staff, whatever. The other thing I do - because I'm completely fearless - is literally just go to the C-Suite and go, These are the four things that you need to know and here's where they are, and here's some insights. And you say to yourself 'Thank God, you're engaging with me!' Say, All right, what isn't it you didn't like? I spent 30 hours on this it felt valuable to me. I've obviously misunderstood what your goals are. Articulate your goals to me. From there, you can get to a place where you're getting them to ask you good business questions. And that will get you there - slow and steady." 

Tell Me More

"Play dumb in their area of expertise and they will become avuncular. Tell them, I saw this weird thing and I think it means this. What do you think? I'm just looking at the data; what does it suggest to you? That gives them the chance to become the mentor and explain the business to you. Oh thank you, you say, that's interesting! Tell me more...."

Wherein Jim's Mind is Blown

An (almost) verbatim quote:

"We're like a little agency inside the company. One product team only has three metrics. But if you're the sales team for that product, you might have six other metrics. If you expand it out towards all the potential products, every page, or when you think of corporate social media, the email campaign; when you think of the model builders, the AI team... you can see how actually 900 is quite limiting. We could have gone completely bonkers."


The Job Interview

"I'm going to an interview tomorrow for a senior marketing analyst role. I just want to ask: What would you ask in the interview? Any tips you can share?

  • You can't uncover the problem solving and procedural mindset that an analyst needs to have by just asking questions.
  • When they ask you those weird interview questions or some random thing, they just want to see how you attack that problem. There is no black and white, mathematical endpoint, answer. They want to see how you think.
  • I am hiring right now, and I really hire for intellectual curiosity. And I am always impressed by people who say, I don't know the answer to that question, but here are the resources I would draw on.
  • I would want them to ask questions around data governance and what I think their biggest challenge will be as a new person in the first 30 days. I'd hire that person in a second just by the fact that they're here and engaged.
  • Ask them more questions than they ask you. Come across as being interested and enthusiastic. Curiosity and enthusiasm cannot be taught. I can teach you any technology or process, but I cannot teach you to be enthusiastic.
  • If somebody is so annoying that they ask you, "What is your biggest weakness?" Reach into your pocket, and hand them a card that says, "Sometimes I over prepare".

What is Analytics Cohorts?

Analytics Cohorts is a small, private conversation between consenting digital analysts. It's a safe space for talking about people, process, and sometimes technology. Careers, corporate relationships, fears, aspirations... it's a free-range exchange.

Apply today

Sterne Measures Newsletter

Sign up and stay up to date on Jim's papers, articles, podcasts, and in-person appearances.

You'll receive a confirmation email because double-opt-in is in my blood.

Videos by Jim Sterne