Maintenance
Part of Five kinds of analytics careers, and what actually separates them
Analytics careers framework as practitioners see it
Analytics careers planned on four axes: technical depth, domain, communication and ownership, with a rule for which one to work on and how to evidence it.
Career planning in analytics usually happens as a list of things to learn. That produces motion without direction, because the list has no way of telling you what is missing.
This is a different shape: four axes you can place yourself on, a rule for which one to work on next, and a way of testing whether a planned move is actually available to you.
What to take away
- Progress happens on four separate axes, and most people advance on the one they already enjoy.
- The axis to work on is the one that is currently blocking the work you want, which is rarely the technical one.
- A move you cannot demonstrate is a move you cannot make. Evidence is the constraint, not ability.
The four axes
Technical depth. What you can build and how reliably. Ranges from writing queries against a modeled layer, through building the modeled layer, to designing systems other people build on.
Four axes and their ranges
Technical depth
- Low end
- Queries modeled layer
- Middle
- Builds modeled layer
- High end
- Designs systems others build
Domain understanding
- Low end
- Knows field names
- Middle
- Knows the process
- High end
- Predicts the number
Communication
- Low end
- Answers the question
- Middle
- Reframes the question
- High end
- Consulted before decisions
Ownership
- Low end
- Owns tasks
- Middle
- Owns a system
- High end
- Owns outcomes and people
Domain understanding. How well you know the business the data describes. Ranges from knowing what the fields are called, through knowing what the process does, to being able to predict what a number will show before you run it.
Communication and influence. Whether your work changes what people do. Ranges from answering the question asked, through reframing the question, to being consulted before decisions are made.
Ownership. What you are accountable for. Ranges from tasks, through a system or a domain, to the outcomes of a function and the development of other people.
Placing yourself takes ten minutes and is more useful done honestly than done generously.
A grid worth filling in
| Level | Technical depth | Domain understanding | Communication | Ownership |
|---|---|---|---|---|
| One | Answers a defined question with existing data | Knows the field names | Delivers what was asked | Own tasks |
| Two | Builds and tests reusable models | Knows the process behind the data | Reframes a vague request into an answerable one | A recurring product or dataset |
| Three | Designs what others build on | Predicts what a number will show | Consulted before the decision | A domain, including its retirements |
| Four | Sets technical direction across teams | Trusted on business judgment beyond data | Changes what the organization measures | Outcomes and people |
Nobody is at the same level across all four, and the profile matters more than any single score. A three, one, three, two profile describes a very different person from a one, three, two, three, and each is short of different things.
Level grid across four axes
Technical depth
- Level One
- Defined question, existing data
- Level Two
- Builds reusable models
- Level Three
- Designs what others build
- Level Four
- Sets technical direction
Domain understanding
- Level One
- Knows field names
- Level Two
- Knows the process
- Level Three
- Predicts the number
- Level Four
- Trusted business judgment
Communication
- Level One
- Delivers what was asked
- Level Two
- Reframes vague requests
- Level Three
- Consulted before decision
- Level Four
- Changes what is measured
Ownership
- Level One
- Owns tasks
- Level Two
- Owns a product
- Level Three
- Owns a domain
- Level Four
- Owns outcomes and people
Choosing what to work on
The rule: identify the work you want to be doing in two years, then ask which axis currently prevents it. Work on that one.
Which axis is blocking you
What work do you want in two years?
Name the axis that prevents it
Technical axis rarely binds past level two
This usually surprises people. The technical axis is the one most under a person's control, so it is the one that gets attention, and it is rarely the binding constraint past level two. What blocks most people is the domain axis or the communication axis, and both improve through exposure rather than through study.
Two practical mechanisms: for the domain axis, sit in weekly meetings where decisions get made, for a quarter, and take notes on arguments.
Communication axis: write a one-page summary for every piece of work, send it, then ask one person what was unclear. Ask analytics foundations questions of the people who own the process to speed domain understanding.
Making a move demonstrable
Ability that cannot be shown does not count in a hiring conversation or in an internal reallocation. Two properties make analytical work demonstrable.
Two properties of demonstrable work
Can be rerun
- What it means
- Others get your outputs
- How to do it
- Reproducibility
- What it proves
- Work can be inspected
- Why it counts
- Not taken on belief
Has a history
- What it means
- Change record is visible
- How to do it
- Version control
- What it proves
- How you decide and work
- Why it counts
- Interviewers can least see this
It can be rerun. Somebody else can take your inputs and get your outputs. That is reproducibility, and it separates work that can be inspected from work that has to be believed.
It has a history. The change record shows how it developed and what you decided. Keeping work under version control is worth doing for its own sake and doubles as evidence of how you work, which is the thing an interviewer most wants and can least see.
Both are economic signals, costly for a weak candidate to fake and cheap for a strong one to produce.
That is why they carry information a resume claim cannot.
A planning pass that takes an afternoon
- Place yourself on the four axes, honestly, with an example for each placement.
- Name the work you want in two years, in terms of what your day contains rather than a title.
- Identify the axis blocking it and write one sentence on why.
- Choose one mechanism for that axis and commit to it for a quarter.
- Name the evidence you will have at the end that you did not have at the start.
- Set a date to repeat this.
Step five is the one people skip and the one that makes the rest count.
Related reading on this site
The five kinds of work you might be planning toward are in analytics careers, and the paths people actually take are illustrated in five career composites. For the moves that quietly close options, see nine career mistakes. The technical axis is grounded in analytics foundations, and the credential question that arrives at most transitions is covered in analytics certifications.
Common questions
What if I am blocked on an axis my employer gives me no way to develop?
That is a finding about the employer rather than about you. Domain and communication can often be developed sideways, through another function; ownership usually cannot, and a role with no route to it is a role with a ceiling.
Should I aim for level four everywhere?
No. Very few roles need it and the trade-offs are real. A profile with two strong axes and two adequate ones describes most senior people accurately.
How do I place myself honestly?
Require an example for each placement. If you cannot name a specific piece of work that demonstrates the level, you are one level lower, and knowing that is the point of the exercise.
Does this work for someone entering the field?
Yes, with the caveat that entry is mostly decided on the technical and communication axes, because there is no domain history to show. The grid is still useful for choosing which first role will develop the axes you are weakest on.
