Is a data analystabove the AI?
Below the line: work that can be generated. Above it: the objective, the authority to act, the evidence, and who owns the exception. This is the typical data analyst week, split along that line. Yours will differ. That is the point of checking.
No link, or a private profile? instead.
15%
Above the AI
Only fifteen percent of your week is spent above the line, mostly acting as the gatekeeper of truth before data reaches the executives. The rest of your time is spent on the plumbing that keeps the reports running.
Read as: someone who translates messy database tables into clean executive summaries.
[ II ] The week, split
Above the line
15% of the week
Deciding which findings are solid enough for leadership
15%This is where you save the company from acting on statistical noise and protect your own credibility.
Below the line
85% of the week
Writing and rerunning SQL queries
25%You are essentially a human API for people who do not want to learn syntax.
Reconciling conflicting numbers between systems
20%Playing detective to find out why two databases disagree on the same transaction is a classic Friday afternoon time sink.
Building and maintaining dashboards
15%You build beautiful charts that your stakeholders will look at once and then ask you to export to Excel anyway.
Fielding ad-hoc data requests
15%Answering urgent Slack messages that could have been solved by looking at the dashboard you built last week.
Drafting the monthly insights summary
10%Summarising the performance metrics into bullet points that busy managers can skim in thirty seconds.
What you own, of the four things that stay human
The objectivewhat good looks like
20The authoritythe power to act and refuse
35The evidencewhat settles a question
65The exceptionwhat lands on your desk at 6pm
30[ III ] Move the number
This number can move.
The ad-hoc data requests and routine query reruns should be automated or deflected first. Setting up a structured intake form or a self-serve semantic layer would free up a third of your week. This allows you to focus on defining the metrics that actually matter rather than just fetching them.
That is a build, not a discipline problem, and it does not need AI expertise in house: Above The AI brings the system, the person keeps the objective, the authority, the evidence and the exception.
Drafted in one pass from what you gave us. A toy, with one serious idea inside: the line is real even when the number is rough. Nothing is stored, and no name appears anywhere, unless you publish your page yourself.