What an analyst does all day

People who want to get into data ask me what the job is like, and I used to answer with the job description: queries, models, charts, insight. Then I kept a time diary for a week. The honest answer is plumbing and persuasion, with some analysis in between.

What this is

One working week, half-hour blocks, seven categories, written down as it happened. Forty hours in total. I averaged the five days into the single ring, and the week view shows the days as they were.

That is the whole dataset. It is one person in one role in one week, and it is here because a real week beats a vague impression. It is not a survey and I have not seen a good one, which is part of why I want yours.

The shape of the week

Finding the data and cleaning it took 15 hours. Not one of those hours felt like analysis while I was in it. It was reading someone else’s schema, asking which of two revenue tables is the real one, and working out why the finance export and the warehouse disagree by two percent.

Meetings and explaining took 9.5 hours. Some of that was standups I could skip. Most of it was the useful kind: walking a stakeholder through a number until they trust it enough to act on it. A chart nobody believes is decoration.

Writing SQL and models took 6 hours. Charts and dashboards took 4.5. Together that is the 26%, and it is the part I would have described as “the job” a few years ago.

Waiting took 3.5 hours: pipelines that ran slowly, reviews that sat in a queue. Learning took 1.5. I would like that last number to be bigger and I suspect that is common.

Why the imagined ring is wrong

The imagined ring is not a strawman. It is what the course syllabus looks like, and what the portfolio project looks like. A course hands you a clean CSV so you can practise the analysis. A real company hands you eleven tables, two of which are deprecated and one of which is right.

The gap between the rings is not a failure of the job. It is what the job is. Data work sits between systems that were never designed to agree and people who need one answer by Thursday. Closing that gap is the value, and the analysis is the small precise part at the end.

What this means if you are learning

My opinion: most people learning the job are practising the 26% and neglecting the rest. That is understandable, because the 26% is what you can practise alone with a laptop. The rest needs a messy company around you, but you can get closer than you think.

The persuasion half is harder to practise, but not impossible. Explain a number to someone who does not care about data and keep going until they can repeat it back. Write the two sentences that would go in the Monday email. Present a chart and ask what decision it changes. If the answer is none, the chart was not finished.

No two days were the same

The daily rings are the part I find most useful. The analysis share went from 19% on Monday to 38% on Thursday, in the same week, with the same role. A single average would have flattened that. If your week has a shape, learn it: protect the day that tends to be deep work and accept the day that is meetings.

Send me your week

I want to publish more rings than mine. If you keep the same diary for one week, in the same seven categories, send it over and I will add it to a follow-up with the rings side by side. Any role that touches data counts. Anonymous is fine.

What to do

• Keep the diary for one week. Half-hour blocks, seven categories, no editing after the fact.
• Draw the ring. If your analysis share is under a third, you are not doing the job wrong. You are doing the job.
• Pick one plumbing skill and one persuasion skill to practise this month. Reconciling two sources, or explaining one number until it sticks.
• Decide what you want the ring to look like in a year, then change the calendar, not the description.