2026-08-06

Analytics onboarding: why it should not take three weeks

Your new marketing hire started on a Monday. Keen, capable, good references.

Three weeks later they send you a message.

"Sorry, which report shows last week's traffic again?"

You are not annoyed with them. You have watched four people go through the same three weeks and two of them left before it paid off.

Here is the part that usually goes unexamined. You have accepted the training time as a property of the job. It is not. It is a property of the tool you chose and it is one of the few costs you can remove by decision rather than by effort.

What the training time actually costs

Do the arithmetic once. It is uncomfortable and it takes two minutes.

A new person needs roughly ten to fifteen hours to become independent in GA4. Not expert. Independent, meaning they can answer a routine question without asking anyone.

Those hours are not free time. They are hours you already pay for.

Then add the second person. Every question they cannot answer alone costs someone senior an interruption and an interruption costs more than the minutes it takes.

Now multiply by your turnover. Marketing roles and internships turn over faster than most. If you onboard three people a year, you are paying for that curve three times a year, forever.

The number is rarely dramatic. It is usually a few thousand euros a year, quietly, in a line item nobody tracks.

The four things a newcomer has to learn

It helps to see why the curve is that long. It is not one big obstacle. It is four small ones stacked.

The vocabulary. Users, active users, sessions, engaged sessions, events, conversions. A newcomer has to learn which one their manager means when they say visitors.

The navigation. Where the answer lives, which is not always where the question suggests.

Report building. Anything beyond the default views requires Explorations, which is a small skill of its own.

The data quirks. Thresholding, modelling, the delay before numbers settle. Until someone explains these, the newcomer assumes their own mistake when a number looks wrong.

Each one is reasonable on its own. Together they mean the tool cannot be handed over, only taught.

"Two of the most important characteristics of good design are discoverability and understanding."

Don Norman, The Design of Everyday Things

Discoverability is whether someone can work out what is possible without being told. Understanding is whether they know what it means once they see it.

A tool that scores well on both has an onboarding time close to zero, because the interface does the teaching.

Still with me? Good, because you can measure this before you commit to anything.

Test learnability before you buy

This takes ten minutes and it is the most useful evaluation step almost nobody performs.

Find someone who has never seen the tool. An intern, a colleague from another department, anyone.

Give them three tasks and say nothing else. How many people visited last week. Where did they come from. Which page was most popular.

Watch and count the questions they ask you. Do not help unless they are stuck.

Time it. Under two minutes with no questions means the tool teaches itself. Over ten minutes with questions means you are buying a training programme along with the software.

Run the same test on every tool you shortlist. The result predicts your onboarding cost better than any feature comparison.

One warning about the test. Do not run it yourself on a tool you already know. You cannot un know an interface and your own fluency will hide exactly the problem you are testing for.

What to write down anyway

Even with a tool that needs no training, keep a one page handover document. Not a manual. A page.

• What counts as a result for us. Name the two or three numbers this company actually decides on.
• What normal looks like. A typical week so a newcomer can tell a real change from noise.
• What we ignore and why. This saves more time than any instruction.
• Who to ask. One name.

That page survives turnover. A twenty page tool manual does not, because it goes out of date and nobody reads it twice.

The part managers underestimate

There is a second cost hiding behind the first one and it outlasts the onboarding.

A tool that takes three weeks to learn is a tool people open less often. They batch their questions. They wait for the monthly report instead of checking on Tuesday.

So the decisions get slower and some of them get made on memory instead of data. That cost never appears in an onboarding budget and it is larger than the training hours.

The reverse is also true. When looking takes fifteen seconds, people look. That is where the value of a simple tool actually comes from.

Where StatJolt fits in

I designed the dashboard so a new colleague can be useful on day one, without a session with you.

The screen opens with the answers already on it. Visitors compared with the previous period, ranked traffic sources, top pages, results. There is no report to build first and no vocabulary to decode.

The words on screen mean what a normal person expects them to mean. That is a deliberate choice and it is the one that removes the vocabulary step from the list of four.

You can hand a new person the link and give them the one page handover instead of three weeks of shadowing.

If you only read one box

Onboarding time is not a fact of analytics, it is a property of the tool and you pay for it again with every hire. Test it before you buy by handing the tool to someone who has never seen it and timing three basic questions. The bigger cost is not the training hours anyway, it is that a tool people find hard to open gets opened rarely and the decisions drift back to guesswork.

Back