Data-Driven Instruction Without Drowning in Spreadsheets

Most classroom data gets collected, reported, and never used. Here is how to track less and actually change your teaching because of it.

Published 2026-08-10 · 5 min read

Most schools collect far more data than they use. Teachers enter scores into systems, systems produce reports, reports go into folders, and teaching continues exactly as it would have anyway.

The problem is not the amount of data. It is that most of it was never collected to answer a question.

The test for any piece of data

Before recording anything, ask: what would I do differently depending on the answer?

If you cannot name a specific action that changes based on the result, the data is not informing instruction. It may still be required — reporting obligations are real — but be honest that it is compliance rather than teaching.

That question eliminates a surprising amount. It also sharpens what remains.

Three questions worth answering

Nearly all genuinely useful classroom data answers one of these.

Who has not understood this yet? So you know who needs it again before moving on.

What did most of the class miss? So you know whether to reteach to everyone or intervene with a few.

Is this pupil progressing? So you know whether what you are doing is working.

Anything that does not answer one of those three is probably serving a reporting requirement rather than your teaching.

Collect small, collect often

The most useful data is not the termly assessment. It is the two-minute check at the end of a lesson.

Exit tickets. One question on the thing you just taught. Not a test — a check. Sort into three piles as pupils leave: got it, nearly, not yet. Thirty seconds of sorting tells you how to start tomorrow.

Mid-lesson checks. Three questions on whiteboards. Look up, scan, count. You now know whether to continue or stop.

Targeted marking. Rather than marking everything, mark one specific thing across the whole class. Faster, and produces comparable data.

These produce information you act on tomorrow. A termly assessment produces information you act on next term, by which time the moment has passed.

What to actually record

Most quick checks should not be recorded at all. You act on them immediately and that is their purpose.

Record when you need to see a trend over time.

Recording worth doing:

Recording usually not worth doing:

The test again: will I look at this alongside another number later? If not, act on it and let it go.

The structure that stays usable

One sheet per class. Pupils down the side, assessments across the top. Add a column each time you assess.

One row per pupil, always. The moment you have multiple rows per pupil the sheet becomes unusable.

Dates in the column headers. Not "Assessment 1" — the actual date. In six months you will need to know when.

Then three calculated columns that make it a working tool rather than a record:

Class average per assessment, so you can see which topics landed badly across the board.

Individual average, so you can see who is consistently struggling as opposed to having one bad week.

Change since last time, so you can see direction rather than just position.

That third one is the most useful and the most often missing. A pupil at sixty percent who was at forty is a very different situation from a pupil at sixty percent who was at eighty.

Reading it properly

Look at the columns first. A whole column scoring low is a teaching problem, not a learning problem. Reteach it differently to everyone.

Then look at the rows. A pupil low across every column needs something different from one who dipped on two topics.

Then look at change. Steady improvement means the current approach is working — leave it alone. Flat means it is not.

Then look at the outliers. The pupil who scored unusually high or low on one thing. There is usually a reason and it is usually worth knowing.

Ten minutes after each assessment. That is the entire analysis, and it is more than most data gets.

Conditional formatting, used sparingly

Two rules only.

Below your threshold in amber. Whatever counts as concerning in your context.

Declining in red. Compare to the previous column. Decline matters more than a low absolute score, because it means something changed.

Resist more. A sheet where everything is coloured tells you nothing, and the temptation to add rules is strong.

What changes because of the data

This is the part that gets skipped, and without it none of the above matters.

After each analysis, write down one thing you will do differently. One. Then do it.

A data system that does not produce actions is a filing system. The action is the point.

When the data disagrees with you

It will, and this is where honesty pays.

You are confident a pupil is progressing. The data shows flat. One of two things is true: either the measure is not capturing what is improving, or your impression is wrong.

Both are worth knowing. Sometimes the measure is genuinely poor — confidence and willingness to attempt work improve before scores do. Sometimes the data is right and you have been generous in your reading of a child you like.

Sitting with that disagreement rather than dismissing it is what separates using data from collecting it.

What to stop

Tracking anything you have never looked at twice. Check your sheets. Columns nobody has referenced since entry can go.

Entering the same data in two places. Common, and always a sign of a process nobody has reviewed.

Assessing more often than you can act. Weekly assessment with monthly response means three quarters of it is wasted effort.

Collecting data to prove you are collecting data. If a system exists to demonstrate diligence rather than to inform teaching, say so plainly and spend the minimum on it.

The aim is less data, looked at properly, and acted on. That is harder than collecting more, and it is the only version that changes anything.