Free test
English

Blog: German for the exam — writing and speaking

Mock test or real exam: why your score drops

15 November 2026 · 5 min · Read

In the model test you had 80 out of 100 points, in the exam it was 55. That is not an outlier and not bad luck — it is two different measurements. Here is why your score at home is almost always higher, which four conditions make the difference, and how to practise so that your practice score tells you something about the exam.

Short answer: A model test measures whether you know the tasks. An exam measures whether you can solve them under unfamiliar conditions: an unknown text, real time, no breaks, no help, one single chance. Anyone who practises at home with familiar test sets and without a clock typically loses part of their points in the real exam — not because of missing knowledge, but because of the conditions. So train with unfamiliar test sets, real time and no interruptions.

The four conditions that make the difference

  1. Unknown text. Anyone who does the same model test three times recognises the questions and answers from memory instead of from the text. That is no longer reading comprehension.
  2. Real time. At home nobody stops the clock. In the exam, how you divide the time decides whether you even reach the last part — in reading and in writing this is the most common loss of points.
  3. No interruption, no help. No dictionary, no phone, no looking things up, no second look the next day. What you do not have ready in the exam does not count.
  4. One single chance. Tension costs points: anyone who shakes reads more slowly and writes less. You can only train this effect by creating it.

What a model test measures wrongly

At homeIn the exam
familiar test set, partly from memorya text you have never seen
time by feel, breaks possiblefixed time per part, no looking things up
on the laptop, with correction helpby hand, without help
solutions available for comparisonno feedback, only after weeks
you can stop if it goes badlysee it through, even if it goes badly

The point is not that model tests are worthless — they show you the format and the task types. They are just not an exam simulation. Anyone who confuses the two will plan too tightly.

How to test realistically

  1. Take a test set you do not know. Not the one whose solutions you have already read. We have eleven exams with several test sets and over 1.100 audio files — the choice is big enough to get something new every time.
  2. Create the conditions, not only the time. Phone away, empty table, a chair, no drink next to the answer sheet, pen instead of keyboard when writing. Twenty minutes without interruption.
  3. See it through, even if it goes badly. That is exactly the exam situation: you cannot start again, and you still have to keep working.
  4. Calculate with the official scale. Not by feel. Every exam has a points distribution and a pass mark — they are in the examination regulations and in the results of your run.
  5. Only then evaluate. And separately: how many points came from tasks you really understood, and how many from guessing?

What your score says — and what it does not

A score from a cold run is a tendency, not a forecast. Experience shows: anyone who is clearly above the pass mark in a cold run will pass with high probability. Anyone just above it may pass. Anyone below it does not need more practice at the same pace, but targeted work on the part that eats the points.

That is why the more interesting question is not “how many points do I have”, but “in which part do I lose them”. A drop in points in writing has different causes from one in listening — and in listening it is almost always the time or hanging on to a missed passage. How to survive gaps in the listening text is in Listening: points.

The three mistakes that artificially lower your score

  1. Time divided wrongly. Anyone who gets stuck on the first text in reading loses the easy points at the end. Look at everything first, then solve — and set a fixed upper limit per part.
  2. Task not read. “Write an email to your teacher” is a different task from “to your friend”. Text type and addressee are points, not decoration.
  3. Writing without a plan. Two minutes of planning save more time in writing than they cost: four bullet points, then write. How to check it is in Writing without correction.

Frequently asked questions

Why do I have more points in the model test than in the exam?

Because at home you work under different conditions: familiar text, no real time limit, breaks possible, help within reach. In the exam all of that is gone — and with it part of the points, without your German having become worse.

How many model tests should I do?

As many as it takes until the score stays stable — with unfamiliar material every time. Three runs with new test sets say more than ten runs with the same one.

Is the score from the sample exam here comparable?

It follows the criteria and the points distribution of the respective exam, but it is a practice assessment. How exact it is and where its limits lie is in AI assessment: limits.

I fail listening, but I am good at reading. What does that mean?

That you have a partial problem, not a level problem. Listening usually fails because of the time or because you hang on to one spot — not because of vocabulary. Practise with short sections and fixed pauses instead of long runs.

When should I register?

When you are above the pass mark in two cold runs in a row. Plan several weeks of lead time for the registration itself, see Getting an exam date.

Does my result from the model test count anywhere?

No. For residence, recognition or a job application only the certificate of a recognised test centre counts.

Status: 15.11.2026. Basis: examination regulations and implementation notes of the providers Goethe-Institut and telc (time limits per exam part, no aids, single administration) as well as our own implementation of the sample exam (unfamiliar test sets, time pressure, assessment against criteria). The statements about lost points describe the effect of the exam conditions; they are not a statistical survey.

Start now

80 points at home, 55 in the test: four conditions explain the gap. How to practise with unseen texts so the number means something.

Start a practice test B1

Article language: Deutsch · English · Türkçe · Українська · العربية · فارسی · Русский · Español · Français

Read next

Back to the blog