Why your practice test score is lying to you

Scoring 92% on a question bank you have seen four times measures your memory of that bank. It does not measure whether you can configure a network.

There is a specific and very common way to fail the CCNA: grind a question bank until you are scoring in the nineties, walk in confident, and meet a lab question that does not care what you have memorised.

Three reasons the number is inflated

1. You are being tested on the same questions

After two passes through a bank you are not recognising the concept, you are recognising the question — its phrasing, the shape of the options, which one looked odd last time. That is real memory, but it is memory of the bank, and the bank is not the exam.

2. Multiple choice hides what you cannot produce

Recognising the correct command among four options is a much easier task than typing it at a blank prompt. The gap between "I knew it when I saw it" and "I could produce it" is exactly the gap a lab question is built to expose.

3. Small samples swing wildly

Twenty questions is not enough evidence to justify a percentage with a decimal point in it. Three lucky guesses move a twenty-question score by fifteen points, and nothing about your actual knowledge changed.

What an honest score has to do

• Shrink toward the mean when evidence is thin. With ten answers, the right output is "we do not know yet", not a confident number.
• Weight by exam domain. Cisco publishes the domain weightings. Being excellent at the 10% domain and weak at the 25% one is not an average.
• Decay. What you knew two months ago and have not touched since is not what you know now.
• Count production, not just recognition. A configuration you built that actually forwarded traffic is stronger evidence than a multiple-choice hit.
• Track calibration. How often were you certain and wrong? That number predicts exam failure better than raw accuracy does.

Calibration is the one nobody measures

Being wrong is survivable. Being confidently wrong is what fails people, because you never go back and check it. If you record how sure you felt alongside each answer, you get a second, more useful score: the gap between your confidence and your accuracy.

Candidates with a large gap tend to fail even at high practice scores, because their study time flows to topics they already know — they never revisit the ones they are wrong about, since they do not believe they are wrong about them.

How to test yourself honestly

1. Configure from a blank prompt. No options to choose from. If you cannot produce it, you do not know it.
2. Fix broken networks, do not just build working ones. The exam gives you symptoms and asks for a cause.
3. Rate your confidence before you see the answer. Sure / think so / guessing. Then look at how often "sure" was wrong.
4. Use fresh questions for measurement. Keep a bank you have never seen for the two weeks before the exam. A score on seen questions is a study tool, not a measurement.
5. Re-test the things you got right weeks ago. If it has decayed, better to find out now.

The uncomfortable version

A readiness score is only useful if it is willing to tell you no. Any number that always agrees with you is marketing, not measurement — and the cost of believing it is an exam fee and another month.

ITWars puts readiness on Cisco's real 300–1000 scale, weights it by domain, shrinks it toward the mean while evidence is thin, and tracks calibration separately. It will tell you when you are not ready. That is the entire point of it.