Term
Online test
An online test is a set of tasks delivered over the internet, with responses evaluated against predefined rules. Its outcome may be a score, feedback on errors, or a suggested next learning step.
Daria Lisovenko29 September 2026≈ 10 min.
What is an online test?
An online test is a set of tasks delivered over the internet, with responses evaluated against predefined rules. It can check understanding, explain errors, or suggest the next learning step. Participants answer in a browser and receive a result automatically or after a reviewer evaluates their work.
In brief: a test has evaluation criteria as well as questions. Those criteria connect responses to an understandable outcome.
“Online” describes access, not assessment quality. A five-question self-check about a guide and a formal examination can both open through a link, yet serve different purposes and have different requirements. A polished percentage screen does not establish that the tool reliably assesses an entire subject.
How an online test works
The participant receives tasks and selects options or enters responses. The system stores attempt data, checks it against an answer key, or sends it for evaluation using a rubric. A result is then calculated and explained through points, feedback, and possible next steps.
Evaluation does not have to be automatic. A single-choice response is straightforward to compare with a key, while an extended answer may need a teacher to read it. With mixed marking, some points can appear immediately and the final outcome later. Disclose that distinction before the test begins so a provisional score is not mistaken for a final result.
The components of a test
- Purpose. The specific understanding or action that the tasks are intended to assess.
- Tasks and responses. Prompts, options, input fields, and instructions.
- Answer key or rubric. What counts as correct, partially correct, or incorrect.
- Scoring. Points, task weights, rules for omissions, and percentage calculations.
- Feedback. An explanation of the outcome and a meaningful next step.
Tasks should assess the stated purpose. If the subject is using a guide, questions should require understanding that guide rather than guessing an unfamiliar term. One successful response does not demonstrate mastery of the whole subject. The broader the promise, the more care the task selection and evaluation criteria need.
Common task formats
| Format | Evaluation | Rule to clarify |
|---|---|---|
| Single choice | Compare the selection with a key | Whether exactly one option is correct |
| Multiple choice | Compare the complete selection | Exact match or partial credit |
| Number | Check a value or acceptable interval | Units, rounding, and tolerance |
| Short text | Compare with accepted responses | Case, spaces, and equivalent wording |
| Extended response | Apply a marking rubric | Who reviews it and when results are final |
A format is not a scoring policy. Multiple choice, for instance, does not mean every selected option should automatically earn a point: otherwise selecting everything might increase the score. Decide whether evaluation requires an exact match, allows partial credit, or applies a penalty, and test actual answer combinations.
A scoring example
Consider an illustrative self-check about a product guide. Its five tasks have maximum scores of 1, 1, 2, 2, 4. A fully correct response earns the task’s full weight; an incorrect response earns zero. This example has no partial credit and no penalties.
The participant answers tasks one, three, and five correctly. Points earned are 1 + 0 + 2 + 0 + 4 = 7. The maximum is 1 + 1 + 2 + 2 + 4 = 10. The percentage of available points is 7 / 10 × 100% = 70%.
Both percentages are correct, but they measure different things. They coincide when all tasks have equal weights; with different weights, that is not guaranteed. Instead of an unexplained “Your result: 70%”, say “You earned 7 out of 10 points”. Weights themselves should represent an agreed importance, rather than arbitrary numbers.
From a score to a result
For the example, define three illustrative feedback bands: 0–4 points means review the basics; 5–7 means review errors; 8–10 means proceed to the next module. These are chosen rules for one self-check, not universal standards of achievement. A participant with seven points receives error-focused feedback.
Checking zero and the maximum is not enough: gaps or overlaps may exist between them. Test each boundary, such as 4, 5, 7, and 8. These bands apply to integer scores. If partial credit introduces 4.5 points, the existing wording does not cover it; redefine the conditions, for example through explicit inequalities.
Different routes also need a clear maximum. If one participant sees tasks worth ten points and another sees tasks worth eight, do not silently divide both totals by ten. Define scoring for each route. Even identical percentages do not automatically make different task sets comparable.
Useful feedback after completion
A result is more useful when it explains the meaning of the number. In a learning test, it might include the total, topics to revisit, and a link to the relevant material. Practice feedback can follow each response; a controlled check may withhold explanations until the attempt is complete.
“7 out of 10 points. Review the topics in tasks 2 and 4, then try again.”
“You are bad at using the product” — a claim about overall ability based on five questions.
This outcome is not a medical finding, a psychological diagnosis, or proof of professional suitability. Important decisions require appropriate methods and additional evaluation. For an ordinary self-check, describe responses and topics rather than making sweeping judgements about the person.
Test, quiz, survey, or calculator?
A quiz is a broader question-based interaction. It may assess knowledge or recommend a service based on preferences, where no answer is right or wrong. An online test can therefore be a quiz, but not every quiz assesses knowledge.
A survey collects opinions or information rather than marking responses against a key. A registration form collects participant details. An online calculator computes a numerical quantity from parameters. A test can use a formula too, but its purpose is to evaluate tasks, not simply to calculate a price or an area.
Attempts, results, and saved data
Opening the page, beginning responses, completing tasks, and successfully saving an attempt are separate events. A number can appear in the browser before a record reaches the system. Define completion using an event that actually confirms storage of the necessary data.
A useful record includes responses, the total, the version of the tasks and key, time, and review status. If multiple attempts are allowed, decide whether to count the latest, the best, or each one separately. An omitted response should not be silently treated as identical to an incorrect response.
Contact details are not an assessment. An email address can deliver results or enable a follow-up, but should not add points. If contact details are required to see the outcome, disclose that condition before the test begins.
Measuring completion
The share of completed attempts differs from the share of correctly answered tasks. The first describes progress through the interface; the second describes responses against a key. The form conversion entry helps choose an event and denominator for the participant’s journey.
A low score may reflect difficult material, an unclear prompt, or a wrong key. Leaving midway may reflect an error, lost interest, or unsuitable conditions. Neither the number of fields nor the average score explains the cause by itself. Investigate the specific tasks and events first.
Checks before launch
- Tasks and key. Prompts are unambiguous and accepted responses agree with the material.
- Reference totals. All-correct, all-incorrect, and mixed responses produce the expected scores.
- Result boundaries. No reachable total is left without a result or assigned to two results.
- Changed responses. Old points do not remain after a different selection or a return to an earlier step.
- Devices and storage. The test works on a phone and with a keyboard, and responses and totals are actually saved.
Do not confuse an incorrect solution with an input error: one affects marking, while the other prevents submission. W3C recommends clear feedback on errors and successful submission. A red border alone is not enough; explain what to correct and whether the action was completed.
Common mistakes
- Meaningless weights. Arbitrary numbers influence the outcome more than the content.
- Unlabelled percentages. Correct-answer counts are substituted for points, or vice versa.
- Overlapping thresholds. One total triggers two outcomes while another triggers none.
- Premature final results. A provisional score appears to be a completed evaluation.
- Unverified protection claims. Shuffled questions are presented as a guarantee of unaided participation.
Online access does not establish identity or prevent outside help. Timers, attempt limits, and proctoring are separate mechanisms, not defining properties of every test. If they are required, verify their availability and actual behaviour in the chosen system.
Online tests in stepFORM
The stepFORM test builder lets you assemble questions, assign option values and formulas, configure transitions and results, and share a link or embed the form on a website. Responses are available in the dashboard, with notifications and integrations for further handling.
The author is responsible for tasks, the answer key, weights, and the meaning of the outcome. The article “How to create an online test for free” covers the practical build. Check plan availability and specialist requirements separately: the definition of an online test promises neither examination security nor automatic marking of any free-text response.
Frequently asked questions
Does an online test have to award points?
No. Its outcome can be error explanations or rubric-based feedback. Evaluation rules still need to be defined; otherwise it is simply collecting responses.
Are a test and a quiz the same thing?
The terms overlap. A quiz may check knowledge or recommend an option based on preferences. An online knowledge test evaluates tasks against an answer key or rubric.
Why can the points percentage differ from the correct-answer percentage?
Tasks can have different weights. In the example, three correct tasks out of five is 60% of tasks, while seven points out of ten is 70% of points.
Can a test evaluate text responses?
Yes, but define accepted answers or a rubric for human review. Matching a single string does not automatically cover every equivalent wording.
Does every test need a time limit?
Only if it serves the purpose. A timer does not improve the questions or guarantee the absence of outside help. An ordinary self-check may not need one.
What counts as a completed attempt?
An event confirming that tasks were completed and the required data was saved under the test’s rules. A button click or a number in the browser does not guarantee this on its own.
The key point
An online test connects tasks, responses, and predefined evaluation. Its quality depends on the content and marking rules, not just the interface. Understandable points, complete score bands, and honest feedback help participants use the result without turning it into an unjustified label.