Essay questions

Free text, reviewed by a person. Nothing about an essay is scored automatically.

Writing the question

The same rich-text editor as multiple choice: headings, lists, inline code, code blocks, tables, images, colour and alignment.

Essays reward specificity in the prompt. "Describe your approach to testing" produces essays you cannot compare; "You have inherited a service with no tests and a weekly release. What do you test first, and why?" produces answers you can rank against each other. The narrower the prompt, the cheaper the review.

Points and time

Points default to 20 — higher than multiple choice, reflecting that an essay usually carries more of the assessment's weight. Time limit defaults to 30 minutes.

Be generous with the time limit. A candidate racing a clock writes a worse answer without revealing anything useful about how they think.

Word limit and evaluation guidelines

Two optional fields sit alongside the basics.

Word limit is shown to the candidate as they write: their word count appears against your limit, and going over it warns them, telling them plainly that they can still submit.

It is guidance, not a gate. Nothing truncates the answer or refuses a save. That is deliberate — cutting someone off mid-sentence, or blocking a save, loses real work over a target you set as a steer. If the length genuinely matters, say so in the prompt as well.

Evaluation guidelines are notes for whoever marks the answer: what a full-mark response contains, the common wrong turn, what separates adequate from good. They are never shown to the candidate.

Write the guidelines before the first candidate sits the assessment. Doing it afterwards means writing them against the answers you have already read, which is how a marking standard quietly becomes a description of whoever answered first.

What candidates see

The prompt, a text area, and a live word count — shown against your limit if you set one. Their answer is saved automatically as they go, the same as every other question type.

They are not shown a score afterwards, and no email tells them an outcome — see Reviewing results.

Marking

Essays are the only type that needs you after submission. The attempt appears under the assessment's Results tab with the candidate's text and a field to award marks.

An attempt containing an essay is not finally scored until someone marks it. The overall score shown before that is incomplete. On an integration this matters: test.attempt.evaluated fires after marking completes, which is why it — and not test.attempt.submitted — is the event to key a downstream workflow on. See Webhooks.

Reviewing consistently

Two people marking the same essay differently is the failure mode of this question type, and it is worse than it looks: it makes candidates non-comparable without anyone noticing.

The cheapest defence is to write down what a full-mark answer contains before the first candidate sits the assessment, and to mark in batches per question rather than per candidate — reading twenty answers to the same prompt in a row calibrates you far better than reading one candidate's whole paper.

Cost

Budget the review time before you add an essay to a high-volume funnel. At 200 candidates and five minutes each, one essay question is about seventeen hours of someone's week. That is often worth it at the final stage and rarely worth it at the top of the funnel.