Features
Asking the question
Everything involved in building an assessment and getting it in front of a candidate. A test is only as good as the question, so this is where most of the product lives.
18
question types
12
languages an assessment can be written in
143
ready-made assessments
Every type, and what it asks a candidate to do
- Short text
- Long text
- Multiple choice
- Checkboxes
- Dropdown
- Number
- Slider
- Code editor
- Video recording
- Audio recording
- File upload
- Instruction
- Date picker
- Yes or no
- Matrix grid
- Image choice
- NPS rating
- Forced ranking
The list comes from the editor itself, so it is what you can actually add to an assessment today. Three of them are gated by plan.
Who this is for, and what changes
Three people who build assessments, and what the builder does for each of them.
The recruiter with no time to write
Starts from a template built from the work of the role, reads the questions on the page, and sends it the same afternoon.
The hiring manager who knows the job
Writes the one question that separates candidates, in the form the work takes: a recorded answer, a piece of code, a ranking.
The team hiring across markets
Sends the same assessment in the language each candidate reads, without a translator or a second build.
Every type, and what it asks a candidate to do
The list comes from the editor itself, so it is what you can actually add to an assessment today. Three of them are gated by plan.
To watch someone think
- Long textA written answer with room to reason. Read by the review and scored against what the question asked.
- Video recordingIncluded from StarterThe candidate records from a browser when they are free. Transcribed, then scored, with the recording kept.
- Audio recordingIncluded from StarterA spoken answer without a camera, for people who think out loud. Transcribed and scored like video.
- Code editorIncluded from ProCode written in the browser, run against your test cases, then read and scored for the approach as well as the result.
To check knowledge
- Multiple choiceOne right answer among several. Marked automatically against the key you set.
- CheckboxesSeveral right answers at once. Marked against the full set, with partial credit if you want it.
- DropdownOne choice from a long list, when the options would crowd the screen as buttons.
- Yes or noThe shortest question there is. A screen, a knockout, or a declaration the candidate signs off.
- Matrix gridSeveral statements rated on the same scale in one grid, for a compact self-assessment or a knowledge check.
To measure preference and scale
- Forced rankingThe candidate orders the options. Used by the Big Five inventory, because ranking is harder to game than agreeing.
- SliderA position on a range rather than a pass or a fail, for confidence, preference or estimate.
- NPS ratingA zero to ten rating with the standard promoter bands, when you want a number people already understand.
- NumberA figure the candidate types, checked against a range or an exact value.
- Image choicePick the picture. For a design eye, a product catalog, or anything words describe badly.
To collect and to guide
- Short textA word or a line: a name, a term, a value. Marked against a key or read as is.
- File uploadA portfolio piece, a spreadsheet, a deck. Size limit per plan, kept with the submission.
- Date pickerA date, for availability, a start date, or a scenario that turns on one.
- InstructionNot a question: a block of context, a scenario or a rule, placed wherever the candidate needs to read it.
Writing the screen by hand, and building it here
The same set of questions, two ways of getting them in front of candidates.
By hand
- 1Invent questions per call
- 2Email them one by one
- 3Collect answers in an inbox
- 4Retype them into a sheet
With the builder
- 1Start from a template or a job description
- 2Send one public link
- 3Answers arrive in one place
The questions are still yours. What moves is everything around them.
Question types
On every plan, including the free one
A question is answered in the form the work actually takes. A support role answers a customer by writing one, a designer talks through a decision on video, a developer writes code that runs. Forcing all three into multiple choice tells you who is good at multiple choice.
Each type is scored the way that type deserves. A ranking question compares the order against a key, a scale question reports the position rather than a pass or a fail, and an open answer is read by the model and then by you. The full list is on the pricing page, per plan, because three of the types are gated.
Building the assessment
Included from Starter
Drag a question into place, set what it is worth, and reorder as you go. Nothing is published until you say so, and an assessment can be edited after it has answers without invalidating the ones already in.
Or describe the role and let the model draft it. It produces a full assessment with questions in the types that suit the work, which you then edit: the draft is a starting point, not a decision. Generation is where an AI credit is spent, and the allowance is on the pricing page.
Video and audio answers
Included from Starter
A candidate records when they are free, which is the whole reason this exists. There is nothing to schedule, no timezone to negotiate, and a person in another country is not filtered out by a calendar before anyone has read their work.
The answer arrives with a transcript beside it, so you can read first and watch the parts that matter. Recording minutes are metered, and the rate is printed on the pricing page rather than hidden in a footnote.
Behavioral assessment
On every plan, including the free one
A Big Five inventory, using forced ranking rather than agreement scales, which is what makes it harder to answer the way you think the employer wants. It reports twelve traits separately: the five Big Five dimensions and seven future-of-work competencies.
It does not produce a hire or no-hire. A trait is a description, not a threshold, and a report that reduced a person to one number would be easier to read and worse to act on.
Public links
On every plan, including the free one
One link, and the candidate needs no account to answer. That removes the step where a third of applicants quietly give up, and it means you can put the link in a job post rather than in a sequence of emails.
A link can carry a password, a time limit and a close date. None of them is required, and an assessment with all three still needs no account from the person answering it.
URL parameters
On every plan, including the free one
Carry a source, a campaign or your own candidate id into the assessment through the link, and get it back attached to the result. That is how an answer gets matched to the row it belongs to in whatever system you already run.
Nothing about the parameters is shown to the candidate, and they are not required. A link without them behaves exactly as it did before.
Questions about building an assessment
Can I edit a template after I copy it?
Yes. A copied template is yours: change the wording, drop a question, add one of your own from any of the question types, and set what each is worth.
Can the AI write the assessment for me?
It can draft one from the job description, in the question types that suit the work. You then edit it; the draft is a starting point rather than a decision, and generation spends an AI credit.
Which question types are on the free plan?
Text, choice, ranking, scales, file upload and the rest of the written and structured types. Video and audio start on Starter, and the code editor on Pro. The pricing page lists them per plan.
Can I change an assessment after candidates have answered?
Yes. Answers already in are kept as they were given, and the change applies from then on. Who changed what is recorded.
Does the candidate see who built the assessment?
They see your company name, and your logo and colours on plans that include branding. They do not see the scoring criteria or any answer key.
Read the questions before you build anything
Every ready-made assessment shows sample questions on its own page, with no account.