A TestGorilla invitation is not one test. It is a bundle of up to five tests from the platform's library, plus however many custom questions the employer wrote, and every one of those pieces carries its own separate clock. Inside a timed test you move forward only, and a question you skip is scored as unanswered rather than saved for the end. Budget 30 to 60 minutes, plan for a single attempt, and read the instructions screen before each test instead of assuming the first one set the rules for all of them.
What is actually inside the link
The word "assessment" hides the structure. A TestGorilla assessment is assembled from two different things, and only one of them is capped.
| Layer | What it is | How it is timed |
|---|---|---|
| Library tests, up to 5 | Ready-made tests the employer picked: cognitive ability, language, software skills, programming, personality and culture, role knowledge | Each test has its own limit, commonly around ten minutes. Many also run a countdown on every individual question |
| Custom multiple choice | Questions the employer typed themselves | Employer sets a limit, or leaves it untimed |
| Custom essay | A free text answer, usually about their company or the role | Employer set, often more generous than the library tests |
| Custom video | A one-way recorded answer, closer to a short HireVue clip than to a test question | Think time, then a recording limit. Retakes are an employer setting and are frequently zero |
| Custom file upload | You attach a document, a portfolio piece or a worked sample | Loose, and not the part that eats your hour |
Five is a hard ceiling on library tests. It is not a ceiling on the assessment, because custom questions sit outside that count, which is exactly how a "five test assessment" still consumes an hour. The invitation email and the overview screen you land on list every part with an estimate beside it. That overview is the only place you see the whole shape at once, so read it properly before clicking into test one.
The four mechanics that catch people out
Five clocks, not one
Each test is timed on its own. Finishing the numerical reasoning test four minutes early does not hand those four minutes to whatever comes next, it deletes them. You pace inside a test and never across the bundle, and a slow start on test one cannot be repaid on test three.
Forward only, with no review screen
Within a timed test the previous question closes behind you. No flagging, no returning, no summary page listing what you left blank, no last sweep with the spare minutes. The school exam habit of doing the easy ones first and looping back does not exist here.
A skip is permanent
Skip does not mean park, it means discard. The item is scored as unanswered and nothing gathers skipped questions for later. Where a test does not publish a penalty for wrong answers, a guess after eliminating one option beats a blank every time. If a specific test's instructions state otherwise, believe that screen over this page.
One attempt, no pause button
You get one run at the invitation. Closing the tab is not a break and a lost connection is not a reset. The moment you press Start is the moment you are unavailable, so the phone goes face down and the water is already poured.
Put those four together and the assessment stops resembling an exam and starts resembling a sequence of short sprints. The strategy that survives contact is unglamorous. Answer every question. Leave nothing blank. Accept a rough answer on a hard item rather than burning ninety seconds the next four questions needed.
A worked bundle, minute by minute
Here is a plausible hour for a support or operations role: three library tests, one personality test, and two custom questions. Contents differ per employer and the numbers below are typical rather than promised, but the shape is what you are pacing against.
- 0:00 to 0:04, setup. Browser and connection check, camera permission if the employer switched on webcam snapshots, and a prompt to enter full screen. Leaving full screen can be logged as an event on your report, so sort out water, door and phone now rather than at minute twenty.
- 0:04 to 0:06, the overview. Every test listed with its estimate. Note which ones warn about a per-question timer, because those are the ones where hesitation is expensive.
- 0:06 to 0:16, test one, cognitive. Something like attention to detail, numerical reasoning or problem solving. Around ten minutes, frequently a countdown on each question that advances by itself when it expires. Items tend to get harder as you go.
- 0:16 to 0:26, test two, language or role knowledge. Roughly ten minutes again, the same forward-only movement, usually a gentler per-question pace.
- 0:26 to 0:36, test three, software or programming. If a code editor appears, expect your submission to be run against test cases rather than read by a person. Handle the obvious edge cases before you polish anything.
- 0:36 to 0:46, personality or culture test. Usually untimed or generously timed, no right answers, and longer than you expect. This is where people who rushed the first three suddenly relax and start overthinking every item.
- 0:46 to 0:52, custom question one, an essay. Written by the employer. Answer it about their company and their problem, not in the abstract.
- 0:52 to 1:00, custom question two, a recorded video answer. Think time, then a recording limit, and retakes only if the employer enabled them. Assume they did not.
Notice where the real risk sits. It is not in the hard cognitive items, which are hard for everyone and scored against other candidates rather than against perfection. It is in minutes 46 to 60, when your attention is spent and the only two parts a human being will genuinely read are still ahead of you. If you have any say over your start time, do not begin this at eleven at night.
The parts that are not scored the way you assume
- Cognitive tests are ranked, not graded. Scores are reported alongside how other candidates did on the same test. A raw result that felt mediocre can still place well, so do not judge your own performance by how hard the questions felt.
- Personality and culture tests have no score to beat. They are matched against a profile the employer defined. Answering as the person you imagine they want produces inconsistent responses across items that ask the same thing in different words, and inconsistency is one of the few things these instruments genuinely detect.
- Situational judgement items are format traps. Some ask for the single best response, some ask for the best and the worst, some ask you to rank. Misreading which one costs you the whole item, and the per-question clock is running while you work it out.
- Essays and video answers are the human-read parts. They also carry the least time pressure and the most upside. Spend your remaining energy there rather than on a cognitive question you already left behind.
What the employer sees afterwards
A results page per candidate: a score for each test, a comparison against other people who sat the same tests, your custom answers in full, and whatever integrity signals the employer turned on. Those commonly include webcam snapshots taken at intervals, whether you left full screen, your IP address, and how long you spent on each question. Not every employer enables them, and what the platform offers changes over time, so treat this as the general picture rather than a specification. The practical version is simpler. If the setup screen asked for your camera, assume snapshots, and behave as though the session is observed from the first click to the last.
Where OfferPilot AI is the wrong tool
Worth saying plainly, because people often reach this page from a neighbouring question. A timed skills assessment is a poor fit for a live copilot, and the reasons are practical before they are anything else.
- There is no interviewer to listen to. Interview Copilot works by picking up an interviewer's voice through system playback and drafting something you can say out loud. An assessment is a text prompt and a countdown in a silent room. There is no audio for it to work from.
- Capture exclusion solves a different problem. The overlay is excluded from screen capture at the operating system level, which keeps it out of Zoom, Teams, Meet, OBS and screenshots during a call. That does nothing about a webcam taking stills of your face, nothing about a lockdown browser, nothing about a person in the room, and nothing about per-question timings that look unusual on a report.
- This is a scored evaluation with the employer's own rules attached. One attempt, monitored, and explicitly compared against other applicants. How you choose to sit it is a decision about those rules, and it is not one we are going to dress up as a product feature.
- Windows 10 and 11 only. No macOS, no Linux, no mobile, no browser extension.
The stage where a copilot earns its place is the conversation that follows, when a person is asking questions in real time and you answer in your own voice. The preparation that pays before then is knowing which questions that person is likely to ask, which is what the company question bank generator builds for a named company and role.
The short version
Up to five library tests plus the employer's custom questions, each with its own separate clock, and often a second clock on every individual question inside them. Forward only, so there is no review pass. A skip is a permanent blank, so guess instead of skipping. One attempt, no pause. Read the instructions before each test rather than only the first, spend your time inside a test and never across the bundle, and keep something in reserve for the essay and the video at the end, because those are the parts a human actually reads.