// help-center

Frequently asked questions.

Everything about how TrialDay scores candidates, our anti-cheat methodology, and the product.

// methodology

How TrialDay scores candidates.

Four principles power every score on the platform.

Three weighted dimensions

Every session is scored on Technical, Communication, and Collaboration. The weights vary by role — a SWE role weights Technical heavily, while a CSM role weights Communication. The overall score is a weighted average, not a simple mean.

AI-graded responses

When you reply to a teammate or write a doc, an LLM judges whether the response is technically precise, vague, or off-topic. Every grade and reasoning is shown in the report — no black box.

Adaptive to difficulty

Simulations run at Introductory, Standard, or Challenging difficulty. Scores are adjusted ×1.35 on Challenging and ×0.85 on Introductory so candidates can be compared fairly across difficulty levels.

Focus score (anti-cheat)

We monitor tab-switches, large pastes, and copy-outs during the session. The focus score (0–100) reflects engagement integrity. Recruiters see the focus chip next to every candidate score.

01

How scoring works

Is the score a multiple-choice quiz?

No. Every answer is open-ended and AI-graded against a role-specific rubric. We don't use fixed-answer questions — the same problem might have multiple correct approaches.

Can I see why I got a score?

Yes. The candidate report shows every graded response with the AI verdict, the rubric category, and concrete moments cited from your session. Pro and Business reports include the full HR debrief.

Are scores comparable across roles?

Yes within a role, with caveats across roles. The weights and rubrics differ by role, so comparing a SWE score to a CSM score isn't apples-to-apples — but within the SWE pool, scores are directly comparable.

What is the focus score?

It's a 0–100 measure of engagement integrity during the session. We start at 100 and deduct points for tab-switches, large pastes, and copy-outs. Candidates with focus scores under 70 are flagged for recruiter follow-up.
02

Anti-cheat and authenticity

Do you use webcam or screen monitoring?

No. We don't take screenshots, record video, or monitor your screen. We only track in-browser behavioral signals — tab visibility, paste size, and copy-out — through standard web APIs.

What if I switch tabs to look something up?

You can. A short tab-switch (under 1.5 seconds) is ignored. Real work involves looking things up — we only flag patterns that suggest the candidate left the simulation entirely or pasted in large external content.

Will using ChatGPT during the sim be detected?

Most likely yes. Pasting an answer from ChatGPT will show up as a large paste; switching to a ChatGPT tab will show up as a tab-switch. Neither alone proves dishonesty — but a high-scoring session with several large pastes is a red flag recruiters can follow up on in interview.

Is the focus score the same as proctoring?

No. Proctoring tools lock the browser, record you, and require special software. TrialDay never installs anything. We surface signals; recruiters interpret them in context.
03

For candidates

How long does a simulation take?

Most simulations are 45–75 minutes of real time. You can pause and resume — your progress is saved automatically.

Will employers see my mistakes?

Only if you take a company-issued simulation through their share link. Self-practice simulations are private to your account.

Can I retake a simulation?

Yes. On Pro (including your 14-day trial) you can retake any simulation as many times as you want. On Free you get 1 simulation per month.

What gets me a certificate?

A score of 65+ or identifying the root cause of the scenario's problem. The certificate has a verified URL and a LinkedIn share button.
04

For companies

How do I create a custom simulation?

Paste your job description into the company portal. Our AI generates teammates, Jira tickets, a scenario challenge, and a screening rubric. You get a share link to send to candidates.

Can candidates take it on their own time?

Yes. Share the link; candidates run the sim whenever they want. You see results in your portal as they finish.

How does Business plan compare candidates?

Sort by score, filter by behavioral flags, view per-candidate behavioral scorecards, and select up to 5 candidates for a side-by-side comparison view.

Do you integrate with ATSes?

Yes — Business plans can configure a webhook that pushes completed candidate sessions into Greenhouse, Lever, or any ATS that accepts webhooks. ATS export to CSV is available on all Business plans.

Can we customize the difficulty per role?

Yes. When generating a simulation you can pick Introductory, Standard, or Challenging. Scores are difficulty-adjusted automatically.
05

Billing and data

How does the free tier work?

You get 1 full simulation per calendar month, across any role. No credit card required.

How does the 14-day Pro trial work?

Every new TrialDay account starts on Pro for 14 days — unlimited simulations, full AI feedback reports, shareable scorecards, the works. No card required to start. If you don't upgrade before day 14, your account quietly downgrades to the Free tier (1 simulation/month). There are no surprise charges and we don't auto-bill you.

Can I cancel anytime?

Yes — cancel from billing settings. You keep access until the end of your billing period.

Is there a refund policy?

If you're not satisfied in the first 7 days, email info@trialday.app and we'll refund you. No questions asked.

Where is my data stored?

In a Supabase Postgres database in the US. We don't sell or share your data. Hiring companies only see your score and feedback report on simulations they issued — never your raw messages.

Does Anthropic train on my session?

No. We send simulation activity to the Anthropic API to generate AI responses. Anthropic's API data is not used for training by default — see anthropic.com/privacy.

Still have questions?

Email us at info@trialday.app — we usually respond within a few hours.

Email supportSee pricing
we never see your card details, never sell your data, and never train models on your sessions.