Proof to apply

OFM Jobs got an employer to a shortlist, but a shortlist is a stack of claims. "Fluent English," "fast typist," "great on calls": everyone writes it, few can prove it, and you only found out it was false after an interview you'd already paid for. Tests move the proof all the way to the front. An employer sets the required skills when they post the job, chosen from a shared test library, and a candidate can't apply until they've cleared the bar. Candidates prove it once, on their own time, and carry a verified profile from job to job. I designed and shipped it end to end, from the job post to the gate, the five tests, and the way proof lands back on the board.

The Story

A résumé is a promise, not proof.

For these roles the job is the skill itself, and almost none of it survives a CV.

  • Can they actually hold a conversation in English, or just tick "B2"?
  • Do they type fast enough to run three chats at once?
  • Is their connection even stable enough to work?
  • A CV answers none of it, but the job depends on all of it.

You found out too late.

The only place to check a skill was the interview, the most expensive filter in the funnel.

  • Thirty minutes in, you learn they type 22 WPM.
  • A call that drops every two minutes, discovered on the call.
  • The filter existed; it just sat at the wrong end of the pipe.

The Decisions

Test the job, not trivia.

Five tests, each mapped to something the work actually needs, and nothing it doesn't. An off-the-shelf vendor would have been faster and wrong: generic tests screen generic skills, and they run on someone else's site, so the candidate and the score both leave OFM. So the tests are built native, shaped to chat work, and the result lands straight on the board.

  • English: reading and grammar, the baseline for chat work.
  • Verbal: a spoken prompt, recorded, for roles that get on calls.
  • Listening: audio comprehension, because half the job is catching what's said.
  • Internet speed: the boring thing that quietly ends chat work.
  • Typing: raw words-per-minute and accuracy under a clock.

Proof to apply, not proof to shortlist.

A test that runs after the shortlist still lets a stack of claims into the pile. So the test became the door. A candidate only sees Apply on the roles their verified skills already clear; the rest stay locked until they prove them. Nothing is submitted below the bar, so the pile an employer receives is pre-filtered, by the standard the role needs, not by a person.

  • Required skills are chosen when the job is posted, not bolted on later.
  • Roles a candidate has already proven are one-click apply; the rest stay locked until they take the test.
  • Proof sits at the very front of the funnel, where the leak used to start.
  • Clearing the bar opens the door, not the job. The employer still chooses among everyone who qualifies.

Grade it the moment it's done.

Waiting on a human to mark tests would just move the bottleneck. So the machine grades what it can, the instant it can.

  • Typing and internet speed are measured, not judged. Numbers, not opinions.
  • English and listening auto-score against a key.
  • Verbal is scored by AI, with the reasoning shown, never a number from nowhere.

A score you can trust.

A result nobody understands is a result nobody acts on, and a test that punishes a bad day tests the wrong thing.

  • Every score opens into how it was reached.
  • Pass marks are the employer's to set, per test and per role.
  • A weak connection or a shaky first take is flagged for a retry, not scored as a fail.
  • Retakes and flags stay visible, not buried in an average.

Prove it once, carry it everywhere.

Making a candidate re-test for every employer is its own kind of leak. So the proof lives on the candidate, not the application. Take a test on your own time, and the verified score sits on your profile, ready for the next gated job that needs it.

  • Take any test proactively, not only when an employer invites you.
  • Verified scores live on the profile and travel from job to job.
  • Already clear a job's bar? Apply instantly. "Fluent English" is now "English: 92, verified."

The Full Flow

Set the bar when you post.

Requirements aren't an afterthought; they're part of posting the job. Add the skills the role needs and the score each one has to clear.

  • A "Required skills" step, right inside the job post.
  • Pull each test from the shared library and set its pass mark.
  • The job goes live gated: no proof, no application.

A library, not a quiz you build.

Nobody writes their own typing test. The tests are standardized and shared, so the same skill means the same thing on every job across OFM.

  • Pick from a catalog of ready, chat-work tests.
  • Same test, same scale, every employer, so scores compare.
  • Choose it, set the bar, and it's attached to the role.

The job, with a gate.

On the candidate's side, the listing is honest about what it takes. The required skills sit right up top, and Apply is locked until they're met.

  • The verified skills the role needs, shown before you start.
  • A locked Apply button; the bar is visible, never a surprise.
  • Green ticks for what you already hold, from earlier tests.

Prove it once, on your own time.

A candidate doesn't have to wait for an invite. They can take any test proactively and build a verified profile that opens doors before they even knock.

  • Take a test whenever, not only when an employer asks.
  • Verified scores live on the profile, ready to reuse.
  • One good result unlocks every job that needs it.

What the candidate opens.

Whether it's to clear a gate or top up a profile, the battery opens as a plain, unintimidating page: here's what you'll do, here's how long it takes.

  • The set laid out up front, with an honest time estimate.
  • A mic and connection check before anything counts.
  • One button to begin; the battery runs itself from there.

English.

A short reading-and-grammar set that mirrors the messages they'd actually send.

  • Real chat snippets, not textbook sentences.
  • One question at a time, auto-advancing.
  • Auto-scored against the key the moment it ends.

Verbal.

A prompt on screen, a recorder underneath: say your answer out loud.

  • A scenario to respond to, like a real customer.
  • Record, hear it back, submit.
  • AI scores fluency and clarity, and shows why.

Listening.

Play a clip, answer what it asked, the half of the job that isn't typing.

  • Short audio, played once or twice.
  • Comprehension questions that follow the clip.
  • Auto-scored, no employer time spent.

Internet speed.

A live test of the thing that silently decides whether someone can do chat work at all.

  • Download, upload, and ping, measured live.
  • A clear pass line for what the work needs.
  • Flagged, not failed, so a bad reading can be retried.

Typing.

A timed passage with words-per-minute and accuracy ticking up as they type.

  • Live WPM and accuracy, no waiting for a result.
  • The same passage for everyone, so scores compare.
  • One number the employer already understands.

Scored, and saved to the profile.

The moment they finish, one card sums the battery, and the results don't just vanish into one employer's inbox.

  • Five results in a row, each with a pass or a flag.
  • The AI's notes on the verbal answer, in plain words.
  • Saved to the profile, not sent once and gone.

Application unlocked.

Back on the gated job, the proof does its work: every required skill turns green, the lock falls away, and applying takes one tap.

  • Cleared the bar, so Apply lights up.
  • The verified scores attach to the application automatically.
  • You apply already proven, not just hopeful.

A pre-qualified pool.

On the employer's side, the applicant list looks different now: everyone in it has already cleared the bar. There's nothing to weed out, only the best to find.

  • Every applicant already meets the required skills.
  • Sort among the qualified, not sift through the claimed.
  • Verified badges on every row, comparable at a glance.

Verified, from column one.

The pipeline starts where it used to end. The very first column is already proven, and the AI match score rests on evidence, not a résumé's word.

  • The first Kanban column is verified, not hopeful.
  • Open a card and the full scorecard is right there.
  • Testing and pipeline, the two halves of OFM Jobs, become one flow.

The Outcome

Impact.

Claims became proof, and the filter moved all the way to the front. The board stopped being a place to weed out bad fits; every candidate on it had already cleared the bar. The interview became the reward for passing, not the place you discovered the truth.

  • Applications arrived pre-qualified, so the "types 22 WPM" surprise never reached the board.
  • Candidates who tested proactively applied in one tap and reused a single result across roles.
  • A role went from open post to a verified, ranked pool without a manual screen in between.
  • Hiring, from post to offer, now runs start to finish on OFM, so the platform earns a seat at every hire.
OFM JobsTests
Build note pending
Build note pending