Last updated · 2026-08-23
Seven dimensions × three peers. One trust rubric.
The public methodology behind the 0–100 Worker Trust, Employer Score, and Agency Score — the same rubric the live data plane on /trust/explainer uses, written out in plain language so a reader does not need a JS round-trip to understand what is being measured.
This copy is a product draft supplied by Kayanin and is pending qualified legal review before launch. It states what Kayanin does and does not do; it is not legal advice.
The methodology
Nine sections, in plain language.
The seven dimensions, side by side
The seven dimensions, in the canonical reading order on the trust explainer:
- 01 · Identity & registrationWorker: Government ID verified, account attributes complete, and the profile flag is unambiguous.Employer: Registration / tax-id cross-referenced at sign-up, before any listings can be posted.Agency: Agency registration, accreditation, and POEA / DMW licence fields match the public registry.
- 02 · Licence & advisoriesWorker: Worker-side flags surface when the platform sees a name match on a DMW / DFA advisory; clean when the screening returns no match.Employer: Migrant-worker licence cross-reference plus posted-pay diff-checked against the actual offer letter on listed roles.Agency: POEA / DMW licence on file, licence stays current, and no prior advisories — checked nightly against the public registry.
- 03 · History & follow-throughWorker: Verified employment history filed at submission; closed applications vs withdrawn vs active, weighted by recency and volume.Employer: Closed, withdrawn, and reposted requisitions weighted by recency and volume + time-to-response / offer-revocation tallies.Agency: Placements divided by total tracked events across companies; cross-company history separate from any one Employer score.
- 04 · Profile depthWorker: Profile fields completed (bio, photo, resume, optional AI course progress) reflect depth of practice on the platform.Employer: Company profile completeness, posted-pay disclosure, and presence of a verifiable on-platform surface all contribute to the score.Agency: Agency profile fields (name, services, industries, locations, worker types, communications, available workforce) complete and current.
- 05 · Reviews & ratingsWorker: Stars + outcome-context from Employers and from Agency intermediaries; weighted by editorial review.Employer: Stars + outcome-context from applicants, plus the response, applicant-update, and offer-terms follow-through a posting promised.Agency: Stars from Workers placed, from Employers served, and the ghosted / disputed / resolved ratio from worker reviews.
- 06 · Advisories & flagsWorker: Government-ID re-verification, work-history discrepancies, and third-party reports all flow into the advisory channel.Employer: POEA / DMW / BI advisory hits, unlicensed recruitment claims, off-platform-only contact, and "no interview" pressure all flow here.Agency: Ghosted Workers, pay disputes, advisories filed against the agency, and placement-integrity flags all flow here.
- 07 · Recency & cadenceWorker: Daily-login streak ties the rest of the dimensions together over time; a stale streak is a noisy signal in either direction.Employer: Posted-pay audits surface payout drift; recent postings carry more weight than older ones in the score.Agency: Recent placements and recent worker reports carry more weight than older ones; the cross-company history stays current.
01
What a score is
Every signed-in Member on Kayanin sees three peer scores on the platform: the 0–100 Worker Trust score, the 0–100 Employer Score, and the 0–100 Agency Score. Each score is computed by the same trust engine, but with a different rubric — the seven dimensions are the same conceptual axes, weighted differently per peer.
A score is a signal, not a verdict. The score reflects what the data plane has measured; the platform surfaces it so that two peers can decide whether to hold a direct conversation. A high score does not mean "this person / company will place you / hire you / pay promptly" — it means "the data on the platform has measured X, Y, Z, and X, Y, Z are positive". The same score, read by a reader who brings outside context, can land differently for different conversations.
02
The seven dimensions, three peers
The seven dimensions are: Identity & registration, Licence & advisories, History & follow-through, Profile depth, Reviews & ratings, Advisories & flags, and Recency & cadence. The same seven are applied to every peer; the inputs differ, but the rubric shape is constant.
Below: every dimension × every peer, in plain language. The same labels feed the live /trust/explainer data plane; readers who want the signed, real-time numbers can pull them from there.
03
Worker Trust — seven dimensions, in context
A Worker’s Worker Trust score is the 0–100 composite of the seven dimensions listed above. The score is re-computed nightly and is visible to Employers and to Agency intermediaries before any conversation is opened.
The per-peer row at the top of this page is the canonical reference; the three notes below name the axes that carry the most weight per peer.
- History & follow-through — verified employment history, closed applications vs withdrawn vs active, and the daily-login streak tying the rest of the dimensions together over time.
- Advisories & flags — government-ID re-verification, work-history discrepancies, and third-party reports flowing into the trust layer’s hard-skip signal.
- Recency & cadence — daily-login streak weights are calibrated so a stale streak is a noisy signal in either direction; the score does not freeze on first-measure.
Each row above is the same dimension set reduced to a short note — the canonical reference is the per-peer panel at the top of the page.
04
Employer Score — seven dimensions, in context
An Employer / Business’s Employer Score is the 0–100 composite of the same seven dimensions. The score is re-computed nightly and is visible to Workers and to Agency intermediaries before any application is opened.
The three axes below carry the most weight per peer.
- Licence & advisories — migrant-worker licence cross-reference plus posted-pay diff-checked against the actual offer letter on listed roles.
- History & follow-through — closed, withdrawn, and reposted requisitions weighted by recency and volume, plus time-to-response and offer-revocation tallies.
- Reviews & ratings — stars + outcome-context from applicants; the response, applicant-update, and offer-terms follow-through a posting promised.
The posted-pay diff-check on Licence & advisories is the layer that catches payout drift before a Worker ever sees the listing; posted pay that does not match the actual offer letter drops the score on the next nightly re-score.
05
Agency Score — seven dimensions, in context
An Agency’s Agency Score is the 0–100 composite of the same seven dimensions, computed across every company the agency works with — the cross-company history stays separate from any one Employer score.
The three axes below carry the most weight per peer.
- Advisories & flags — placement-integrity catches "placement fee" / "processing fee" charges to Workers; the boundary is non-negotiable.
- History & follow-through — placements divided by total tracked events across companies, weighted by recency and intact placement sequence.
- Identity & registration — agency registration, accreditation, and POEA / DMW licence fields match the public registry, current as of every nightly re-check.
A confirmed "Worker charged" report drops the Agency Score sharply on the next nightly re-score; the cross-company history rollup does not forget the old pattern.
06
Data sources
The data behind every score on Kayanin comes from one of three sources. None of them are sold, traded, or shared with a third party for any non-platform purpose.
- On-platform signals — the profile, the application history, the verification status, the reviews, the daily-login streak. The Worker owns this data; Kayanin holds it under the Privacy Policy.
- Public-registry signals — registration / tax-id, POEA / DMW licence cross-reference, posted advisories on DFA / DMW / BI. Kayanin looks these up at sign-up and re-checks them on a nightly cadence; if the registry updates, the score updates with it.
- Editorial / review signals — moderation decisions, resolved reports, and the human-gated verification + payout-release decisions. These flow into Advisories & flags.
07
Cadence — what is automatic and what is human
Scoring is automated and re-runs nightly. The nightly job reads every new on-platform signal, refreshes the public-registry cross-references, and recomputes every active score. The human side is narrower: verification at sign-up is human-gated, payout release is human-gated, and moderation decisions flow into Advisories & flags as editorial signals.
A nightly re-score does not mean a signal can be ignored. A red flag — a confirmed scam report, a posted-pay discrepancy, an unlicensed recruiter — drops a score on the next nightly re-score, not after a manual review.
08
The trust-signal-not-endorsement boundary
Kayanin is a trust platform, not a placement agency.
Kayanin does NOT guarantee employment. A 0–100 Worker Trust score, an Employer Score, or an Agency Score is a trust signal, not a job offer or a referral. The score does not say "this is the right person / company / agency for you"; it says "the data plane has measured X, Y, Z, and the measurements are positive on those axes". A reader with more context — a Worker who already has one offer in hand, an Employer who has not finished interviewing, an Agency that has just lost a major client — will read the same score differently.
Kayanin is NOT a recruitment agency. Kayanin does not place Workers into jobs. The score is the input to a decision the two peers make; Kayanin is not party to that decision.
09
Changes to this methodology
We update this page when a dimension is added, removed, or re-weighted — for example, when a new signal becomes available, or when enforcement language on an existing dimension needs to be tightened against a new scam pattern.
The "Last updated" date at the top of this page is the source of truth for the current revision. Material changes are also announced on the home page. The live data plane on /trust/explainer inherits the same rubric definitions; updates to the rubric land there on the same day.
010
Contact
Questions about this methodology — and any disputes about how a score is computed — go to kayanin@proton.me. The team reads every message; nothing routed to that address is ignored.