A match score should help you decide what to open first. It shouldn't pretend to know whether a recruiter will call, and it shouldn't hide its evidence behind a confident number. That is the idea behind how Oink scores: rank a posting against your current brief, show concrete reasons for the fit and the concerns, and leave the application decision to you.
I came to this from my own search. I auto-applied to hundreds of jobs because I didn't feel I had time to read every position, and most of them were never a good fit. Reading postings against a clear brief is the part I wanted a tool to do.
Your profile is more than a resume
Oink keeps what you can do apart from what you want to do. The resume is evidence of roles, skills, domains, and seniority. Your current preferences are the target: desired role, skills you want or want to avoid, work arrangement, location and eligibility, salary floor, languages, and dealbreakers.
This avoids a common failure. If your last job was in ad-tech and your brief now excludes ad-tech, your history shouldn't override the brief. A skill missing from a short resume also doesn't prove you lack it, so Oink keeps unknowns as unknowns and doesn't turn every blank into a confident pass or fail.
If your brief is still vague, here is a practical guide to describing the job you want. Clear constraints make the reasons more useful. Keyword stuffing doesn't.
Two stages of relevance
Oink first uses a cheaper pass to cut a very large pool down, then scores each surviving job in more detail. Checks based on the posting's own fields catch explicit conflicts, such as a disallowed location or work mode, without asking a model to reinterpret the facts. The unclear cases go on to a closer read.
In the 30 days to 27 September 2026, Oink scored job postings against the briefs of 18 users 139,505 times. 90.6% scored 0 and 0.94% reached 3 or higher. That is a small group of users, a posting is scored again when someone changes their profile, and only fetched jobs are counted, so read it as a feel for the scale and not as a benchmark. Not every fetched job gets the same AI enrichment either. Providers fail and fields go missing, and when that happens the card should show an unknown or a concern.
What the 0 to 5 score means
- 0 is a rejection for an explicit or very strong mismatch. It isn't a normal delivered recommendation.
- 1 and 2 mean weak alignment or important conflicts.
- 3 is a plausible stretch, worth a look when the reasons make sense to you. It is the default send level.
- 4 is a strong "apply today" candidate against your brief.
- 5 matches your target requirement by requirement, based on the evidence available.
These numbers rank relevance inside your search and aren't calibrated probabilities. A 5 doesn't mean a five-in-six chance of an interview, and a 2 doesn't mean you can't do the job. The employer knows things Oink can't see, and the posting may be incomplete or out of date. A friend who is a researcher found a position through Oink that fit perfectly, and the job went to someone else.
Scoring has been the hardest part to get right. Early user ratings in July 2026 showed the scorer repeating two mistakes: sending roles a level above the person, and sending postings that had a title and no description. Both now score at most 2, below the send level. It is still not finished.
Read the reasons before the number
A useful verdict names specific support and specific gaps: the role or stack, seniority, work mode, location, salary, visa or language signals when the posting has them, and missing facts that matter. The posting text and its source fields are the evidence. AI-enriched fields help organize the record, and they shouldn't erase where a fact came from or contradict the source text.
A quick test: can you explain the score in one sentence without repeating the number? "Strong backend fit, EU-remote, salary above my floor, but less MLOps ownership than I asked for" helps you decide. "Great match for you" doesn't.
The guide for people who applied to hundreds of jobs without a reply shows the same evidence-first checks done by hand. The remote eligibility guide covers one of the constraints that plain keyword search flattens most.
Feedback changes evidence, not reality
"More like this" and "Not for me" are extra signals about what you want. Oink records them so later matching can use them, and enough distinct feedback can trigger a profile rebuild. One tap doesn't retrain a private model, and it can't change what a posting says. Feedback helps most when it reveals a steady preference that your brief missed.
Marking a role applied is separate. It updates your tracker and doesn't submit the employer's form. Delivery is separate from scoring too: the strongest eligible results are ranked against your score floor and a batch cap before they become Telegram cards.
Use the score as a queue
Open the high scorers first and read both the strengths and the concerns. Confirm the listing at the original source, then decide whether the unknowns are acceptable and whether the job deserves a tailored application. If a reason is wrong, fix the preference or send feedback, and don't treat the number as final.
Next step: pick the top card from your last batch and write down, in one sentence, why it scored what it did.
Continue with how Oink works end to end or see how the reasons look in Telegram job alerts.
Describe the role you want and add a resume where required or useful. Oink searches the public sources available to your run, ranks viable postings 0–5, and delivers explained matches on the web or in Telegram.