INSIDE JOBRIC

By Ric @ Jobric · October 2026

A match score tells you how well a job fits. It does not tell you whether to apply.

That sounds like a small gap. It isn't. You can sit in front of an 82% for ten minutes trying to work out whether 82% is a yes, and the number will not help you, because a number is a measurement and what you actually need is a decision.

September was mostly about closing that gap, in the matches themselves and in everything leading up to them.

Researching a company isn't hard because the information is hidden. It's scattered, and pulling it together takes an evening you do not have.

So people skip it. They walk into interviews having read the careers page and nothing else. Then somebody asks what interests them about the business, and the honest answer is that they needed a job and this one was open.

A Company Briefing is a dossier Jobric compiles on demand, drawing on six or more independent sources. It comes back in three parts: who the company is, what the sources say about working there including where they disagree, and interview preparation built from both. It takes a couple of minutes.

The aesthetic is deliberate. It's a dossier, it's called a dossier, and the sections are field notes. Partly because that's more fun to read, and partly because it's honest about what the thing is. You are gathering intelligence on an organization that has already gathered plenty on you.

Every match now ends with a recommendation. The written analysis attached to a match used to grade how well the job fit, which left you doing the final step yourself. It now ends with one of four answers: apply, apply and prepare, check first, or not recommended.

Anything worth addressing before you apply is collected under a heading that says exactly that, grouped by what to do about it, in neutral type rather than warning colors. If there's nothing to address, it says so rather than manufacturing a concern for balance.

Two other changes matter as much. It no longer guesses at your motives, which previously meant a move from a corporate role to an agency could read as a step backward to a system with no idea why you made it. And it no longer treats a job title as proof you hold a specific skill. A title is evidence of what somebody called you, which is not the same thing.

It's also about 17% shorter, and the recommendation now agrees with the label on the match card. Those two were measured differently, so a card could say Good Fit while the explanation underneath sounded distinctly lukewarm. One answer is more useful than two that argue.

Jobric tells you what it read off your resume. We read every resume that comes in, and when there's something worth flagging as a possible job-matching issue, we write it down. That reading now reaches you: a notification the moment it's ready, linking straight to it, repeating the same figures the page shows rather than teasing them.

The report itself got considerably more careful about what it raises. It now separates "Jobric could not work this out" from "here's something that may affect your matches," and only the first takes over the report. Previously any possible job-matching issue outranked everything else, which meant a resume with a clear work history and well-dated roles could be told we couldn't read it, because it opened with a phone number instead of a summary line.

Findings also come with the count behind them. "None of your six roles has a description" is something you can hold up against your own page and decide about. A note with no number behind it is just an opinion.

We got substantially better at reading what you do and where. Two separate improvements, same underlying idea.

An early step in our matching used to settle what kind of work you do by looking at your most recent job title alone. A title that merely resembled a known one was close enough. Now anything short of an exact match goes to the model that actually reads your resume, which is meaningfully more accurate about what you do.

Separately, we improved how industry is read off incoming job listings, so we can now identify the sector on the large majority of jobs in the pool.

Both of these sit upstream of everything you see. A job sorted into the wrong career category cannot be scored against you properly no matter how good the scoring is.

Your week arrives, and it arrives at a sensible hour. Finding matches happens in two steps: we pull candidate jobs, then we score how well each fits. The scoring finishes roughly half an hour after the search does, and those finished scores now land on your board as they complete, inside the same weekly allowance. A week that comes up empty no longer discards the scoring work it just did, and when a weekly run does come up empty, it now records why.

Our outbound mail had also been arriving at unpredictable hours. It now lands in a single window at 10am Eastern, which is early afternoon in the UK and a civilized hour on the west coast.

The weekly digest itself also compares your match score against last week rather than yesterday, which is the comparison that actually tells you something.

Two that were quietly getting in the way.

The status that ran ahead of the facts. If a resume upload didn't go through, the screen could still show a green tick and the words "Resume analyzed," which is a confident thing to say about a document you do not have. The status now reflects what actually happened, so a failed upload looks like one and you can do something about it instead of waiting on matching that was never going to start.

The digest that stopped writing parts of itself. A couple of sections of the weekly email, including the salary summary and my own note at the top, had quietly stopped generating. The email still arrived, on time, slightly hollow, as though someone had left the building partway through. Both are back.

Telling us where you want to go, not just where you've been. The last onboarding step will show what we read from your resume and ask which career category you actually want to pursue, with room to describe in your own words what you're after. That shows up as a line on each match and never touches the score. It's built and not yet switched on. The reason I'm interested in it goes beyond matching: every answer is a direct read on how accurate our categorization is, from the one person qualified to grade it.

Refreshes that spend their effort where the matches are. We found that match refreshes were putting most of their work into a catch-all category that rarely produced anything good. The fix is ready and waits on a related measurement before it goes live.

Career Guides, close now. Guides are being recruited and interviewed, and the interview opens with a twenty-minute live sample coaching session before any agreement is signed, so nobody is approved on the strength of a resume and a conversation. Seekers will also be able to say whether they'd prefer a woman or a man as their guide, which moves matching guides up the list without hiding anyone. Booking isn't open yet. No date, because a date that moves costs more than saying nothing.

The quieter work in September was about measurement, and it keeps producing the same lesson.

We now measure, end to end, how good the matches on your board actually are. Getting to a figure we trusted took longer than producing one would have, and it is now measured the same way every month, so it can move and we will see it move.

The internal page measuring the resume report was built the same way. It records what happened when somebody replaced their resume after reading their report, and if a shorter, sharper resume moves the counts down, down is what gets recorded. A measurement that can only ever show improvement is the one thing that feature cannot afford to be.

None of that is visible to you. All of it decides whether anything else here is true.

A product that can't tell when it's wrong will tell you it's right with exactly the same confidence either way.

The job market works for you now.

That's the update. Now go do something that isn't job searching.

Ric @ Jobric