What a portfolio cannot prove
Portfolio advice runs on one assumption: if you are missing something, go build something that shows it. I wanted to know how far that actually gets you, so I measured it across two job markets.
2,840 job postings. 99 capabilities. No language model involved in the measurement.
The assumption worth checking
Every piece of career advice I have read tells you to close a gap by building a project. It is good advice right up to the point where it quietly stops being true.
No side project will page you at three in the morning.
No side project gives you a production system with real users and real consequences when it breaks.
So the interesting question is not whether portfolio projects work. It is where the ceiling is, and whether that ceiling sits in the same place for every kind of work.
How I measured it
I built two capability vocabularies from scratch, one per market, each a set of regular expressions matched against posting bodies.
Business analyst. 45 capabilities derived from 201 LinkedIn postings.
Technical. 54 capabilities derived from a role-balanced sample of 758 postings, then validated against a held-out set of 1,881 more. The two sets agreed within 1.8 percentage points on average, so the patterns travel rather than memorizing the sample.
Then the part that matters. Every capability carries a second tag for what it would take to evidence it honestly.
Solo. One person, public data, public tools. A portfolio project can show it.
Outreach. Needs other people, but not a job. Open source maintainers, a volunteer organization, real user interviews.
Employment. Needs a real organization with real stakes. No portfolio project can honestly evidence it.
Weighting each capability by how often postings ask for it turns that tagging into a ceiling.

The ceiling is the same. What is above it is not.
A portfolio can reach 73% of business analyst demand and 71% of technical demand. Two points apart. If that were the whole finding it would be a boring one.
The split above the line is where the two markets separate. For business analysts, 22% of demand needs other people and only 5% needs an employer. For technical work those numbers invert: 10% needs other people and 19% needs an employer.
A business analyst can reach 95% of what the market asks for without being hired. An engineer tops out at 81%.
That gap is not a matter of effort. It is a matter of access. The business analyst's remaining demand is things like running elicitation sessions, coordinating user acceptance testing, and driving process improvements with stakeholders. All of that is unlocked by finding people who will talk to you. A volunteer organization will hand you a broken process on a Saturday.
The engineer's remaining demand is production incidents, on-call rotations, platform migrations, and setting technical direction for a team. None of that is unlocked by trying harder. It is unlocked by someone giving you the keys.

Eleven against three
Counted as capabilities rather than weighted demand, the asymmetry is even starker. Three of the 45 business analyst capabilities are employment-gated. Eleven of the 54 technical ones are.
Two of those eleven are worth sitting with. Debugging and resolving production incidents appears in 34% of technical postings. Setting technical direction and standards appears in 28%. Those are not niche asks buried in a nice-to-have list. They are in roughly a third and a quarter of everything posted, and there is no honest way to put either on a portfolio site.

They sit at the top of demand, not the bottom
It would be convenient if the employment-gated capabilities were the rare ones. They are not. Rank every technical capability by demand and the third entry needs an employer. Do the same for business analysts and the first unreachable capability is seventeenth of 45.
So a technical candidate building a portfolio runs into the wall almost immediately, while a business analyst can work down most of the demand curve before hitting anything a project cannot touch.
What I would do with this
If you are building a business analyst portfolio
Your ceiling is high and most of what is above it is a people problem, not a permission problem. Find an organization that will let you near a real process. A nonprofit, a church, a local league, a small business with a spreadsheet problem. That single move converts roughly a fifth of the market's demand from unreachable to reachable.
If you are building a technical portfolio
Build for the 71%, then be direct about the rest. Run your own deployed service long enough to break it and write an honest postmortem. That gets you closer to incident work than another tutorial project ever will, and it is real. For what remains, say plainly that you have not done it. A hiring manager who reads "I have not carried a pager" trusts the rest of the page more, not less.
The failure mode in both markets is the same: dressing up a project as something it is not. If a capability needs an employer, no amount of framing makes a side project equal to it, and the people reading your portfolio know that better than you do.
The tool
All of this runs behind jobfit.guide. Paste the postings you are chasing, tick what you can already evidence, and it returns the gap split three ways with projects matched only against the part a project can honestly close. It never recommends a project for an employment-gated capability. It names those and tells you plainly that nothing here will fix them.
Corpora: 201 LinkedIn business analyst postings and 2,639 technical postings, collected November 2026. Raw posting text is not published. The capability vocabularies, the demonstrability tags, and the reference counts are open at github.com/chadwambles/jobfit-engine.



Comments