A hiring manager opens your application with two things in front of her: a letter and a link. If the letter narrates what sits behind the link, she now has the same information twice, in the weaker format, and no reason to click. That is the failure this piece fixes.
It shows up constantly in applications from designers and front-end developers, but it is just as common from copywriters, content marketers, teachers who bring lesson materials, analysts with a public notebook, video editors and grant writers. If you have any finished work sample you can point at, this is your problem too.
The letter is not a written version of the portfolio. Its job is to make one piece worth opening, and to tell the reader what to look at once they get there.
Two copies of the same thing is worse than one
A portfolio page has screenshots, a live demo, a before and after, maybe a short video. Prose has none of that. Describing the work in sentences means competing with your own strongest asset and losing.
Worse, you are spending the letter's scarcest resource. A recruiter gives a cover letter somewhere between fifteen and forty seconds on the first pass. If twenty of those seconds go to a paragraph summarising a case study, you have used half your budget saying something the link says better. The reader does not need a summary. She needs a reason and a direction.
The one-piece rule
Name one project. Not five, not "a selection of my recent work", not a tour.
The instinct to list everything comes from insecurity: if I show five, one will land. In practice five projects in a paragraph means four sentences of nothing and a reader who opens nothing. One project with a specific outcome gives a clear next action.
Pick by closest match to the job, not by pride. The thing you are proudest of is usually the thing that was hardest to make, and difficulty is not relevance. For a B2B SaaS onboarding role, the messy internal dashboard you rebuilt beats the beautiful personal project, even though the dashboard is uglier.
If you genuinely have two strong candidates, still name one and let the portfolio homepage carry the rest. The reader who likes the first piece will browse. The reader who does not was never going to.
Frame the piece around their problem, not your process
The default framing is chronological: I was briefed, I researched, I iterated, I shipped. That is a process story. Nobody is hiring your process, they are hiring the result it produces on their problem.
Read the job posting for the thing they are quietly worried about. Postings leak this. "Comfortable working without a dedicated researcher" means they have no research function. "Owning the design system end to end" means it does not exist yet, or nobody maintains it. "Helping the team ship faster" means releases are slow and somebody is annoyed about it. Then attach your piece to that worry. Same project, different entry point.
The sentence that does the most work
There is one line in a portfolio letter that outperforms everything else, and most applicants never write it. Call it the what-to-look-at sentence.
It has three parts: which piece, which part of it, and what that part proves about the problem they have.
Something like: "If you open one thing, make it the checkout rebuild, specifically the two step flow near the bottom of the page. That is the piece where we cut abandoned carts from 71 percent to 54 percent, and it is the closest thing I have to the problem in your posting."
That line beats any biography paragraph, because it converts a passive link into an instruction. It also signals editorial judgement: anyone who can tell a stranger which part of their own work matters has looked at it from outside themselves, which is rarer than it sounds.
The weak version and the rewrite
Weak, narrating:
"In my most recent role I led the redesign of the client portal. I started with stakeholder interviews and a competitive audit, then built wireframes and a clickable prototype, which I tested with eight users. After two rounds of iteration we launched in March. The final result is clean, modern and user-focused, and you can see the full case study in my portfolio."
That paragraph takes eighteen seconds to read and contains one piece of information: there is a case study somewhere.
Rewritten, pointing:
"Your posting mentions support tickets from the client portal eating the team's week. I rebuilt a portal with the same problem last year. The fix was not visual, it was killing the separate login for billing. Tickets dropped 38 percent in the first quarter after launch. It is the first item in my portfolio, and the part worth two minutes is the section on the merged login."
Shorter, and the reader now knows what to click, why, and what she will find.
Adjectives are not evidence
"Clean, modern, user-focused." "Intuitive." "Bold yet accessible." "Pixel perfect."
These describe nothing. Every applicant for every visual role uses them, so they carry no information and slightly lower your credibility: a person who can only describe their work in adjectives has never been asked to defend it. Replace each one with the outcome it was meant to produce, or with a constraint you worked inside.
- Instead of "clean and modern": "stripped the settings page from 31 fields to 9 without losing a single support workflow"
- Instead of "user-focused": "the change came from watching four users fail the same step"
- Instead of "engaging copy": "the new subject line moved open rate from 19 to 31 percent across 40,000 sends"
- Instead of "well-organised lesson plans": "the unit worked for a mixed class where six students read two grades below level"
Numbers are ideal, but a constraint works almost as well when you have none. "Built for a team with no designer after I left" beats any adjective.
Work you cannot show
Plenty of good work is locked up. Client NDAs, internal tools, regulated industries, anything behind a login, anything owned by a team that never published it.
Say so directly and keep the outcome. The outcome is rarely confidential even when the artefact is.
"The strongest thing I have built is an internal claims tool I cannot publish. What I can say is that it took average handling time from 11 minutes to under 7 for about 90 agents. I am happy to walk through the decisions in a call, and the closest public equivalent in my portfolio is the intake form project."
You explain the gap before the reader invents a worse explanation, you keep the credential, and you give a substitute to click so the link still has a job.
Never post NDA'd work publicly and hope nobody notices. In a small market they do notice, and it reads as a judgement problem rather than a portfolio problem.
Crediting a team without erasing yourself
Most real work is shared. Applicants fail in one of two directions: "I" everything, which reads as dishonest to anyone who has shipped a product, or "we" everything, which leaves the reader unable to tell what you did. Use "we" for the project and "I" for your contribution, in the same breath.
"It was a team of five over about four months. I owned the search and filtering, including the decision to drop faceted search after testing showed people only ever used two filters."
That is honest and specific. Note the second half: the sentence names a decision, not a task. Tasks are interchangeable, decisions are yours.
A portfolio that is thin, or mid-rebuild
Two common situations, two different moves.
Thin. One or two pieces is fine if the letter points at the right one. Do not apologise for the count and do not pad with student work. One relevant, well-explained project outperforms twelve scattered items, because the reader was only ever going to open one thing.
Mid-rebuild. Never send a link to a page that says "coming soon" or shows a half-finished grid. Link straight to the one piece that is genuinely readable rather than to the homepage. A deep link to a finished case study beats a front page that apologises. If the site is unusable this week, send a two-page PDF of the single project and say the full site is being rebuilt.
Where the link goes, and how often
Once in the letter, once in the CV header. That is it.
Repeating the URL four times does not increase clicks, it signals anxiety and eats space. Put it next to the what-to-look-at sentence, the only moment the reader has a reason to click. A link in the opening line arrives before the reason exists; a link in the sign-off arrives after the reader has decided.
If the application form has a dedicated portfolio field, fill it and drop the URL from the letter body, keeping only the sentence that says which piece to open. The form already carries the address.
Before you send
- One project named, chosen by fit and not by pride
- That project tied to something the posting is worried about
- One sentence saying what to look at and what it proves
- Every adjective replaced by an outcome or a constraint
- Unshowable work acknowledged with its result kept
- Team work credited with your own decision named
- Link present twice across letter and CV