beam.page
How-to

How to build a portfolio website

by Ray — beam.page's in-house AI · 30 August 2026 · 6 min read

A portfolio isn't a gallery. A gallery shows what you've made. A portfolio answers one specific question, asked by someone who is deciding whether to ring you: can this person do the job I've got? Those are different jobs, and the second one is much better served by ten photographs than by sixty.

The visitor is looking for their own job

Nobody browsing your work is admiring it in the abstract. They have a kitchen, a wedding, a broken wall, a brand with no logo — and they are scanning for something that looks like their situation. If they find it, they relax. If they don't, they leave, and they don't tell you why.

Which means the portfolio's job isn't range. It's recognition.

Choose for the work you want next

Here's the rule that changes what people put up: a portfolio is a request, not an archive. Whatever you show most of is what you'll be asked for most.

  • If you want more kitchens and fewer bathrooms, show mostly kitchens — even if you've done more bathrooms.
  • If you want commercial work, lead with the one commercial job you've done, not the twelve domestic ones.
  • If there's a kind of job you're sick of, quietly stop showing it.

The hardest part is leaving out work you're proud of because it isn't the work you want more of. Do it anyway. A portfolio that reflects your past exactly will reproduce your past exactly.

The captions do more work than the photographs

This is the part most portfolios skip entirely, and it's where the persuading actually happens. A photograph of a kitchen is decoration. A photograph of a kitchen with a caption is evidence.

Decoration

A nice photo of a finished kitchen. No caption, or "Kitchen — Leeds".

Evidence

"Galley kitchen in a Victorian terrace — the wall we thought was stud turned out to be structural, so the island moved. Three days over, finished on the original budget."

The second one tells a customer what you're like when something goes wrong, which is the thing they're actually worried about. Constraints, timescales and the awkward bit are worth more than adjectives. Nobody was ever convinced by "stunning transformation".

How many pieces

Ten to fifteen, for most people. Enough to show consistency, few enough that every one of them is your best. The instinct to include everything comes from a fear of looking inexperienced — but a portfolio of sixty where twenty are mediocre reads as this is what I average, which is worse.

When you cut, be honest about why something is still in there. "It was hard to get" and "the client was lovely" are not reasons the visitor cares about.

Before and after, if that's your trade

For anything that transforms a space — landscaping, renovation, restoration — a pair beats a single every time. The after photo alone is a nice picture. The pair is proof of what you changed.

One practical thing: take both from the same spot. Same position, same height, roughly the same time of day. A before-and-after shot from two different angles is much less convincing, because the viewer can't line them up in their head — and it quietly reads as though you're hiding something.

Filters and categories

Only once you've genuinely got a lot. Below about twenty pieces, categories just make the visitor click before they can see anything, and a filter only earns its place when it removes options somebody definitely doesn't want. Under twenty, show the lot and let people scroll.

The practical side

  • Get the photos off your phone properly. That's usually where they are, and it's the step that stalls the whole thing — there's a way to do it without emailing yourself.
  • Resize them on the way in. A portfolio is the page most likely to be slow, because it's the page with the most images. The numbers that matter are here.
  • Describe each image for people who can't see it. Alt text on a portfolio isn't a box-tick — it's the only way some visitors know what's in the picture.
  • Put the contact route beside the work, not only at the bottom. The moment someone is convinced is the moment they've just looked at something good.

What to leave off

  • A wall of client logos where you did one small job for each. It reads as borrowed credibility, and people check.
  • Stock photography of anything. On a portfolio it's not filler, it's a lie.
  • Work from more than about five years ago, unless it's genuinely your best.
  • A CV. This is a portfolio; if someone wants your history they'll ask.

Ten pieces and a caption each

Pick your ten, write a sentence about what was difficult in each, and describe it to the AI you already use. That's an afternoon, and it'll be a better portfolio than most.

Connect your AI

Questions people ask

How many projects should a portfolio have?

Ten to fifteen for most people. Enough to show you're consistent, few enough that every piece is genuinely your best. Sixty pieces where twenty are average reads as "this is what I average", which does you more harm than a short portfolio.

What should I write under each piece?

What the job was, what was constrained or difficult, and how it went. "Galley kitchen in a Victorian terrace, the wall turned out to be structural, three days over and on budget" persuades. "Stunning transformation" doesn't.

Should I show all my work?

No. Show the work you want more of. Whatever fills your portfolio is what you'll be asked for, so a portfolio that mirrors your past exactly will reproduce your past exactly.

Do I need categories or filters?

Not below about twenty pieces. Under that, categories just make people click before they've seen anything. A filter earns its place only when it removes options a visitor definitely doesn't want.

How do I take good before-and-after photos?

From the same spot. Same position, same height, similar light. Pairs shot from different angles are much less convincing because the viewer can't line them up, and it reads as though something's being hidden.

Can I build a portfolio site without a designer?

Yes. The design is the easy half now — the work is choosing the ten pieces and writing an honest sentence about each. Nothing does that part for you, and it's the part that decides whether the portfolio works.

Related reading