How Many Projects Should a Portfolio Show?

Ask how many projects in a portfolio is right and you will be given numbers within seconds. Three to five case studies. Twenty to forty images. Reviewers spend three to seven minutes before deciding. These figures appear in hundreds of articles, in course material, in advice from people who hire. They are quoted with the confidence of measurement.

We went looking for the study behind them and could not find one. Not a small one, not an old one, not a vendor's marketing survey. No named sample, no stated methodology, no publication. What exists is a large body of professionals repeating a range that sounds right to them, which is a legitimate thing but is not data, and should not be presented as data.

What the received numbers actually are

For design and UX work, the settled position across the industry is three to five case studies presented in depth. For photography, sources converge on twenty to forty images in a public portfolio, with twenty to thirty treated as the sweet spot, and a refresh habit of replacing the three to five weakest images every three to six months. For attention, the figure is three to seven minutes per portfolio, with the accompanying line that reviewers are looking for proof of thinking rather than proof of pixels.

The consistency across independent sources is genuine — these ranges do not vary much from one writer to the next, which is itself mild evidence that they reflect something real about how people review work. But consistency and verification are different things. A number repeated a thousand times is still one number, and the three-to-seven-minute figure in particular has the texture of a plausible estimate that hardened into a citation somewhere around 2019 and has been passed along since.

Why we are saying so

Because the alternative is to write "studies show" and hope nobody checks. That phrase does a great deal of work in this corner of the internet, and in this case there is nothing underneath it. If we are willing to tell you that a platform's own price page was unreachable, or that a vendor's commission structure could not be confirmed from a second source, we should be willing to say that the most repeated statistic in portfolio advice has no traceable origin.

It also changes how you should use the advice. Unverified consensus is not worthless; it is testimony. Hundreds of people who look at portfolios professionally converged on the same rough shape, and that is worth taking seriously in the way you take a colleague's judgement seriously. What you should not do is treat the boundary as a rule. Nobody was ever rejected for showing six case studies rather than five.

The reasoning that survives without the numbers

Strip out the statistics and the underlying argument holds up perfectly well, because it does not depend on them.

You are judged by your weakest inclusion, not your strongest. This is the load-bearing idea and it needs no survey. A portfolio is a claim about the standard you work to, and the weakest piece is the most credible evidence of that standard, because anyone can produce one good thing by accident. A reviewer reads the floor, not the ceiling. Three strong pieces make a cleaner claim than three strong pieces plus four adequate ones, and the arithmetic of that never changes with sample size.

Curated, not comprehensive. A portfolio is not an archive, and the impulse to include everything is usually anxiety rather than strategy. Every additional piece dilutes the average and costs the viewer attention that was going to run out anyway. Editing down is uncomfortable precisely because the work was expensive to make, which is why almost everyone under-edits.

Avoid duplicates in disguise. Three e-commerce checkout redesigns are one project shown three times. Three portrait sessions in the same light with the same lens are one look. Repetition reads as narrowness even when it represents a great deal of work, so keep the strongest of any cluster and spend the freed slot on something that shows a different capability.

Sequence matters more than count

If attention is short — and whatever the true figure, it is certainly short — then order does more work than volume. Lead with the strongest piece, not the most recent, and not the one you found hardest to make. Effort is invisible to a viewer; only the result is on the wall. Put the second strongest last, because openings and closings are what people retain, and let the middle carry the range.

Sequence for contrast as well as quality. Two similar images side by side flatten each other; the same two separated by something different both survive. This is exactly how a print show is hung, and the reason is the same: the eye tires of a rhythm faster than it tires of pictures.

Numbers by discipline, offered as judgement

With the caveat above fully in force: for photography, twenty to thirty images reads as edited and forty starts to read as everything you had. For design and UX, three to five case studies with real process visible beats eight that stop at the final screen. For writers, ten to fifteen clips grouped by beat or format say more than sixty in reverse chronological order. For models and actors the constraint is different again, since a book is expected to be reasonably complete — but the digitals should be current and the tearsheets should be the ones you would show a booker in person.

When more is the right answer

The minimalist advice is not universal, and there are conditions under which volume is the correct signal. Commercial and event photographers are frequently hired on evidence of consistency across many jobs rather than on peaks, and a client booking a two-day conference wants to see that you can do it eighty times. Illustrators pitching for series work are asked to demonstrate a sustainable style, which requires enough examples to prove it is not one lucky drawing.

The resolution is structural rather than numerical: a tightly edited front, and depth available behind it. A landing selection of ten, with a fuller archive one click away for the person who wants it, satisfies both readers. What fails is the undifferentiated grid of ninety, because it makes the viewer do the editing and they will not.

Keeping it alive

The most useful habit in circulation is also the least glamorous: every few months, remove the three to five weakest pieces and replace them with better ones. It keeps the count stable, forces a comparison against your current standard, and turns curation into maintenance rather than an annual crisis. It also quietly requires a platform where editing is cheap — if changing the portfolio means a weekend of rebuilding, it will not happen, which is one reason our ranking of the best portfolio website builders weighs how easily a site can be revised.

So: no, there is no study. There is a wide professional agreement, a set of arguments that stand on their own without it, and a decision that remains yours. Show the number of pieces you would be content to be judged by if a stranger looked at only one of them, chosen at random. That test is harsher than any range, and it is the only one that has ever produced a good portfolio.