Portfolios are the first thing everyone checks and the least useful way to choose a studio, because they show what a studio made and hide how it decides. A screenshot cannot tell you whether the site was fast, whether it brought in enquiries, whether the client could edit it, or whether the studio argued for what was right or built what it was asked for. A better way to choose is to examine how a studio thinks, using questions a portfolio cannot answer.
What a portfolio can and cannot tell you
A portfolio can tell you about taste and craft: whether the typography is careful, whether the layouts have a point of view, whether the details survive a change of screen size. Those matter, and a portfolio is a fair test of them. What it cannot tell you is any of the things that decide whether the site worked. The work you are looking at was shaped by a client, a budget, a deadline and a brief you will never read, so a handsome result may belong more to a good client than to a good studio, and a plain one may be the best decision anyone made.
Treat the portfolio as a filter for taste, not as evidence of outcomes. If you like the work, you have a reason to continue the conversation. You do not yet have a reason to hire.
Ask them to decide something in front of you
Describe your real situation in three sentences and ask what they would do first. A studio that starts with a mock-up has told you something; one that starts with questions about who buys, what those people do today instead, and what would count as success has told you something else. You are not looking for the right answer, because you cannot know it yet. You are looking for visible reasoning: a first move, a reason for it, and a stated thing it depends on.
Notice what they ask you as well as what they say. Good questions are expensive to ask, because they come from having been wrong before.
Ask what they would talk you out of
Every studio will tell you what it can build. Few will tell you what it would not build, and that list is the most reliable description of its judgement. A studio that would tell you the animated intro is a mistake, that the fourth language can wait, or that you do not need a booking system yet is protecting your budget at the cost of its own revenue. One that agrees with every idea is pricing your enthusiasm.
This is also the test with the sharpest edge, because a studio that answers honestly is telling you it will sometimes say no to you after you have paid it. Better to learn that now.
Find out who actually does the work
Many studios sell with their most senior people and deliver with whoever is available. Ask who will design your pages, who will write the code, who you will speak to week to week, and whether those are the people you are meeting now. None of the answers is automatically wrong; a larger team can be the right choice. But a surprise on this point arrives in the second month, and it is much cheaper to hear it in the first meeting.
Ask what happens in month seven
Launch is the moment of least risk for a studio and most risk for you. Ask what is included after it: who fixes what breaks, how long a content change takes, what happens if the person who built the site leaves, and whether you could take the site elsewhere with everything intact.
The answers show whether the studio expects a relationship or a handover, and whether it has built things so that you are able to leave. A studio confident in its work does not need to make leaving difficult.
Look at the studio's own website, and talk to someone who left
A studio's own site is the one project where it had full control and no client to blame. Check how fast it loads on a phone, whether its headings and pages are organised sensibly, whether it says something specific or something that fits any studio, and whether the writing sounds like people. If the studio sells search work or performance work, its own site is the first piece of evidence for it.
A reference list is curated, so asking for it tells you little. Ask instead to speak to a client whose project had difficulties, or one who has since moved on. Few studios will offer that, and the ones that will have handled friction well enough not to fear the call. Whatever the reference says, ask what they found hardest about the project; the answer shows where the studio is weakest, which is what you actually need to know.
The one question to end on
Finish by asking what they would need to see to tell you this is not the right project for them. Almost nobody asks it, and the answer is more informative than any reference. A studio that can name its own limits has thought about them. One that cannot has not yet had to.