By Pierre Dragicevic. January 2022, last updated March 2022.
This guide was originally written for students doing a Master's thesis in France in the areas of human-computer interaction (HCI) and information visualization, and who are already doing their research internship. Some of the advice generalizes to other cases. Thanks to members of Aviz who contributed.
First off, think of whether a PhD thesis is for you. This is a huge investment, which requires the ability to work hard on the same topic for three years, and which requires certains types of skills, personality traits, and attitudes. You can check the six criteria I consider important. But note that this is only my own subjective opinion, and that some of the criteria are specific to the areas of HCI and information visualization.
You should start searching early on during your M2 internship, e.g., around March/April. Do this as a background task, a little bit every week, and don't wait until the last minute. Have an updated CV ready, as well as personal web page, even if you don't have much to show. Think of what research topics (research fields, application areas, technologies,...) you find exciting.
To do a PhD thesis, you will need an advisor and a source of funding. Broadly speaking, there are two types of fundings:
I advise you to look for both.
You need to subscribe to job announcement mailing lists early on (see below), and monitor your emails carefully. You will start receiving lots of emails, but bear with it, this is about your future after all!
Once you subscribe to a mailing list, consider browsing the archives to see if you haven't missed important postings. Some mailing lists don't have public archives, but you can ask someone (e.g., your internship advisor) to check their mailbox and forward you interesting postings.
Other resources with job postings in HCI / visualization:
Each time you find an interesting topic, contact the people and send the stuff they ask (CV, etc.). Prepare to be interviewed. It's considered OK to be applying to several positions at the same time, as long as you're transparent about it. Both parties (prospective student, prospective advisor) should ideally negotiate a deadline for making their decision and notifying the other party. Don't apply to too many positions (5 is fine, 30 is probably too much). Don't apply to a position you're sure not to take, either because of the topic, the location, the team or the advisor (see below). Don't send a generic application – show that you've read the topic carefully.
Not all PhD positions are equivalent. If you are just seeking to obtain a diploma, the choice of PhD position won't make an enormous difference, except maybe in terms of making the process more or less pleasant. But if you want to continue in the academia, your chances of success are greatly dependent on the PhD position you choose.
In considering a PhD position, you need to choose your advisor and research team very carefully. There's a high variability in terms of how dynamic, productive, friendly, helpful, etc. research teams and advisors are. Look at the advisor and team's web page, and ask around you about them (including your internship advisor). Pro tip: look at the advisor's Google Scholar page. Are they active research-wise? Do you like their papers? If everything appears to be great about an offer (team, topic, location, advisor), then try to speak with the advisor's students or ex-students (discreetly). Don't hesitate to email them and ask to chat over Skype/Zoom. Often it's the only way of detecting a terrible (unavailable or even abusive) advisor. You don't need to do all of this before you apply, but you should definitely do it before deciding whether you will accept or reject an offer.
There's a trade-off between how famous an advisor is, and the time they have for you. It's good to have an advisor who is famous for several reasons: often they have lots of experience and expertise for getting papers published in impactful venues, they have connections to help you find a good post-doc later, etc. At the same time, an advisor who is famous tends to have many PhD students, and may have little time to advise you. Some hire many PhD students to work on their own, and wait to see who is good enough to succeed. Fortunately, there are rules and limits in France to prevent this. Still, ask your prospective advisor how many PhD students they have, and how often they meet them (once a week is standard). It's perfectly OK to ask this.
Deciding where and with whom do a PhD thesis is a very important decision. But obviously, also pay attention to the topic: there's nothing worse than working on a topic you don't like for 3 years. Often the topic is not clear from the description itself, so you'll need to talk to the advisor and ask questions. Topics can be negotiated to some extent: it depends on the project and the advisor. Don't sign for a topic you're not at all excited about, or a topic you don't understand at all. On the other hand, don't be too uncomfortable if the topic appears vague/unclear: this is research. Part of your work will be to define your topic more clearly and more narrowly. Don't wait for the perfect topic to fall from the sky. You need to know what you're interested in, but you also shouldn't be interested in something too narrow. There are not so many PhD topics out there.
To get an idea of what information to seek about your potential supervisors, check out the "The Definitive ‘what do I ask/look for’ in a PhD Advisor Guide": https://www.cs.columbia.edu/wp-content/uploads/2019/03/Get-Advisor.pdf. I personally find it's a bit extreme in terms of the amount of information to ask yourself, but there's definitely a sweet spot behind this and completely shooting in the dark.
Be aware that many researchers will not hire a PhD student unless they previously had them as an intern (either a Master's internship, or an extra internship afterwards). Three years of advising is a big investment for a researcher and a bad match can be quite stressful for them too, so many researchers are more comfortable if they have the prior opportunity to work with the student during a few months and see if all goes well. So if things are going very well with your current internship advisor from your perspective, you should consider them as a potential PhD advisor. Discuss with them, and see for example if you want to apply to a generic funding together (more on this further below). But it's always a good idea (and perfectly accepted) to be searching elsewhere in parallel, unless (i) your advisor has 100% guaranteed funding for you, (ii) is allowed to advise you for sure (e.g., they have the ''habilitation à diriger les recherches (HDR)''*, and they don't have their number of students maxed-out) and (iii) you're both extremely excited at the idea of working together and you've negotiated a research topic you both like. Keep in mind that a generic funding is never 100% guaranteed unless your advisor got confirmation that it was accepted.
(*) often, a PhD student is co-supervised by a senior researcher having the HDR and a more junior researcher without an HDR. When this is the case, the junior one will typically supervise the PhD student more closely, and so the choice of junior supervisor may be more important.
This section is work in progress, sorry.
Generic fundings are often local to a specific area. Most of the options below are for area of the Université Paris-Saclay, where I was before. For all of those options, your future advisor needs to do most of the paperwork so you need to coordinate with them ahead of time.
Some generic funding sources have a website where prospective advisors post PhD topics looking for PhD students. However, the funding is rarely limited to those advisors who post a topic. Feel free to search for potential advisors by using academic search engines like Google Scholar or by browsing University websites, lab websites or team websites. You can contact researchers directly by e-mail, and ask them whether they'd be interested in supervising you. Researchers love to be contacted directly, but hate generic e-mails. Look carefully at what the researcher has done and think carefully about how it overlaps with your areas of interest.