'Tis that time of year again. The time when grad students receive offers from multiple universities, and faculty are trying to attract the best possible younglings.
Mine is a large public R1 university, the state's flagship, with several top-10 ranked STEM departments and many in the top 20 in their respective fields (mine is ranked 14 or 15). So it's a very good school, but still not a FancyPants Private U. We almost invariably lose good candidates to private schools. I am in a department that does not offer many fellowships or teaching assistantships, so most financial offers to incoming graduate students have to be research assistantship offers, made by individual faculty. (For biomed readers: we don't do rotations.) At best, you meet some applicants at the open house (only those from a few neighboring states are invited in the case of my department), but I rarely find any students I like at these events, and those I do like typically have other attractive options.
The work I do requires a very strong math and physics background, which is often found in international candidates, so my group has a number of international students. With international students, it's really hard to recruit. You go by the BS school's reputation, letters of reference, test scores, talk to them on the phone, exchange emails. It's still hit or miss. For instance, from the same excellent international school I have had one stellar student and one complete dud, who both came in with very strong and very similar records. With international students you have less ability to do a thorough pre-screen because they are remote, so there is a greater chance for an advisor/advisee mismatch. On the other hand, the commitment to fund an international student is quite serious: if without funding, international students cannot pick up a McJob to support themselves or pay the tuition, they cannot always qualify for TA-ships because of language skills, and their dropping out of good standing as a student endangers their ability to stay in the country. If a student -- domestic or international -- doesn't work out, you have to suck it up and pay them until they find another advisor, but this aspect has added importance for international students who usually have absolutely no financial safety net.
I have also been trying to grow my own grad students, i.e., recruit them in my undergraduate courses and have them do research with my group for a summer or two, or at a lower pace over the semester. The good ones usually do it for the letters of recommendation and the fact that they got research experience, and then go on to greener pastures. And that's quite alright. Sometimes, though, you invest considerably more time in some students and they aren't very forthcoming about their plans: my most recent disappointment is a kid, an undergrad who was with the group for two summers and three semesters, coauthored a couple of papers (was one of the middle authors, did small, specific calculations in a large team). He sent a couple of grad school applications to private schools, but was saying the entire time that he planned on staying here for grad school because of family. I helped him write 3 fellowship applications (based on ideas in my larger grant proposals) to federal agencies to help support his studies here. Now he got admitted, with a lucrative first-year fellowship, to a FancyPants Private U, and has jumped on the offer like a kid into a pool of ice cream. I know that pedigree of the PhD institution matters, and that for his career the move is the right one. Still, I am disappointed as I spent a lot of time and energy on him, and some other place gets to cash in on a well-trained student, experienced in research, who may also end up getting a fellowship based on my grant proposal ideas. It's a great deal for the FancyPants Private U. (Ironically, their department is not more highly ranked than mine, but the overall name recognition trumps everything.)
The part that I hate the most is "the recruitment courtship." To stellar candidates, someone in a school like mine is never first choice; I know that very well and ask each potential candidate to tell me honestly what their first choices are, and that if they don't work out we can talk. But, many good candidates feel they need to pretend that my group is their first choice, so we go through this elaborate dance where they swear they are dead serious about coming here, make me issue RA offer letter and everything, only to yet again reveal that they were waiting for an offer from a more prestigious place the entire time. What I hate the most are the patronizing email apologies for not coming here, which come from feeling very smug because multiple places want them. *eyeroll* Whatevs, kid, happy trails.
I don't think there is a magic formula for recruiting good students while not at a FancyPants Private U (perhaps even if you are). When recruiting, I mostly go by the BS school's reputation and look for candidates with excellent undergraduate records and some research experience, but those who may be passed up by FancyPants Private U's because the record has a blemish (e.g., perhaps less than stellar GRE scores). My most recently hired student, who looks like he will be awesome, fits this recruitment pattern. Another way is going by recommendations: my best student so far looked nothing special on paper, but I hired him based on a recommendation from a trusted colleague, and it worked out great.
What is your recruitment strategy for graduate students -- especially if you have no benefit of testing the student out through rotations, but rather have to take him/her on, with pay, essentially sight unseen?