Applications
How to ask for a reference
In short
Ask six weeks ahead, in person or by a specific email, and send an evidence pack: the programme, the deadline, your CV, what you did in their class or lab, and what you would like the letter to be able to say. A referee given nothing writes a letter that says nothing.
Weak reference letters are usually not the referee's fault. They are the predictable output of a two-line request sent three days before a deadline.
The letter you get is a function of what you supply, and this is what to supply.
Who to ask, which is not who is most senior
Ask the person who saw you do something specific. A lecturer who supervised your project and watched you rebuild it after it failed can write two concrete paragraphs. A department head who taught you in a hall of two hundred can write that you attended and did well.
Seniority helps only when it comes with specifics. Given a choice between a professor who barely knows you and an assistant professor who supervised you closely, take the second.
When to ask
Six weeks before the earliest deadline, and earlier if the portal emails a form rather than accepting an upload. Academics have grading cycles and conference deadlines, and a request that arrives in the wrong fortnight gets a rushed letter or a decline.
Ask once, clearly, and then send one reminder a week before the deadline. Two reminders is fine; five makes the letter about your anxiety.
The evidence pack
This is the part that changes the letter, and almost nobody sends it. Attach it to the request rather than promising it later.
- The programmes, with links and deadlines, and the submission mechanism for each.
- Your CV and your statement, so the letter can corroborate rather than contradict.
- A short note on what you did in their course or lab: the project, your specific part, what went wrong and what you did about it.
- What you would like the letter to be able to speak to, framed as a suggestion rather than a script.
The awkward part: being asked to draft it yourself
It is common, particularly in India, and it is not automatically improper. What makes it acceptable is that the referee reads, edits and signs it, and that every claim is something they can verify.
What is not acceptable is submitting a letter a named referee has never seen. That is misrepresentation and it is treated as such. If you are asked to draft, write it in their voice, keep it to what they actually observed, and send it as a starting point rather than a finished document.
Giving them a way to say no
Include one line: if you do not feel you can write a strong letter, I completely understand. This feels like inviting rejection and it is the most useful sentence in the email.
A referee who agrees reluctantly writes a lukewarm letter, and a lukewarm letter from a credible source does more damage than a warm one from a less senior one. Giving them the exit gets you a better letter or a faster no, and both are wins.
Related questions
How many references do I need?
Two for most taught master's programmes, three for US graduate programmes and most PhD applications. Send exactly what is asked for; an extra letter is not a bonus and can read as not following instructions.
Can I use a workplace reference instead of an academic one?
Where the programme allows it, and it is often a better choice for applicants several years out of study. Check the wording: some programmes require at least one academic referee regardless of how long ago you graduated.
What if my referee misses the deadline?
Contact the admissions office rather than panicking. Many portals accept references shortly after the applicant deadline, because universities know this depends on a third party. Ask early rather than after the date passes.