Candidates spend weeks on a personal statement and ten minutes on the references, which is the wrong way round given how often the reference is what fails.
It fails in predictable ways: submitted late, or not at all; written by someone senior who barely knows the candidate; or generic, because the referee had nothing specific to work with. All three are things the applicant controls.
Choose someone who knows your work, not someone impressive
The instinct is to ask the most senior person who will say yes. A head of department who taught you once produces a letter that says you were a good student, which is what a letter from someone who does not know you looks like.
A lecturer who supervised your project can describe how you handled a problem when it went wrong. That is worth far more, because it is evidence rather than endorsement, and panels can tell the difference immediately.
Ask properly, and in person if you can
An email asking for a reference by Friday, from a student the lecturer has not spoken to in a year, is easy to leave unanswered.
Ask in person or on a call where possible, explain what you are applying for and why, and ask directly whether they feel able to write a strong reference. That last phrasing gives them a graceful way to decline, which is a much better outcome than a lukewarm letter you never see.
Ask someone who supervised your actual work, not the most senior name available. Give them weeks, not days. Send a brief with the award, the deadline, your CV and a reminder of what you did with them. Then confirm it was submitted — do not assume.
Give them a brief
Referees write better letters with material in front of them. Send, in one message: what you are applying for and the link, the deadline and how to submit, your CV, your draft statement, and a short reminder of the work you did with them and roughly when.
This is not presumptuous; it is the normal courtesy, and referees who write many letters expect it. It is also the difference between a letter that names your project and one that says you were conscientious.
| What goes wrong | Why | The fix |
|---|---|---|
| Reference never submitted | Asked once, never followed up | Confirm receipt with the funder yourself |
| Generic letter | Referee had nothing to work from | Send a brief with the request |
| Referee barely knows you | Chosen for seniority | Choose for proximity to your work |
| Rushed and late | Days of notice, not weeks | Ask as soon as you decide to apply |
| Wrong kind of referee | Requirement not read | Check whether academic or professional is required |
Notice is the whole game
Referees write letters between marking, teaching and their own work. A request arriving weeks ahead gets a considered letter; one arriving on the deadline gets whatever can be produced in the time available, if anything.
Ask as soon as you decide to apply, even before your own draft is finished. Nothing about the request depends on your statement being ready.
Check what kind of referee is required
Some awards require an academic referee, some a professional or employer one, some one of each. Some specify that referees must hold a particular position or must not be related to you.
A perfectly good letter from the wrong category of referee is invalid, and it is usually discovered too late to replace. Read the requirement before choosing, not after.
Line your referees up so they do not overlap
Where an application takes two references, the pair should cover different ground. Two lecturers from the same department describing the same coursework tell the panel one thing twice.
A supervisor who can speak to your academic work and an employer or project lead who can speak to how you operate outside it give the panel two independent readings. If you can choose, choose for coverage.
Follow up, and thank them
Confirm with the funder that the reference arrived rather than assuming it did. A polite reminder to a referee a week before the deadline is expected, not rude, and it rescues a meaningful number of applications every cycle.
Afterwards, tell them the outcome either way. Referees rarely find out what happened, most of them would like to, and you will very likely need them again.
