You can identify most at-risk students before the final exam by reading marks you already hold. Look for low assignment scores, missed deadlines and a downward trend across two or three pieces. Thin participation matters too. This method turns those signals into a short list you can act on in a single afternoon.
Acuity, Lectimax's cohort analytics module, reads your marks data and surfaces the same patterns automatically. It hands you the short list of students who need attention; acting on it is your call.
Which signals predict trouble?
Four signals consistently appear before marks collapse. They are sustained low scores, missed submission deadlines and a downward trend. Participation that falls away is the fourth. A single poor mark is noise; a student who shows two or more of these across the term is a real risk.
The order of the signals matters for how you act. Sustained low scores say the student is underprepared or not engaging with the material. Missed deadlines say something else. They usually point to a load or motivation problem rather than a comprehension one. That is why a missed deadline and a low score together call for a different conversation than either one alone. A downward trend is the most urgent. An early strong student sliding across the term is either losing interest or hitting an unstated problem. Both are easier to fix in week eight than in week twelve. Each signal changes what you say and who you think you are talking to.
How do you build a simple risk model in a spreadsheet?
Give each signal a score of zero, one, two or three and add them for every student. Sort the list and read the top. A student scoring high on missed deadlines and downward trend needs a conversation this week, not a warning at the exam. This model runs in an afternoon with a column for each signal.
The scoring is deliberately coarse, because the point is ranking, not precision. A zero to three score per signal keeps you from agonising over the difference between a 47 and a 49 when both are simply low. Sum the four columns and sort descending, and you have an ordered list in the time it takes to enter the scores. Do this across a spreadsheet of the whole cohort and the top of the list is where your afternoon goes. The scoring only ranks the list; the conversation still belongs to you.
Read the pattern, not the single mark
One bad week happens to anyone. The reliable warning is a student who misses a deadline and then slides again across the next pieces of marked work. Read the pattern across time instead of reacting to one score in isolation.
The single mark is the trap. A student who bombs one test for a flu or a family crisis is not at risk, and flagging them wastes attention and can feel like surveillance. A student who misses a deadline, lands a low mark, and then goes quiet in class is showing a shape that repeats. If you react to every dip, you become noise. If you only react to patterns, students hear the signal when it comes.
Act early with a targeted list
Turn the top of the list into a short outreach list for the week: an email, a check in, or an offer of a consultation. Early contact beats a late warning, and students usually respond when the message is specific about the pattern you noticed. Keep the tone about support, not surveillance.
The outreach is where the whole method pays or fails. A generic "let me know if you need help" that goes to the whole module is filed with the forty other emails and changes nothing. Write a specific message that names the pattern: "you have missed two deadlines and your last mark was below the class average". It reads as someone who noticed. That is what opens the conversation. Offer a concrete next step, like an office hour slot or a draft you can review. Keep it about getting work back on track rather than policing behaviour. Students on the list often already know they are struggling; the value is that you tell them early, in a way they can act on.
When the list grows, let software do the scan
A spreadsheet works for one module. When you teach two or more, the signals live in different places and the short list rebuilds slowly. Lectimax's cohort analytics reads participation, submissions and marks across modules at once and keeps the at-risk list current as new results land. See the full Lectimax feature set.
The spreadsheet is a fine first tool. Its limit matches any manual model: it only sees the columns you remember to fill and updates when you remember to update it. It dies the moment you teach more than one cohort. Software earns its place for a simpler reason: the data is scattered. When a student is slipping in your Python module but steady in your statistics module, the single module view misses it. The cross-module view catches the student who is quietly sliding everywhere at once.
FAQ
How early can you spot an at-risk student? The signals are visible as soon as two or three pieces of work have been marked, which is well before the final exam.
What counts as at risk? A student counts as at risk when they show a downward trend, missed deadlines or falling participation across several pieces. Any single signal alone is usually not enough; the combination is the reliable warning.
Do you need a data science team? No. A spreadsheet with a column per signal ranks your list in an afternoon, and Lectimax does the same scan across modules automatically.
How do I avoid sounding like surveillance? Keep it about a specific pattern and a concrete next step. Students respond to somebody who noticed and offered help, not to a monitor reporting behaviour.
Where to go next
Pair early identification with a marking flow that gives you the data in the first place. See what AI grading in higher education really is, or review Lectimax pricing and the full feature set.