← All field notes
How-to

Running your first COIL project

Most of the work happens before a single student joins a call. This is the order we would tackle it in, and the four decisions that matter more than any of the tooling.

A first Collaborative Online International Learning project is usually won or lost in the weeks before it starts. Once students are in the room, what determines whether it works is whether two faculty agreed, in advance and in writing, on four fairly unglamorous questions. None of them are about software.

This is a general guide rather than an account of any one partnership. It is the pattern underneath most of them, drawn from published COIL practice and fifteen years of designing online international projects for students.

1. Agree one shared outcome, not two parallel syllabi

The most common first draft of a COIL project is two courses running alongside each other with some video calls bolted on. Students meet, exchange pleasantries, compare notes and go back to their own assignments. It is pleasant, everyone reports enjoying it, and it teaches considerably less than it appears to.

The difference between an exchange and a collaboration is whether something exists at the end that neither cohort could have produced alone.

AN EXCHANGE Two courses, a few calls between them. Each side still hands in its own work. A COLLABORATION one output Something neither cohort could have produced on its own.
The test is not how much contact there is. It is whether the contact is necessary to the output.

The practical version of this test: write one sentence describing what a mixed group will hand in. If that sentence still makes sense with only one institution’s students in the room, it is not a shared outcome yet.

Outcomes that hold up well:

What these have in common is friction. The partner is not a guest speaker or a cultural bonus; the partner holds half the material.

2. Map the calendars before anything else

Two academic terms almost never line up. Different start dates, reading weeks in different places, national holidays that fall nowhere near each other, exam periods that begin while the other institution is mid-term, and — across hemispheres — terms that barely overlap at all.

Institution A 14-week term Institution B 15-week term overlap: 8 weeks minus a reading week and two public holidays → 6 usable
Both terms are long. The window in which both cohorts are actually available is not.

Build the project on the overlap, not on either term. Six usable weeks is normal, and it is enough. What it is not is fourteen, which is what each side will assume if nobody does this exercise on a calendar with both sets of holidays marked on it.

Place the collaborative block in the middle of both terms. The start of term is lost to registration, late enrolment and students still finding the module. The end is lost to assessment, when the collaboration is the first thing a stressed student drops.

Then decide how much has to be live

Design asynchronous-first and treat every synchronous session as a scheduled exception. A useful default is one live session at the beginning — introductions genuinely work better face to face — and a second only if the shared outcome requires real negotiation.

Past roughly six hours of time difference there is no comfortable shared hour, only a choice about whose evening gets used. Announce live sessions weeks ahead, alternate which side is inconvenienced, and record everything.

3. Decide who grades what, in writing, before week one

In almost every case each institution grades its own students against its own criteria. This surprises people who expect joint marking, but joint marking is usually impossible — regulations, moderation processes and external examiners do not travel.

INSTITUTION A INSTITUTION B Participation own own The shared deliverable own students own students Individual reflection own own A student who stops contributing flag it flag it
The usual shape. What matters is less which arrangement you pick than that both sides wrote the same one down.

Four things need agreeing explicitly:

One page. Both of you sign it. It takes an hour and it is the difference between a disagreement in week five being an administrative question or a personal one.

4. The minimum viable tech stack — and what to ignore

Tooling consumes far more planning time than it deserves. You need three things:

The blocker almost nobody predicts is the second one. Your virtual learning environment probably cannot issue accounts to another institution’s students without an IT request that takes longer than the project runs. Assume this is true, check it early, and design around it rather than waiting.

Things to ignore on a first run: bespoke platforms, anything requiring procurement or a data protection review, anything students have to pay for, and immersive or VR tooling. All of it is a second-project problem.

Before you start, you should have

  • One sentence describing what a mixed group hands in
  • A shared window with real dates, after subtracting both sets of holidays
  • A one-page assessment agreement, signed by both sides
  • Three tools, all reachable without an institutional login
  • A named person on each side to contact when something breaks

Five things. If you have them, you are further along than most first partnerships, and the rest can be worked out once students are in the room.

This is a general guide and does not describe any single partnership. Companion piece: Where COIL partnerships tend to come unstuck.
More field notes