How do you set research goals and milestones that show when you're behind?

On 24 August one of our example lab's goals turns amber. Its target date is more than nine weeks away, and not one of its milestones is overdue.

That's the whole point of a goal with milestones: it tells you you're late while there's still time to act. Most research goals can't. They're written as intentions and checked on the deadline, which is a smoke alarm that goes off once the house has burnt down. Below is how the check works, the three goals one small biotech set for a year's work, and the places the rule gets it wrong.

What makes a research goal checkable?

An outcome, a target date, and milestones that count towards it. "Work on suspension adaptation" is an activity; you can do it forever. "Serum-free suspension HEK293, banked, by 30 October" is an outcome. On 30 October it's either in the freezer or it isn't.

The example lab is a small biotech moving its HEK293 cells from growing stuck to plastic, in medium with 10% serum, to growing in suspension in serum-free medium. It sets the goal on 1 June, and five milestones count towards it:

Milestone	Due
A1 In suspension at 5% serum	Fri 26 Jun
A2 Growing at 1% serum	Fri 24 Jul
A3 Growing at 0% serum (two passages at 90% viable or better)	Fri 14 Aug
A4 Three stable passages at 0%	Fri 4 Sep
A5 Adapted bank frozen, 20 vials	Fri 25 Sep

Each milestone is an event you can check at the bench, and each is a few weeks of work. Both matter, for reasons that come up below.

How do you know a goal is falling behind?

You compare the share of milestones completed with the share of time used. If 60% of the time between setting the goal and its target has gone, about 60% of the milestones should be done.

CCA Labs works the status out every time it's read, from the milestones that point at the goal. The rules, first match wins:

1. No milestones: Not Started.
2. Every milestone completed: On Track. A finished goal can't be late.
3. Target date passed with work outstanding: Off Track.
4. Progress more than 15 points behind the elapsed share: At Risk.
5. Otherwise On Track.

The 15 points are a tolerance. Time moves every day and milestones complete in steps, so with no slack a goal would turn amber a few days after every milestone and back to green at the next one. It's the same reason a thermostat doesn't fire the boiler at a tenth of a degree under the setpoint.

One goal through the summer

The goal runs 151 days, from 1 June to 30 October. A1 is done on 25 June and A2 on 22 July, each a day or two early: 40% by late July, ahead of the line.

Then the cells hit 0.5% serum. On 3 August viability is down to 82% and the cultures are clumping, which is a common place for HEK293 adaptation to stall. The lab holds at 0.5% with an anti-clumping agent. On 10 August the cell culture scientist and the head of R&D re-plan: A3, A4 and A5 each move three weeks, to 4 September, 25 September and 16 October. The goal's target stays 30 October.

On 31 August, day 91 of 151, the expected progress is 91 / 151 = 60%. Actual progress is still 2 / 5 = 40%. That's 20 points behind, so the goal is At Risk. It crossed the line around 24 August, when expected progress went past 55%. Had A3 landed on 14 August as first planned, progress would read 60% against 60, On Track.

That's deliberate, and it's the one habit I'd insist on. Move milestones when the plan changes. Move the goal's date only when the goal itself changes, after a conversation, because moving it stretches the timeline the whole check measures against.

What about the other two goals?

The programme has two more goals, and each shows a different limit of the rule.

"A 2 L process that repeats", set on 3 August with a target of 26 February 2027, has four milestones. The first, "Vessels and probes received", is done on 27 August. That's 25% against an expected 14%, so it's On Track, even though two of its later milestones moved at the same re-plan, by three weeks and two, because the first 2 L run needs the adapted cells. Early in a long goal, the check has little to say. The slip will show in December, not now.

"Adaptation method written up as an SOP", target 18 December, has no milestones at all. It reads Not Started, and it will go on reading Not Started past its target date too, because the first rule matches before the date is ever checked. When the programme counts goals on track it counts Not Started as on track, because an empty goal is unplanned rather than late. So the programme reports 2 goals on track out of 3. It'd report the same thing if nobody ever wrote a word of the SOP.

Where it goes wrong

Milestones aren't weighted. A1, three weeks of passaging, counts exactly as much as A5, freezing the bank. Progress moves in steps of 20 points. If one milestone is a day's work and the next is two months', the goal's progress means very little. Make them similar in size, or split the big ones.

The line is straight and the work isn't. The check assumes progress arrives evenly. Adaptation often doesn't: here the first serum steps were quick and the last one stalled. A straight line flatters the start and punishes the end. It's still the right default, because the alternative is a curve somebody drew to make their own goal look fine.

Blocked milestones don't change the colour. A blocked milestone shows up as something to decide about, not as a verdict on the goal. That's a sensible split, but it means a goal can read On Track with a blocked milestone in it.

No target date, no warning. A goal without a target date can never be behind. That's correct arithmetic and useless management.

How CCA Labs tracks goals and milestones

Goals live on a programme. Milestones live on projects, and each milestone has an optional Goal field: pick the programme goal it counts towards, which must be in the same programme. A goal stores no progress and no status of its own; both are worked out from those milestones whenever someone looks. A goal with no owner inherits the programme's owner and says so, which isn't the same as having no owner.

The goals page leads with Goal Attainment: milestones completed over milestones counted, across all goals, not an average of goal percentages, so a goal with one milestone can't outvote a goal with twenty. On this programme's own page on 31 August that's 3 of 9, 33%.

The table sorts the worst goals first. Needs a Decision lists what someone has to act on: here, that the programme is at risk with "1 of 3 goals off pace", and that the SOP goal "has no milestones". At the review it gets two: a draft by 20 November and a released SOP by 18 December.

You can set this up when you create your lab: a programme, its goals, and milestones on the projects that do the work. The planning tools are on the all-in-one lab page, and the budget behind this programme is in how to split one budget across programmes and projects.

A goal tells you that you're behind. It doesn't tell you which project, by how many days, or whether the date that moved was the right one to move. That's what a roadmap is for, and it's the next part: how to see which lab projects are slipping.