How long to give a new method before you judge it
You have tried three approaches in a fortnight and none of them worked. That is very likely why none of them worked. Two nights of data is not data.
3 min read
Babies have bad nights for no discoverable reason. A tooth moving, a bit of wind, a dream, a slightly warm room, a day with more stimulation than usual. Night-to-night variation is large, and it does not politely stand aside while you run an experiment.
On top of that, the first two nights of any change carry an extinction burst - an increase in protest as your baby tests whether the old arrangement still applies. So the first two nights are not merely uninformative. They are systematically the worst nights, and judging on them guarantees you abandon things that were about to work.
What counts as progress
Shorter, not fewer. The length of each settle or wake drops before the number does. Watch the minutes, not the count.
A better trend, not a better night. Forty minutes, thirty-five, fifty, twenty-five is progress. Any single night in that sequence is noise.
Bedtime before nights. Bedtime almost always improves first. Night wakes often resolve on their own a week or two later.
Calmer protest. Grumbling where there used to be crying is a real change, even if the duration is identical.
The timeframes
Four nights is the minimum before drawing any conclusion. Seven is better. Gradual methods need three to five weeks overall, with three or four nights per step. Faster methods often show clearly within a week.
Write it down, because memory is actively unhelpful here - a bad night overwrites the three good ones before it, and by the end of the week you are certain nothing has changed when the numbers say otherwise.
The first two nights are the worst by design. Judging on them means abandoning everything that was working.
When it genuinely is not working
After two weeks of consistent application with no trend in any measure, something else is wrong. Usually the timing: wake windows or bedtime that do not fit. Occasionally the method is a poor fit for the age - pick up put down on a ten-month-old, for instance.
Check those two before changing approach. Switching method while the schedule is wrong just produces another failed fortnight.
What to actually write down
Four things, every night, in a notebook by the bed. What time they went down. How long until they were asleep. How many times they woke. How long the longest wake was. It takes under a minute and it survives the memory distortion that makes every fortnight feel identical.
The reason it matters so much here is that improvement in the early weeks is almost entirely in the durations, not the counts. Without numbers you will see four wakes on night one and four wakes on night ten and conclude nothing has changed - when the total time awake may have halved.
Why the first two nights get worse
When a reliable arrangement changes, the first response is to try harder at the old one. Behaviourally this is called an extinction burst, and it is entirely predictable: more protest, for longer, before anything improves.
Knowing it is coming changes how you read it. A worse night one and two is the method working as designed rather than evidence against it. It also explains why abandoning halfway is so costly - a baby who protests harder and then gets the old arrangement back has learned exactly what escalation is worth, and the next attempt starts from a worse position.
When to stop troubleshooting
Stop immediately if your baby is unwell, in pain or feverish, and resume when they are better. If several weeks of consistent effort with correct timing produce nothing, or there is snoring, mouth breathing or pauses in breathing, speak to your GP or health visitor.
Want this worked out for your baby instead?
The quiz takes three minutes and asks about your nights, not just their age.
Take the 3-min quiz →General guidance, not medical advice. Your paediatrician's advice takes precedence.
