Earned Secure Attachment: A 90-Day Plan

Before day 1: two things you fix in advance

The start date, and the thing you are going to count.

Two decisions, both made before anything starts, because both are easy to skip and neither can be added later. First the start date, written as an actual date inside the next seven days rather than as next month. Second the measure. It has to be something a person watching you could count without asking how you felt, which rules out calmer, better and more secure. Unmeasured plans do not fail loudly. They fade, and around week six you find you cannot tell whether anything moved, which is when most people quietly stop.

Days 1-30: the asking block

One behaviour: ask plainly, and keep the count.

The behaviour is asking for something plainly and the measure is how many times. Small and frequent beats large and rare by a wide margin: five trivial asks in a week does more than one significant one, because what you are building is a default rather than an achievement. Pick things where the answer barely matters, so the practice sits in the asking rather than in the outcome. The usual failure in this block is making one important ask, having it go badly, and treating that as the result of the experiment instead of as one entry in it.

Days 31-60: the staying block

One behaviour: hold a fixed delay where you usually move.

Everybody has a move at the moment discomfort peaks. Some people leave the conversation, some fill the silence, and most can name theirs in one sentence once asked. The behaviour for this block is a fixed delay at that exact moment: ten more minutes in the room, or two hours before the follow-up message. Not a resolution, a delay. Count the times you held it and the times you did not, and keep both figures, because the ratio is the interesting part. Keep the delay short enough to repeat. Heroic delays get held once and then abandoned.

Days 61-90: the repair block

One behaviour: go back inside a day, in two sentences.

Repair goes last because it needs the other two blocks underneath it. The commitment is narrow: any rupture gets a return inside twenty-four hours, and the return is two sentences rather than a case. Count ruptures, count returns. The usual failure is not forgetting. It is waiting to see whether they go first, which feels like patience and functions as a policy. If you notice you are keeping score of who returned last, write that down as well, because it tends to be the most informative line anywhere in the block.

Day 90: the review, and the honest bit

Three counts, one question, and a decision about the next block.

Put the three figures next to their targets, then ask the question the figures cannot answer: what would somebody close to you say is different? If the honest answer is nothing, that is worth knowing rather than worth explaining away, and the most common reason is a plan run alone inside a situation that does not respond to it. Ninety days is a working unit and not a course of treatment. What it buys you is a handful of behaviours that no longer need a decision, which is modest and is the part that turns out to matter.

About the timescale, honestly

Ninety days will not produce a new pattern and no plan that claims otherwise is describing the research. Attachment patterns are moderately stable, and the movement that has been documented happens across years, mostly through relationships and through therapy rather than through exercises done alone at a kitchen table. What a plan can do is narrower and still worth having: it makes two or three behaviours automatic enough that they are available to you under pressure, which is exactly when they are otherwise unavailable.

The parts that make it stick are unremarkable. One behaviour at a time. Dates written before you start. A measure somebody else could count. And an answer, written in advance, for the week you skip. Plans of this kind almost never die on a bad week. They die on the decision afterwards to write off the whole thing.