Appendix: Building a Record · The Machinery of Compliance

This appendix is not new argument. It distills, into something you can actually use, the discipline that Chapter 49 lays out in full: how a pattern built to be deniable gets turned into a record that survives a skeptic. Every point below is documented in that chapter. If you are reading this because you need it, take only what you can carry today.

The one thing that changes everything

Coercive control is a pattern, not an incident. Any single act — a comment, a transferred bill, a checked phone — looks minor or even legal on its own. The harm is only visible in the accumulation. So the record you build is a timeline, not a highlight reel. Evidence assembled one incident at a time will always look weak, because it was never meant to be read one incident at a time.

Before anything else: safety

Leaving is the most dangerous moment, not the safe one. Separation is the period of highest risk, not its resolution.
But “dangerous” is not one setting, and knowing which one you are in matters more than any note you will ever write. Separation is a single item on a validated instrument that exists to tell apart two frightened people who are not in the same danger — and it does it the way this book does everything, on markers a stranger can check rather than on how it feels from inside. If you have ever been strangled, start there. In the case-control research behind the instrument (Glass, Laughon and Campbell, Journal of Emergency Medicine, 2008), women who had been strangled by a partner and survived carried roughly seven and a half times the odds of being murdered by him later, compared with abused women who had not been — and it counts even if it happened once, even briefly, even if it left no mark. England and Wales made non-fatal strangulation a crime in its own right in 2022, needing no proof of injury, for exactly this reason. An abuser’s access to a firearm, a previous threat with a weapon, threats to kill, escalating frequency or severity, stalking, forced sex, the other person’s own threats of suicide, and violence during pregnancy all carry weight too. If any of these is in your history, say so to an advocate today, before the record and before the plan. The assessment is free and takes minutes, and it is the one thing on this page that is about staying alive rather than being believed.
For that reason, documenting and leaving run at the same time, and covertly — never “leave first, then document.” The surveillance is often still live, and often most alert, exactly when departure is being planned.
If you are in acute danger, documentation is not an obligation you owe anyone. Safety comes first; the record can wait. Treating someone in danger as though they were already a witness building a case file is its own harm.
A pet is a safety-planning factor, not a sentimental aside. Threatening or harming an animal is a documented coercive-control tactic — a large majority of pet-owning women entering shelters report the abuser injured, killed, or threatened a pet — and it works as a leash: between eighteen and forty-eight percent of victims delay leaving, or return, out of fear for an animal left behind, because most domestic-violence shelters cannot house one (a review of twelve studies; the pattern traces to Frank Ascione’s foundational research). Plan for the pet as you plan for yourself. Safe-haven and fostering programs — run through humane societies, veterinary offices, and a growing number of shelters — exist for exactly this, and an advocate can find the nearest; arrange it before you go, because “I couldn’t leave the animal” is one of the most common and most preventable reasons a plan fails.
Good domestic-violence services build evidence preservation directly into safety planning rather than treating them as two separate jobs — which is the argument of this whole appendix, arrived at independently by the people who do this for a living. Ask yours to do both at once. The risk instrument described above is Jacquelyn Campbell’s Danger Assessment (dangerassessment.org); an advocate will either use it or use the local equivalent. Find your country’s service through the directories in the “Resources and Help” appendix.

The four principles — this is the checklist

Contemporaneous beats remembered. Write it the day it happens: date, time, exact words. A note made in the moment is evidence. The same account reconstructed six months later is testimony — weaker, and directly attackable, because trauma-affected memory compresses and reorders under stress.
Corroboration beats one witness. A text sent to a friend in real time, a neighbor who heard something, a medical or financial record created for an unrelated reason that happens to note the fact. Anything that moves the record out of the two-person, he-said-she-said frame, which is the frame a denial-and-reversal is built to win.
Originals beat summaries. Keep the original artifact and its metadata, not a paraphrase. Export and back up before you lose access — shared devices and joint accounts are exactly how evidence disappears before a person realizes it needs preserving. Store copies with an independent person, outside the household.
Write for a reader who does not believe you. The register is an incident log, not a diary. Interpretation, editorializing, and emotional processing — the things that make a private journal useful — are exactly the things that make a record easy to attack.

Corroboration you can build on purpose

Principle 2 is usually described as something you find — a friend who happened to be texted, a record created for an unrelated reason. But you can also create corroboration deliberately, and it is one of the most useful things you can do while still inside the situation:

Tell one trusted person, in a dated message. Not a vent — a plain, time-stamped account of what happened. It moves the fact out of your own head and onto a second, independent record with its own metadata.
See a doctor. Stress, sleeplessness, an injury, a panic response noted in a neutral medical file becomes a dated, third-party record made by someone with no stake in your dispute — which is exactly the kind of record a later reader weighs most heavily.
Use work channels where they exist. Raising something with HR, or an employee-assistance program, generates an institutional record outside the household.

The value is the same in every case: an independent party, holding a dated record you did not author alone, is the hardest thing for a denial-and-reversal to explain away. Do this the same careful way as everything else here — from a safe device, aware that some of these steps can be noticed, and timed with an advocate where the danger is high.

The incident-log format

For each entry, record only:

Date and time · what was directly observed or said, in plain words · who else was present · what happened immediately after.

Keep interpretation, feeling, and your own sense of the pattern in a separate, clearly labeled companion record. It is genuinely useful — for your own processing, and for eventually seeing the shape of the whole — but it is kept out of the evidentiary log so that it cannot be used to make the log itself look unreliable.

Digital evidence, specifically

What you see on your own phone is a curated view. Content can vanish from that view while still existing intact in the database underneath it, and things can appear there that you never wrote. This means:

The device alone proves nothing for either side — deletion, fabrication, and innocence all look the same from the screen.
What settles it is provenance at the back end: server logs, message metadata, timestamps, account records. Preserve the original and its metadata together, and keep an independent copy off the shared system.
Do this on a device the other person cannot see or reach — a friend’s phone, a library or workplace computer — never on a shared or monitored phone, and never into a cloud account they may control. The surveillance described in this book is often still live, and most alert, during separation; the act of copying, screenshotting, or backing up can itself trigger an alert. Export before access is lost — but never at the cost of revealing that you are building a record.

A note on recording

Recording can be strong evidence — or a criminal charge against you. Which one it is turns on a single question: were you a party to what you recorded?

Record only what you are part of, or present for — a call you are on, a conversation happening to you, an incident in front of you. In most places this is lawful, and it can be powerful.
Never plant a device to capture others. A hidden recorder left running, a phone left behind, a bug, a camera aimed into someone else’s private space — capturing people when you are not there, or a conversation you are not part of, is the classic wiretap crime. It is illegal even when a survivor does it to gather evidence, it usually cannot be used in court, and it can be turned against you.
Assume your intuition is wrong here, because it usually is. Almost everyone believes that a conversation they are part of is theirs to record. In many places that is exactly right. In others it is a crime — and, in some countries, the answer changes from one state or province to the next inside the same border, so “it’s legal in my country” is not an answer either. There is no rule of thumb that survives the trip, and the belief that there obviously must be one is the specific thing that gets people charged. Find out what the law is where you are, before you record anything. A domestic-violence advocate or a lawyer can tell you in a single conversation, and it is free.
Your own jurisdiction may not be the only one that counts. If the other person is somewhere else when the call happens, their rule may reach you where you sit — some places apply their own consent rule to a caller outside them, so the stricter of the two can end up governing. Check both ends of the call, not only the end you are on, and do not assume your own law is the only one in play.
Recording your child’s calls with the other parent is its own question, and a dangerous one. Whether a parent can consent on a young child’s behalf — the argument is usually called vicarious consent — is answered differently in different places and is not settled anywhere. Do not assume that being the parent, or being the protective one, makes it lawful. And getting it wrong is not only a possible crime: in a custody case it hands the other side the exact story the chapter on the family court describes — the parent seeking protection, recast as the one doing the surveilling. This is the point in this appendix where the gap between lawful and wise is widest, and it is the one to take to a lawyer in your own jurisdiction before you act, not after.
Cameras follow different rules again. Filming a space where a person reasonably expects privacy — a bathroom, a bedroom, inside someone else’s home — is voyeurism, a crime, whatever the reason; and adding audio to any camera re-triggers the recording-consent rules above.
Admissibility is never guaranteed, and safety comes first. Even a lawful recording may or may not be admitted — that is a judge’s call, jurisdiction by jurisdiction. And a recording the other person discovers can escalate the danger the whole effort was meant to reduce. Preserve the original and its metadata, keep it somewhere they cannot reach, and never let the act of recording reveal what you are doing.

The principle underneath all of it is the one this book turns on its own material: document what is done to you — do not become the surveillance you are documenting.

Why any of this works — one line

A contemporaneous record defeats a denial not because of what it says but because of when it was made. It existed before the confrontation that will try to discredit it, which is the one thing memory recalled under pressure can never claim for itself. That is the whole of it.

When you did not document it in time

Most people find a book like this in the middle, not at the start — months or years already behind them, and little written down. The record can still be rebuilt, because the operation left traces you did not make on purpose. Bank and phone records establish a timeline you never kept. Emails and messages carry their own dates in their metadata. Medical and employment files noted things contemporaneously, for unrelated reasons. Third parties remember. The task is the one this whole book describes — reconstruct the sequence from what could not help but be left behind — turned on your own past: gather the dated traces, assemble them into the timeline, and let the timeline show the pattern that no single trace does. A gap in your own notes is not the end of a record. It is the start of a reconstruction.

The discipline this turns on you

The rigor that makes a record un-attackable has to be turned on your own conclusions too, or it stops protecting you. Three rules:

Reconstruct; do not retaliate. The purpose of a record is recognition and legitimate recourse — a lawyer, a regulator, a court, a safety plan — not running the tactics back at anyone. The moment you act on the pattern instead of documenting it, you hand your credibility away.
Make each layer earn its place. For any specific event, ask whether it is truly the operation or whether it is ordinary incompetence, coincidence, or someone simply behaving badly — and require evidence before you count it. What is real survives the question easily; this is what keeps a record from being dismissed as a theory.
Count layers; do not assume them. Most harm is a single layer. Resist the pull to read one bad event as proof of a coordinated whole. The discipline that protects you from being disbelieved is the same one that protects you from seeing a pattern that is not there — and both failures end the same way: with no one able to act on what you have.

The tools here are the same whether the record belongs to a person leaving a marriage, a nurse exposing a hospital, or an analyst facing a state. Only the scale of what they are used against, and how long the record has to survive before someone is willing to read it, ever changes.