How to track changes in someone's account of events
People's descriptions of the same event shift over time. Sometimes this is normal - memory is imperfect, and details fade or get reordered as time passes. Other times, the changes are more significant. Tracking how someone's account evolves, using their own messages as the record, can reveal whether the shifts follow predictable patterns.
Why accounts change
Before looking at how to track changes, it helps to understand why accounts change in the first place. There are several reasons, and they are not all equally concerning.
Normal memory variation includes losing peripheral details, compressing timelines, and blending similar events together. Someone who says "it happened on Tuesday" in one message and "I think it was Wednesday" a month later is probably just remembering imperfectly.
More significant changes include shifting who initiated an event, changing the emotional framing of what happened, adding details that were not mentioned before, or removing details that were included in earlier accounts. When the core narrative changes - not the peripheral details - that is worth examining more carefully.
Setting up a comparison log
To track changes systematically, you need a structured way to compare versions. A simple format works:
Event being described: [one-sentence summary]
| Date of account | Source | Key details | Notable differences from earlier versions |
|---|---|---|---|
| March 2, text message | "I told them I wasn't comfortable and they backed off" | First version - baseline | |
| March 15, text message | "I told them to stop and they kept going" | "Backed off" changed to "kept going." "Wasn't comfortable" changed to "stop." | |
| April 3, text message | "I never said anything because I was scared" | Now claims they did not speak up at all |
The comparison table makes the evolution visible. Without it, you might notice that something feels different about how someone is describing an event but not be able to articulate what changed.
What to compare
Focus your comparison on the structural elements of the account:
- Who did what. Does the description of actions change? Does someone who was passive in version one become active in version three?
- Sequence of events. Does the order shift? Do cause and effect get rearranged?
- Presence and absence of details. Do new details appear that were not in the original account? Do previously mentioned details disappear?
- Emotional framing. Does the same event get reframed from neutral to threatening, or from concerning to harmless?
- Other people's roles. Do bystanders get added or removed? Does someone who was present in the first version disappear from later ones?
When inconsistencies are meaningful
Not all inconsistencies carry equal weight. The question is whether the changes alter the meaning of what happened or serve a pattern.
Minor inconsistencies - times, dates, the color of someone's shirt, who said which specific word - are normal and expected. Major inconsistencies - who initiated an action, whether consent was given, whether someone was present, whether something happened at all - are significant.
Look for directional consistency. If someone's account of an event gradually shifts in a direction that serves their interest in a dispute, that is different from random memory variation. Random variation goes in all directions. Motivated revision tends to move in one.
Also look at what triggers the revision. Does the account change after the person learns new information about what you know? Does it shift after a confrontation? Does it change to match a new audience? The timing of revisions is data.
Being fair with this analysis
Tracking account changes is a powerful documentation tool, but it requires honesty. You have to be willing to find that the inconsistencies are minor and meaningless. You have to resist the temptation to treat every small variation as evidence of deception.
Apply the same standard to your own accounts. If you described the same event at two different times, would your versions match perfectly? Probably not. The goal is not to prove that someone is lying. The goal is to see whether the pattern of changes is consistent with normal memory or with something else.
When you present this kind of analysis - to a lawyer, a mediator, a therapist, or just to yourself - present all the versions, including the ones that do not support your interpretation. Let the record be complete. A partial record that only shows the damaging inconsistencies is cherry-picking, and it weakens your credibility if the other versions surface later. The full record is always stronger than a curated one.