Who you actually reached - people and interactions in your Instagram export
Your Instagram export tracks actions between you and specific people. This is the volume half: not who is on your lists, but how much you reached them, what kinds of interactions you had, and the honest limits on that count.
S1 is state. This is volume. The social graph shows who is on your lists right now. The interaction counts show who you actually reached - how often, and in what ways.
State vs volume
Relationships are a snapshot of ties. Interactions are the events that accumulate over time.
The vocabulary for "who I reached"
The per-person index uses a consistent vocabulary for kinds of interaction. These are the same kinds used when LMKFR builds a people mix, so definitions never swap between views.
Common kinds include: direct messages, group messages, reactions to messages, DM media; story reactions; polls, Q&A, notes you answered; saves, reposts you made; reels watched, stories viewed, posts viewed; searches for a person; opening their profile link; and relationship actions.
Per-person totals are arithmetic
Each person gets an interaction total that sums real event counts by kind. Relationship annotations (muted, close friends, blocked, pending) may be relationship actions, but when ranking engagement they are not mixed with events to inflate one at the expense of the other.
Conversation states
For 1:1 DMs, conversations can be classified by direction (never replied, one-way them, one-way me), size (single message, >= 200 messages), dormancy (>= 90 days since any message), content (media-only with no typed words) and who spoke last. Group chats use a separate, simpler classification. These are rules based on counts and timestamps, not guesses.
Honest limits
- The archive does not record who liked or commented on your posts. That is not present in Download Your Information for a typical account.
- Reels watched and stories viewed are tracked separately and must not be blended.
- Name-only identities may appear without a confirmed @handle. Events are real, but the handle is not verifiably an Instagram account from the export.
- History depth is limited to what Instagram included in your request.
A measured way to read it
Start from the person with the highest real count for the kind that matches your question (DMs if conversations; saves/reposts if collecting; reels/stories if consumption). Never let a single strength number hide the breakdown by kind.
Does the export show who liked my posts?
No, for a typical Download Your Information archive the export does not list other people who liked your content. Only events you took are consistently represented in the sections used here.
What is the difference between watched reels and stories viewed?
They are counted separately in the export and are not added together.
Can a DM be classified without guessing?
Yes. For 1:1 chats the classification uses message direction, count, last speaker, days since last message and whether any typed words exist. Group chats are treated separately.
Do relationship actions count towards interaction volume?
They are real actions but are treated as relationship actions rather than engagement events. When ranking who you reached by speaking/reacting/consuming/collecting, they are generally kept out of that engagement total.
Sources of truth
packages/shared/src/types/analytics.ts-UserIndexEntry(kinds,interactionCount,dmMessages,groupMessages,handleConfirmed,storiesViewed,reelsWatched).packages/shared/src/analytics/user-index.ts-INTERACTION_LABELS.packages/shared/src/analytics/people-mix.ts-buildPeopleMix,PERSON_KIND_GROUP,unmappedPersonKinds.packages/shared/src/analytics/engagement-intel.ts-EngagementIntel.packages/shared/src/analytics/conv-states.ts- states and thresholds (>=90 days dormant, >=200 = high-volume).