How Situationship Autopsy measures a conversation — and what it can’t tell you

Messaging effort is not the same thing as feelings. Situationship Autopsy measures what a conversation shows — who writes, who starts, who asks, how fast replies come — and deliberately stops there.

Where the analysis happens

The export is read in your browser, in a background worker, and never uploaded. Wabiro, analytics and Stripe never receive messages or names. Nothing is sent to the other person. The same parser as Group Chat Court is used — see its method for how exports, dates, system lines and media placeholders are read.

The autopsy needs exactly two people and at least 30 messages. A group chat is routed to Group Chat Court instead.

What is measured, per person

• Messages, words, and questions (text messages containing a question mark).
• Conversations started: the first message after 4 hours without any message.
• Plan vocabulary: messages containing words from a fixed list (tomorrow, tonight, this weekend, coffee, dinner, drinks, meet up… and their French, German and Spanish equivalents). It is a keyword count — it does not know whether a plan happened.
• Follow-ups without a reply: writing again after 20 minutes or more with no reply in between.
• Restarts: writing first after 3 days or more of silence.
• Time of day, weekend share, consistency (weeks with at least one message ÷ weeks spanned).

Reply time is an approximation

Exports do not say which message answers which. When one person writes directly after the other inside the same conversation session, the time between the two messages counts as a reply time. Long silences start a new session and are not counted as slow replies. The result shows the median of these times per person, labelled approximate. “Replies faster” needs at least 5 measured replies on each side and a 1.5× difference.

Effort balance

For messages, conversations started and questions, Wabiro takes the share carried by one person and measures its distance from 50%: 1 − |share − 0.5| × 2. The effort balance is the weighted average (messages 40%, starts 40%, questions 20%), from 0 to 100. 100 means an exact 50/50 split on all three; it does not mean anything is better or worse.

Patterns

The first matching rule gives the pattern:

1. Burst & Disappear — export spans 21+ days, messages on fewer than 25% of days, 20+ messages per active day.
2. One-Sided Starter — 6+ conversations started and one person started 70% or more of them.
3. Mutual Yappers — 40+ messages per active day, messages split within 40–60%.
4. Weekend Conversation — 55% or more of messages on weekends.
5. Slow Burn — both medians of reply time are 60 minutes or more (5+ replies each) and messages on 30%+ of days.
6. Balanced Ping-Pong — messages within 40–60% and starts within 35–65%.
7. Uneven Split — messages outside 40–60%.
8. Steady Rhythm — everything else.

What it never says

Whether someone likes you, is interested, is lying or cheating, whether you should stay or leave, or whether a relationship will work. No field in the analysis can hold such a claim, and the copy is reviewed for it. The reflection prompts in the Full Autopsy are questions, not advice.

The Full Autopsy

The optional one-time unlock adds every dimension, month-by-month messages and starts, the reply-time trend, recovery after long silences, hour-by-hour rhythm, reflection prompts and a share-safe summary — all computed on your device. Payment is bound to a one-way fingerprint of aggregate numbers, never to chat content.

Versioning

Analysis version 1. Any change to a rule, keyword list or threshold creates a new version.