Bring your conversation
Upload a TXT or CSV export, or paste messages directly. You stay in control.
MEASURED · NEVER PREDICTEDBUILT ON 33 PEER-REVIEWED STUDIES
Private by design
Am I getting
dumped?
Here’s what actually changed in your conversations.
Ten things researchers actually measure in how couples talk — reply timing, curiosity, warmth, shared planning — drawn from peer-reviewed studies of real relationships and read from your own thread. Patterns, not verdicts.
Protected in transitNo training on your dataDeletion controls included
Am I Getting Dumped? identifies patterns—not intentions, diagnoses, or predictions.
READOUT · BEFORE → NOW12 WK WINDOW
MEDIAN REPLY GAP · 4 h BASELINE → 31 h CURRENT · Δ +27 h OVER 12 WK
Your communication rhythmLast 12 weeks · May · June · July
Curiosity has become less frequentQuestions made up 18% of earlier messages and 11% recently.
“I miss feeling curious about each other. What would help us make room for that again?”
Worth discussing · +24% More shared planning · Balanced Conversation startsSAMPLE · 1,240 MESSAGES · 10 MEASURED CONSTRUCTS · 0 PREDICTIONS
Am I Getting Dumped? notices patterns.
It never predicts your relationship.
01 Measured, not assumed
02 Neutral, not judgmental
03 Questions, not verdicts
How it works
Clarity in three gentle steps.
We map the patterns
Am I Getting Dumped? measures rhythms, responses, tone, curiosity, affection, and change over time.
Reflect, then talk
Receive a calm, evidence-based report and thoughtful questions for a real conversation.
The evidence
Every pattern we show you comes from published research.
Not a personality quiz. Not an algorithm we invented. Ten measured constructs, each one drawn from peer-reviewed work on how couples actually communicate — and each one shown to you with the paper it came from.
33peer-reviewed papers
10measured constructs
0predictions made
How much can language actually tell you?
No one can predict a breakup. Not us, not anyone.
The largest review of couple communication and breakups found the link is real but small. Using all of it, you would guess right 61 times in 100 instead of 50. That is the best science on earth — and it is still not a crystal ball. Anyone selling you one is lying.
So we don't predict. We measure. What changed in your conversations — reply times, who reaches out first, which words disappeared — isn't a guess. It's already in the thread.
Kanter et al. (2021), meta-analysis of couple communication and dissolution: d = −0.41, expressed here as a common-language effect size. Positive communication alone was not a significant predictor of dissolution.
Positive against negative markers, tracked as change — never against a “magic ratio.”
Warmth, humor, appreciation and interest in ordinary conversation, not conflict.
Function-word alignment between two people, where decline is the meaningful direction.
We-talk, I-talk and you-talk — reported as asymmetry, never scored as “more we is better.”
Criticism, contempt, defensiveness and stonewalling, each surfaced separately.
How good news gets answered, across the four established response types.
Questions, elaboration and topic follow-up in reply pairs.
A sequence, not a trait — and one that context can make adaptive rather than harmful.
The multi-feature shift observed roughly three months before a breakup.
Assurances, shared plans, positivity and social-network references.
What it will never do
- Give you a breakup probability, a risk percentage, or a score out of 100
- Quote the famous “83–93% accurate” figures — they came from models fitted on the same couples they described
- Diagnose either of you, or infer intentions, feelings, or cheating
The strongest known predictors of a breakup are commitment, love, and how much you include the other person in your sense of self. Text cannot observe any of them. What text can show is what changed — and that is what you get.