What this is built on.

The product’s spine is Amy Chan’s Unsingle: How to Date Smarter and Create Love That Lasts (Abrams, April 2026). Chan’s framework is the sequence Stedi walks you through: name what’s non-negotiable, surface the type pattern you keep falling into, run a structured nine-date experiment to interrupt it, then make a call. Her central thesis: most singles aren’t broken, they’re stuck in patterns, and most people are one or two shifts away from a healthy relationship. The reframe of every date as data collection about yourself, not a search for The One, is hers.

Trade book, not peer-reviewed. The underlying research the framework draws on (below) is solid. Stedi is independent and not officially endorsed.

The four predictive markers we score after every date come from John and Julie Gottman’s decades of work on what predicts relationship stability. Their longitudinal couples studies consistently show that the way partners handle conflict (especially the “four horsemen”: criticism, defensiveness, contempt, stonewalling), the ratio of positive to negative interactions, and the pattern of “bids for connection” predict long-term outcomes far more reliably than chemistry on first dates.

Strength: very strong, with replicated longitudinal data on committed couples. Caveat: most Gottman research is on couples already in committed relationships. We’re applying it to dating, which is informed inference, not direct evidence.

The Date Card asks separately about Pull (magnetism) and Familiar (does this person feel like someone you’ve dated before). When those two correlate strongly, that’s the attachment-theory pattern: an anxious-avoidant pairing produces a high-arousal dopamine-anxiety loop that secure attachment doesn’t. People often misread the loop as chemistry.

Strength: well-established. Bowlby’s original work, Mary Ainsworth’s Strange Situation studies, Mikulincer & Shaver’s adult-attachment research. Levine & Heller’s Attached is a good lay introduction.

The audit’s framing of every pattern with the survival logic that built it is straight from clinical care practices used by therapists. The principle: a behavior pattern that looks dysfunctional today usually emerged as an adaptive response to a real situation in the past. Naming the survival logic makes the pattern interruptible without inducing shame.

Strength: well-validated as a clinical practice. Polyvagal theory (Stephen Porges), Internal Family Systems (Richard Schwartz), Schema Therapy (Jeffrey Young) all converge on this framing.

The audit never tells you what your pattern is. It paraphrases what you said and asks what you make of it. That move is from motivational interviewing, developed by William Miller and Stephen Rollnick to help people change behaviors they’re ambivalent about. The evidence base is large and replicated across substance use, weight management, and clinical settings.

Strength: very strong, hundreds of clinical trials.

The math behind “dates 1-3 are calibration, the first date that beats them is the stop” is the secretary problem from optimal-stopping theory. With nine candidates, looking at the first ~37% (three) and committing to the next one that beats them all maximizes your probability of selecting the best candidate from the pool.

Strength: the math is provably optimal. Caveat: the assumptions don’t hold in real dating. The pool isn’t sealed, you can go back, preferences shift. We treat it as a structured mirror, not a literal instruction.

We have no efficacy data. No one has run a randomized trial showing that doing a Stedi experiment leads to better relationship outcomes than doing nothing. We’re building the outcome-tracking infrastructure to start gathering that data from users who opt in. In the meantime: the design draws on well-established research, applied to your specific data. The specific recommendations are informed inference, not certainty.

It is not therapy, not a clinical instrument, and not a substitute for professional help if you are in distress. The audit’s pattern-naming is hypothesis-generation, not diagnosis. If anything we surface lands hard, please find a qualified person to process it with. In the US you can reach 988 by call or text any time.