we read the research so you don’t have to trust us
Short answer: we collected and graded 331 research claims that circulate behind breakup-recovery apps. Not one earned an A— a replicated causal result in a breakup population. The entire treatment literature for romantic-breakup distress, 1992–2023, is 13 studies. Sixty-six claims were folklore, tested-and-null, or evidence of harm. And when we turned the audit on our own app, seven of our nine load-bearing claims failed — so we deleted them. This page is what survived.
the problem with this category’s numbers
Breakup apps and breakup content love statistics: success rates, “clinical studies show”, day counts at which your brain allegedly rewires. We have spent weeks trying to trace these to sources. The pattern, almost without exception: the number either has no source at all, or its source says something measurably different. A famous example from our own audit — the claim that a craving science review shows urges peaking at 2am traces to a mis-citation: the underlying studies found alcohol craving peaks around 8–9pm, in a model that can only produce one peak per day by construction.
The honest position is less quotable and more useful: almost everything every app in this category does — ours included — is borrowed from adjacent fields: smoking cessation, bereavement, insomnia, depression, loneliness, craving research. Borrowing is legitimate. Pretending it isn’t borrowed is the category’s besetting sin.
what the audit found
331 claims graded. A: 0 · B: 121 · C: 121 · D: 23 · F: 53 · harm: 13. Grade A meant a replicated causal result in a breakup population. Nothing qualified.
The whole treatment literature is 13 studies— five single-case, four quasi-experimental, four experimental, with samples from N=1 to N=152, and exactly one delivered online (Yanez-Peñuñuri et al. 2024, DOI 10.21500/19002386.6577).
Two interventional trials for breakup distress have ever been registered on ClinicalTrials.gov. Neither has reported.
66 of 331 claims came back as folklore, tested-and-null, or actively harmful. Almost every cheap, intuitive, obviously-good idea in this category has been tried, and several made people measurably worse.
what we deleted from our own app
The audit graded us too. Nine load-bearing evidence claims lived in this product; two held. The other seven were removed from every shipped surface, and we keep the list public because the deletions are the credential:
“Same brain regions as cocaine craving” — a reverse-inference fallacy (the same activation appears in people happily in love). Deleted from the flow where it fired mid-spiral.
“The urge peaks at night” stated as a research finding — unmeasured, see below. Now we describe the experience without inventing the statistic.
“Days 1–14 is where nearly everyone slips” — no evidence base in any population. Deleted from five places.
Two invented “most urges pass in twenty minutes” durations. The craving literature does not give you that number.
A tidy grief-wave sequence presented as science — the one intensive-longitudinal dataset points the other way, and the modal trajectory after a breakup is flat resilience.
A day-90 outcome promise, and one marketing superlative. Both gone.
what actually has evidence — and how we use it honestly
Unbroken streaks motivate; broken ones demotivate. Seven experiments in the Journal of Consumer Research. It is why our count never resets to day one: a slip freezes a day, it doesn’t erase your history. The streak framing is motivational scaffolding — we don’t claim the count itself heals you.
If-then plans help — in their positive form. The effect traces to a small 1997 literature, so we treat it as useful rather than proven. And the negated form (“if X, then I will NOT…”) showed ironic rebound — it strengthened the habit (Adriaanse et al. 2011). Our trigger planner literally warns you if you write one that way.
Real recovery stories move meaning-making in two randomized trials — and moved neither hope nor distress, and 11% of reads left people feeling worse. So stories here are one per evening, offered not pushed, each pullable in minutes, and every one is a consented, ethics-approved quote from CC BY research — attributed on the page, because the attribution is the licence fee.
Peer support moves self-reported recovery and empowerment — and nothing else measured.A 19-trial meta-analysis (White et al. 2020, DOI 10.1186/s12888-020-02923-3) found no effect on symptoms, hospitalisation, or network size. We build on the two real effects and refuse to advertise the others.
Distraction games make the wave feel smaller while you play. The same trial found no significant effect on whether people actually indulged the craving afterwards — so our copy says the first thing and never the second.
Structured digital programs can work at scale — the clearest nearby example is insomnia (OASIS trial, N=3,755, DOI 10.1016/S2215-0366(17)30328-0), which is also a caution: half the outcomes rest on the minority who finish.
the 2am question nobody has measured
This app is built around the 2am urge, so here is the most honest sentence we can write about it: no study has ever timed the urge to contact an ex by hour of night. We ran the absence check — the diary studies sample once a day, the one naturalistic ex-contact instrument goes silent overnight by design, and the largest timestamped breakup corpus in existence — over a million posts (Seraj, Blackburn & Pennebaker 2021, PNAS, DOI 10.1073/pnas.2017154118) — has never been analysed by clock hour.
What is supported: at night there is nobody around to answer. The night-risk literature’s own implication section calls for exactly this — support that is available in the hours when it isn’t. That is the half of the thesis we claim, because it is the half we can defend: not “more people are in crisis at 2am”, but “at 2am there is nobody to answer” — and an app whose job is to change that.
method, briefly
Sixteen research passes collected the 331 claims; every claim we actually use was then re-verified against full texts rather than abstracts — which caught errors in both directions, including a famous helping-rate percentage that appears nowhere in its own paper. Where a claim’s best source is an abstract behind a paywall, our notes say so. The graded map runs 745 lines and 113 DOIs; if you are a researcher or journalist and want it, ask— we’ll share the working.
why publish this
Because the reader deciding whether to trust a breakup app at their lowest deserves better than invented percentages — and because an app confident in its honesty should be auditable. If you find a claim anywhere on this site or in the app that doesn’t hold to a source, we want to hear it, and we will fix it in public.
if it's heavier than a breakup
If you're thinking about hurting yourself, please don't sit with it alone. In the US and Canada, call or text 988. In Poland and the UK, 116 123. Anywhere else, findahelpline.com finds a free line where you are — a number you cannot dial is not help. We're peer support, not a crisis service, and this is the one thing we'd never want you to use us for.
the app this audit shaped
NO CONTACT CLUB is what's left when you delete everything the evidence can't carry: a tiny pod of real people at your stage, one shared streak that never resets to day one, and honest silence instead of invented statistics. Free, no account.