the evidence audit

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

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:

what actually has evidence — and how we use it honestly

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.

download on theApp Store
free · iOS · no login wall · no AI pretending to be a person

updated 2026-09-03 · the underlying audit is versioned with the codebase, so this page changes when the evidence does

peer support · not therapy · crisis line on every screenprivacy · support · guides · timeline · quiz · free app · compare · blocking · the evidence