Research
Why a decision layer, and why for neurodivergent adults
Prerenew rests on a small thesis: auto-renewal quietly charges people for things they stopped valuing, adults with ADHD pay that cost more often, and a reminder that arrives as one concrete decision is more likely to be acted on than a vague intention to “sort out my subscriptions”. This page shows the evidence for each part, labelled by how strong it is, and lists the parts of our design that rest on judgement rather than data.
Sources last checked 6 September 2026. Every number links to the primary document.
How to read the labels
Published in an academic journal after independent review. Strongest, but samples can be small and one study is weaker than a meta-analysis.
National statistics agencies, regulators and courts. Authoritative for what they measure or rule on. Prevalence figures rest on self-report.
W3C accessibility standards and guidance. They define what “accessible” means in practice.
Commissioned by companies that sell related products. Useful for perceptions, but self-reported and the questionnaire is rarely published. We quote them as “respondents said”.
Four numbers
of US adults had a current ADHD diagnosis in 2023, about 15.5 million people. Around half were diagnosed as adults.
GovernmentStaley et al., CDC MMWR 73(40), 2024
of adults worldwide had symptomatic ADHD in 2020, about 366 million people. 2.58% had ADHD that persisted from childhood.
Peer-reviewedSong et al., Journal of Global Health, 2021
of US adults are estimated to be autistic, about 5.4 million people. This is a model projected from child prevalence, not a survey of adults.
Peer-reviewedDietz et al., J Autism Dev Disord, 2020 (CDC authors)
a month. What 1,000 US consumers guessed they spent on subscriptions, against what they found when they itemised it. 42% had forgotten they were still paying for something they had stopped using.
Industry surveyC+R Research, 2022
1. ADHD and money
The pattern across registry data, a 30-year cohort and smaller comparison studies is consistent: adults with ADHD default, borrow and miss payments more often than their peers. The mechanism is tangled up with mood and personality, and none of these studies shows that any tool changes the outcome. They show why a tool that catches the routine money leaks is worth building.
Using full-population Swedish data, adults with ADHD start adulthood with normal credit demand and default rates. Defaults then grow sharply in middle age, leaving poor credit scores and less access to credit.
N = 11.55 million (mental health registry) plus a random credit sample of 189,267 people, 2002 to 2015.
Peer-reviewedBeauchaine, Ben-David and Bos, Science Advances 6(40), 2020
Two-thirds of adults diagnosed with ADHD in childhood were still financially dependent in some way at age 30, against 30% of matched controls.
Prospective clinic-referred cohort from Pittsburgh, mostly male. Do not generalise to all adults with ADHD or to women.
Peer-reviewedPelham et al., Journal of Consulting and Clinical Psychology 88(2), 2020
In one study of 96 adults, those with ADHD were far more likely to carry non-mortgage, non-student debt (48.9% vs 15.7%) and less likely to hold a savings account (53.3% vs 86.3%).
n = 45 with ADHD, 51 controls, Germany and the Netherlands. Small sample.
Peer-reviewedBangma et al., Neuropsychology 33(8), 2019
In a community sample of 1,292 adults, people with ADHD symptoms reported more impulsive buying, more avoidant or spontaneous money decisions, and saving less often. The associations shrank to non-significance after controlling for personality, depression and demographics.
The honest reading: ADHD travels with money trouble, and the mechanism is tangled up with mood and personality.
Peer-reviewedBangma et al., PLoS ONE 15(10), 2020
In a UK survey, 49% of adults with ADHD said they miss bill payments occasionally or often, against 18% of adults without ADHD. 77% said upcoming-bill notifications were helpful.
Bank-commissioned YouGov poll. Sample size not published. Self-reported.
Industry surveyMonzo with YouGov, 2022
2. Everyone forgets subscriptions. Sellers are paid for it.
The strongest single piece of evidence for this product comes from payment-network data rather than a questionnaire. When a card is replaced and people have to act to keep a subscription, a large share simply stop paying, which means they were paying past the point of value all along.
When a payment card is replaced and subscribers must actively re-enter details, cancellations jump. One service lost more than 30% of subscribers in the replacement month. The researchers estimate this inattention roughly doubles sellers’ revenue.
Ten subscription services on a large US card network, 2017 to 2021. The most direct evidence for the premise of this product.
Peer-reviewedEinav, Klopack and Mahoney, American Economic Review 115(5), 2025 (NBER WP 31547, 2023)
51% of US adults who have had a subscription have incurred unwanted charges, from a free trial converting, an accidental sign-up, or forgetting they signed up. 34% say cancelling automatic payments is difficult.
n = 2,497 US adults, January 2022.
Industry surveyBankrate with YouGov, 2022
Average reported subscription spend rose from $237 a month in 2018 to $273 in 2021, while 66% of respondents guessed more than $200 below their itemised total (24% in 2018).
n = 2,500 US consumers, June 2021.
Industry surveyWest Monroe, The State of Subscription Services Spending, 2021
70% of respondents had forgotten to cancel a free trial and been locked into a paid subscription.
n = 1,272 US respondents. This survey counted a narrower set of services and found average spend of $35 a month, which is why we present spend as a perception gap rather than a fact.
Industry surveySelf Financial, 2026
3. The rules were supposed to fix this. They did not.
In the United States, the regulator tried to require easy cancellation and renewal reminders. The reminder requirement was dropped before the rule was final, and the rule itself was struck down before it took effect. That is why a consumer-side tool exists.
- March 2023
The FTC proposes a rule that would, among other things, require annual renewal reminders for subscriptions.
Government Primary document
- October 2024
The final "click-to-cancel" rule is adopted 3 to 2. The annual reminder requirement is dropped.
Government Primary document
- July 2025
The Eighth Circuit vacates the rule in full (Custom Communications v. FTC), six days before its compliance date, on procedural grounds.
Government Primary document
- March 2026
The FTC reopens rulemaking with an advance notice. As of this page, no federal rule requires easy cancellation or renewal reminders.
Government Primary document
4. What the evidence says about our design choices
Each row names a choice in the product, the best evidence we found, and our own verdict on how far that evidence goes.
| Design choice | Evidence | Our verdict |
|---|---|---|
| A concrete cue before each renewal, instead of hoping you remember | If-then plans ("when X happens I will do Y") raise the rate at which people act on their intentions, with a medium-to-large effect (d = 0.65) across 94 tests and more than 8,000 participants. Peer-reviewedGollwitzer and Sheeran, Advances in Experimental Social Psychology 38, 2006 | solid |
| One decision at a time, three fixed answers | The popular "paradox of choice" does not replicate as a general law (a 50-study meta-analysis found a mean effect near zero), but overload does appear when options are hard to compare and preferences are uncertain, which describes deciding about a subscription you half-forgot. | conditional |
| A cap of two nudges a day | Phone users receive about 63 notifications a day and more notifications correlate with stress and feeling overwhelmed. Interrupted work is finished faster but with significantly more stress and frustration. Neither study identifies a safe number. Peer-reviewedPielot, Church and de Oliveira, MobileHCI 2014; Mark, Gudith and Klocke, CHI 2008 | direction only |
| Quiet hours, deferred rather than dropped | Sleep problems are reported by 43% to 80% of adults with ADHD across studies. No study measures whether notification quiet hours change outcomes, and delayed sleep phase would argue for a later morning boundary, not an earlier one. Peer-reviewedWynchank et al., Current Psychiatry Reports, 2017 | indirect |
| Predictable timing and plain, non-judgemental copy | Intolerance of uncertainty is elevated in autistic people and linked to anxiety in 9 of 10 studies; lifetime anxiety disorder prevalence in autistic adults is 42%. Adults with ADHD describe frequent criticism and heightened sensitivity to it, and low self-compassion contributes to poorer mental health. | reasonable inference |
| Reduced motion switch, plus honouring the system preference | WCAG 2.3.3 asks that interaction-triggered animation can be turned off because vestibular reactions include nausea and migraine. 35.4% of US adults aged 40 and over showed vestibular dysfunction on a balance test. Standards bodyW3C, Understanding SC 2.3.3; Agrawal et al., Archives of Internal Medicine 169(10), 2009 | solid |
| Atkinson Hyperlegible offered as a preference, with no reading claim | Fonts marketed for dyslexia have failed controlled tests: OpenDyslexic produced no gain in rate or accuracy and worse accuracy than Arial; Dyslexie was not read faster or more accurately and was preferred by 11.6% of children. Atkinson Hyperlegible was designed for low vision and has no published reading test. Peer-reviewedWery and Diliberto, Annals of Dyslexia 67, 2017; Kuster et al., Annals of Dyslexia 68, 2018; Braille Institute, 2025 | preference only |
5. Where the evidence is thin
These are our design judgements. We think they are good ones. We do not have studies that prove them.
- The T-7, T-3, T-1, day-of cadence. No study tests the best lead time or number of reminders before a renewal. It is a design judgement.
- The number two in "two nudges a day". The stress correlation is real; the threshold is ours.
- "Did you use it?" as the single question. The card-replacement study implies people renew without checking value, which motivates a usage prompt. Nobody has tested one question against alternatives.
- Three taps. There is no evidence for a tap count. It is a promise about our own interface, not a finding.
- Amber instead of red. No study compares colour states for shame or completion. The qualitative work supports avoiding blame in language; it says nothing about colour.
- The Sunday digest (Overwhelm mode). Batching notifications has some HCI research behind it that we have not yet verified against the primary paper.
- High contrast for dyslexic readers. Some dyslexia style guides recommend lower contrast. We offer a toggle and make no claim.
- Anything about outcomes. No study, including ours, shows that this app or any app changes what people with ADHD spend. We will publish our own numbers when we have enough users to say something honest.
6. Statistics you will see elsewhere that we do not use
The 2008 Mark study found interrupted work finished faster with more stress. The 23-minute figure is from a different 2005 paper. We cite the stress finding.
Two pre-registered multi-lab replications (2,141 and 3,531 participants) found ego-depletion effects near zero.
A self-estimate from a bank poll with an unpublished sample size. We quote it only as what respondents said.
The Swedish study found a 3 to 4 times higher rate in the highest default-risk subgroup, in men. It is an association and does not belong on a product page.
That is one survey’s itemised self-estimate; another survey counting fewer categories found $35. We present the gap between guess and itemised total instead.
We could not find it in West Monroe’s own material, only on aggregator sites. The primary release says 66% were off by more than $200.
That needs a very lenient cut-off. Peer-reviewed estimates are about 7% in children and 3% to 7% with stricter criteria.
It is not in DSM-5 and has no validated measure. We say that many adults with ADHD describe intense sensitivity to criticism, which is documented.
Controlled tests say otherwise. We offer a legible font as a preference, not as a treatment.
This is one author’s reasoned estimate in a review (Doyle, 2020), not a measurement. We use it only with that label.
See the design choices in the product
The demo is seeded with six subscriptions and a time machine. No sign-up.
Open the live demo