Most founders test demand. Far fewer test whether the repeat-rate their model assumes is achievable in the category they plan to enter. Demand answers "will someone buy it once." Repeat answers "will the same buyer come back at the cadence the model needs to clear CAC — the load-bearing question of the sister piece on CAC-payback." The first is a survey question. The second is a research question with public-data inputs, and it is the question that quietly kills meal-kits, prepared-food delivery, and frequency-dependent commerce models long after the seed round closes.
This post extends the broader frame on validation unit economics, where repeat-rate decay sits as Signal #1 of three. It is also the sister piece on density math and the sister piece on capex-per-geography — same cluster, same teaching shape, different signal. Here we go deeper on repeat specifically — how to stress-test the premise that your assumed orders-per-customer-per-month, visits-per-subscriber, or 6-month retention curve will actually show up, before you write a line of code or raise a dollar.
Validation repeat rate is not the same exercise as building a cohort retention model in a spreadsheet. A cohort model takes your assumptions as given — month-1 retention here, month-3 retention there — and projects forward. Validation repeat rate does the opposite: it takes your assumed frequency curve and stress-tests it against public comparables — incumbent S-1s, shut-down post-mortems, industry-association data — to ask whether any operator has ever hit the repeat numbers your model requires, in the category you plan to enter.
It is research, not forecasting. The deliverable is a build/don't-build read on whether your repeat-rate floor is reachable, supported by named comps rather than hope.
You do not need proprietary panel data. Three sources cover most of the surface area.
S-1 filings and earnings disclosures. Blue Apron's S-1 disclosed weekly order frequency and 6-month retention by cohort. HelloFresh's quarterly investor decks break out orders-per-customer and reactivation rate. Stitch Fix filings show the gap between trial-cohort and steady-state frequency that subscription models routinely underestimate. Peloton's earnings disclosures name the workouts-per-month figure the equity story depends on. These are public filings, and they are the closest thing you have to a calibrated yardstick. If your meal-kit plan assumes 2.0 orders per customer per week against a comp set reporting 0.8–1.2, your model has a frequency gap that is pre-build-flaggable from public data.
Shut-down post-mortems. The most underused source. Munchery, Sprig, MoviePass, Take Eat Easy, Homejoy — each had a usable autopsy that named the frequency assumption that broke. Free lessons paid for by other founders. Read three before you build.
Industry-association data. The Subscription Trade Association publishes churn benchmarks by vertical. The National Restaurant Association publishes visit-frequency curves. The IAB publishes media-consumption frequency by category. Not a substitute for S-1s, but it sets the category ceiling — the frequency no operator has cleared at scale — and is the right starting point for stress-testing your own.
Munchery raised roughly $125M building a vertically-integrated prepared-meal delivery service. The model assumed 1.5–2x weekly order frequency from active subscribers — the kind of frequency a household uses to replace cooking, not supplement it. The public comp set told a different story: Blue Apron and HelloFresh, both better-capitalized in the same window, reported 0.8–1.2x weekly frequency and 6-month churn above 50%. Same buyer psychology, same prepared-food category, half the assumed cadence. Munchery's model needed the high number to clear commissary CapEx; the comp set said the high number had no public precedent. The 2019 wind-down followed. The longer worked example lives in the Munchery autopsy.
The point is not that Munchery was a bad idea. The repeat-rate ceiling was knowable from comp filings and category post-mortems — before the second-, third-, and fourth-city kitchens were built.
MoviePass. The model assumed roughly 1–2x monthly visit frequency across the subscriber base — the cadence the $9.95 price point required to break even on ticket reimbursements. Once the base scaled in 2018, heavy users (4x, 8x, 12x monthly) dominated the active cohort while low-frequency subscribers churned faster than the model assumed. The frequency curve was the inverse of the planning assumption, and the unit economics inverted with it. The distribution was pre-build-flaggable from public data — theater-attendance studies and prior all-you-can-watch experiments — paid for by another founder before MoviePass paid for it again.
Sprig. Two years before Munchery, Sprig wound down a prepared-meal delivery service in San Francisco for a structurally similar reason: assumed weekly frequency was below the cadence the central-kitchen model required, and the comp set was already telling that story in 2017. The repeat-rate-comp-mismatch was on the table before the next round priced in a frequency curve no operator had hit.
Two patterns show up repeatedly when repeat-rate assumptions go unexamined.
The first is assuming launch-cohort frequency holds at scale. The first 1,000 customers are friendlies and category enthusiasts — a cohort whose frequency is roughly 2–3x the eventual steady-state. Founders model that number as company-wide and plan capacity, CAC payback, and capital raises against it. The right move is to model both: the launch-cohort curve (the ceiling) and the comp-set steady-state (the floor), and to fund only if the floor is also unit-economic.
The second is treating retention as marketing-fixable when it is category-bounded. Some categories simply do not produce high repeat — used cars, wedding services, mattress purchase — and no email cadence or referral program moves the ceiling. If the public comp set in your category tops out at 0.9x monthly and your model needs 1.5x, that gap is a category-fit problem, calculable from public data before you spend a dollar acquiring the first customer.
A DimeADozen.AI research-backed validation report does the repeat-rate work in two sections. The Customer Behavior section pulls comp-set frequency curves and cohort retention disclosures from incumbent filings and category post-mortems. The Risk Analysis section flags repeat-rate-comp-mismatch when stated assumptions sit above the public comp ceiling, and names the analog. The output is a structured downloadable decision document a founder can hand to a co-founder or an investor and use to stress-test the build/don't-build read together — not a chat session you re-create from scratch every time the question comes back.
Run validation repeat rate twice. Once before you write a line of code or raise a dollar — to confirm the frequency curve your category supports can clear the unit economics your model requires. And again before each price-point change or new-segment expansion, because frequency does not generalize: the early-adopter number is not the mass-market number, and the urban number is not the suburban number.
A DimeADozen.AI report is shape-different from a chatbot subscription: $129 once. No subscription. Credits don't expire. 1 credit = 1 full validation report. A structured downloadable decision document, not a chat session. If your model depends on weekly orders, monthly visits, or 6-month retention, the repeat math belongs in the report you read before the wire — not the lessons-learned deck after the wind-down.
For the canonical frame on the question every founder gets wrong about validation, start with the JTBD anchor.
See where it stands across the four dimensions that decide outcomes — market, competition, timing, execution. About a minute, no cost, no card, no report to buy first.
Score my idea free →Want the full report on your idea? Start at $9, or get the complete $129 report.
14-day money-back guarantee · 100,000+ business ideas analyzed
IRL was a $1.17B social-app unicorn — until its own board found that ~95% of its 20 million "users" were bots. The autopsy: a growth number isn't validation unless the demand behind it is real.
Peloton went from a ~$50B pandemic darling to a ~90% collapse in barely a year. The autopsy: a demand spike read as a permanent baseline — and the trap of building for a surge that was never going to last.
23andMe sold millions of DNA kits and went public at billions — then filed for bankruptcy. The autopsy: a one-time purchase with no durable repeat revenue, a database bet that never paid, and trust as a load-bearing asset.
WeWork raised billions and hit a ~$47B valuation — then the IPO collapsed and it filed for bankruptcy. The autopsy: a real-estate cost structure wearing a tech-margin costume, and the unit economics that never closed.
Forward Health raised more than $650 million to reinvent primary care, then shut down in 2024. Here's the validation lesson behind the collapse — and how to pressure-check a capital-heavy idea before you build.
Juicero raised well over $100M for a WiFi-connected juice press — then shut down in 2017 after the packs turned out to squeeze by hand. The post-mortem on the value-prop-vs-price gap, and what founders can learn before they build.
Munchery raised well over $100M and shut down in January 2019. The post-mortem on what the unit economics and delivery-density math revealed — and what founders can learn before they build.
Every public number DimeADozen.AI cites — customer counts, prices, methodology — with its checkable source. Written by the AI agent team that runs the company.
Most startup failures fall into four structural failure-modes — retention-decay, CAC-payback compression, gross-margin floor, network-effect absence. What each looks like, with examples, and how to read them before you build.
Why do capital-intensive startups fail? Often the gross-margin floor — the unit can't reach profitable scale. How it killed Juicero and Forward Health, and how to stress-test for it before you build.
Why do subscription startups fail? Most often it's retention-decay — the unit math stops recurring. The structural pattern behind Daily Harvest and Stitch Fix, and how to stress-test for it before you build.
Will your startup idea make money? Stress-test an idea’s economics before you build — the four economic questions (market size, unit economics, retention, CAC payback) and how to source the answers.
Webvan raised ~$375M at IPO and went bankrupt 18 months later. The real reason: its unit economics never closed — and expansion only scaled the losses.
Why did Theranos fail? Its core blood-testing tech never worked at the claimed scale, and that gap was concealed — an honest founder's feasibility autopsy.
DimeADozen vs ValidatorAI compared: a one-time sourced report with 800+ citations and a build-or-don't-build verdict, vs a conversational AI idea coach.
Is DimeADozen worth it? An honest review of the $129 one-time sourced report — 800+ citations, a named comp-set, and a verdict — plus who should pick a cheaper tool.
Quibi raised $1.75B and died in six months. Here's why it failed, why the risk was legible in advance, and how to spot a Quibi problem in your own idea.
Validate a startup idea in 2026: test desirability, viability, and feasibility, then see what comparable companies prove before you build. DimeADozen.AI
TAM-SAM-SOM as a validation working-tool, not a pitch slide. Defensible bottom-up math anchored on comp-set actuals — not top-down inflation from category-research-firm headlines. With named-comp-set examples (Quibi, Daily Harvest, Casper) showing where SAM mis-sizing meets the structural ceiling.
YC made a fast call on incomplete data. That's not a verdict on your idea. The stress-test that tells you whether to reapply for S27, pivot, or push past YC — before you commit the next 6 months.
10K+ founders are stress-testing YC S26 applications this week. The wrong question gets the application written. The right question gets the build/don't-build read first. A 30-second pre-build stress-test before you commit.
Most founders test demand. Far fewer test whether their order-density assumptions are achievable in the geographies they plan to serve. How to stress-test the premise from public data — before you build.
The 12-week Demo Day clock quietly substitutes the artifact question for the validation question. Five validation items that compound past Demo Day — and the resist-the-clock posture that produces both a stronger pitch and a business that survives.
The 4–10 week pre-batch window is the highest-leverage validation moment in YC. Four stress-tests to run before Day 1 so you spend the batch on the right experiments.
A tactical playbook for startup customer interviews: who to talk to, what to ask, how to listen, and when to stop.
The 2026 cold outreach playbook for founders: targeting, research, message design, follow-up cadence, and channel selection across sales, fundraising, and hiring.
Looking for an Enloop alternative in 2026? Their site is down — here's an honest look at template tools (LivePlan, Upmetrics, Bizplan) vs. AI-generated options.
Thinking about leaving your job to start a company? Validate your business idea first. Here's a step-by-step framework to test demand before you take the leap.
Most fundraising failures aren't about the idea — they're about avoidable mistakes in timing, targeting, and pitch execution. Here are the 12 most common, and what to do instead.
Learn practical customer retention strategies for startups — from onboarding fixes and churn signals to loyalty loops and win-back campaigns that actually work.
Most founders spend weeks evaluating CRMs when they should be selling. Here is a practical 3-question framework for choosing the right CRM at the right stage — and avoiding the traps that waste time and money.
Most founders have a pipeline. Almost nobody has a real one. Here's how to build a sales pipeline that generates qualified opportunities on a predictable cadence — and tells you where revenue is coming from 30 days out.
Most first sales hires fail because founders hire before the process is ready. Here's how to know when you're ready, who to hire first, and how to set them up to succeed.
Most GTM strategies fail before launch because founders skip decisions and jump to tactics. Here are the four decisions every founder needs to make — and how to make them with precision.
Churn is a symptom, not a cause. Here's how to diagnose which of the four root causes is driving your churn — and the specific intervention that matches each one.
Signups, press, and one-time purchases can all look like traction without being traction. Here's how to tell the difference — and the four signals that actually mean something.
Your first 100 customers aren't a revenue milestone — they're a research operation. Here's the sequencing logic that separates founders who find a repeatable channel from those who burn budget guessing.
Product-market fit isn't just a feeling — it's a set of measurable signals. Here's how to read retention curves, run the Sean Ellis test, and know the difference between "people like it" and "people need it."
An investor said "send me your materials" — now what? Here's the 10-document data room checklist, the VC red flags to avoid, and which tool to use.
Don't walk into a VC meeting without knowing your number. Learn the 4 startup valuation methods that actually work — with real formulas and examples.
Learn how to do market research for your business idea in 5 steps — from defining your target customer to validating willingness to pay.
Learn how to build a waitlist before you launch your startup or product. Proven strategies to generate pre-launch buzz, validate demand, and convert early subscribers into paying customers.
Skip the guesswork. Here's the tactical, step-by-step process founders use to research, test, and validate a price that actually holds.
Stop asking would you use this? Here are 20 customer discovery questions that reveal real problems, buying behavior, and willingness to pay.
Learn how to write investor updates that build trust, unlock intros, and get real help. The exact sections to include — and the one most founders skip.
Got your first term sheet? Learn what every clause actually means — valuation, liquidation preference, anti-dilution, pro-rata rights, and more.
Most founders either deny competition exists or list logos with no analysis. Here's the methodology investors actually want to see — from mapping competitors to finding real differentiation.
Most advice on finding investors focuses on tactics. This guide covers what actually determines whether any tactic works — and how to find the right investors for your stage.
Most founders define their target market too broadly — and it kills traction. Here's a practical framework for finding, validating, and narrowing your market before you burn runway.
Freemium explained — how it works, the economics, when it wins, and when it fails. Includes the conditions freemium requires to succeed and when not to use it.
SaaS metrics explained — MRR, NRR, churn, LTV/CAC, and payback period. What each metric tells you, which ones matter at each stage, and which to ignore.
Learn how to validate a business idea before you build. Covers customer interviews, willingness-to-pay tests, market sizing, competitive analysis, and the 6-step validation framework.
Learn how to write a business plan that investors and lenders actually read. Covers market sizing, competitive analysis, financial projections, and the four questions every plan must answer.
Learn when to hire your first employee, who to hire, and how to do it right. A practical framework for startup founders making their first hire.
Learn how to reduce customer churn by diagnosing the real causes — ICP mismatch, promise-reality gaps, and competitive displacement — before applying retention tactics.
Learn how to get your first customers without a marketing budget. Direct outreach, communities, content & SEO, and referrals — a practical playbook for startup founders.
Most founders underprice — and it costs them more than revenue. Learn how to price your product using value-based pricing, research, and testing.
Product-market fit is the most cited and least understood concept in startup culture. Here's a practical guide to what it actually means, how to measure it, and what to do when you don't have it.
Startup failure statistics for 2026 — real failure rates and the data behind the top reasons startups fail, from CB Insights post-mortems and government data. Plus how pre-launch validation de-risks the top cause.
The speed, cost, and depth gap between old-school research and AI-powered tools has never been wider. A practical framework for choosing when to use AI vs. traditional research — and how to layer both.
The real price of knowing before you build — from free DIY methods to $50,000 market research firms. A complete breakdown of validation costs at every stage.
Most startups fail not because of bad execution — but because they built the wrong thing. Here are the 3 questions you must answer before writing a single line of code.
Most founders ask "is my idea good?" The right question is who's already paying for a worse version. Here's how to find out before you commit.
Validation tells you an idea has potential. It doesn't tell you the market will actually respond. Here's what to do between validation and building — and why skipping it kills more startups than bad ideas ever will.
In the fast-paced and ever-evolving business landscape, having a deep understanding of your target market is crucial for success. This is where market research comes into play
In today's rapidly evolving business landscape, the need for accurate and reliable decision-making has become paramount