Launch & iterate

Phases 4 and 5 of the pipeline: getting the MVP from building the MVP into the hands of its first users, then running weekly cycles until the numbers say you have product–market fit, need to pivot, or should stop. It covers launch sequencing, channels, post templates, activation, the metrics that matter (and the ones that flatter), the weekly loop, prioritization, pricing, pivots and kill criteria. The whole pipeline is on the playbook; fundraising and founder advice live under startup advice.

Phase at a glance

PHASE 4 (1 week)                     PHASE 5 (weekly cycles, until exit or kill)
soft launch ─► communities ─► ─ ─ ─► public launches as onboarding works
    │               │                      │
    └─ personally onboard every user ──────┘
                                           ▼
          talk to users → review metrics → pick one bet → ship → measure
                     ▲                                             │
                     └─────────────────────────────────────────────┘
          every 6–8 weeks: pivot / persevere / kill review

Entry criteria

Phase 4 starts with the handoff from building the MVP.

Launch is a process, not an event

A single big launch day wastes your one shot at each audience on a product whose onboarding hasn't met strangers yet. Launch in widening circles; each ring fixes the problems the next ring would have bounced off.

LAUNCH WEEK (Phase 4)
 
Mon  Final pre-launch checks. Send 10 personal emails (ring 1).
     Onboard replies personally (call or screen-share).
Tue  10–20 more personal emails. Fix everything hit on Monday.
Wed  Remaining ring-1 list. First cohort check: who activated?
     Call or email everyone who signed up but didn't activate.
Thu  One community post where your users are (ring 2).
     Answer every comment within the hour.
Fri  Review: sign-ups, activation, first core-loop repeats,
     top 5 problems, quotes. Set Monday's first bet.
     Write down who you will talk to next week (≥ 5 people).

First users: do things that don't scale

Paul Graham's essay "Do Things That Don't Scale" (July 2013) is the canonical text for this phase. His core points:

The Airbnb photography story is a separate, well-documented example: Joe Gebbia has described (First Round Review) how, with New York revenue stuck at about $200 a week, the founders went to New York, rented a camera and replaced hosts' poor listing photos, which doubled weekly revenue to about $400. The lesson isn't photography; it's going to the users and fixing the thing between them and value, by hand.

Unscalable tactics for the first 50 users

• Onboard every user on a 20-minute call, and watch where they struggle.
• Import their existing data for them (spreadsheet, old tool, email).
• Reply to every sign-up personally within an hour: "I'm the founder; what made you sign up?"
• Check the events table daily; message anyone who stalled before activation.
• Build tiny features for individual customers when it keeps them (and log them; don't let them become the roadmap).
• Visit, or at least video-call, the five most active users.

Channels

For the first users, rank channels by fit (are your users there?), not by reach.

Show HN rules that trip people up

From the Show HN guidelines (opens in a new tab): a Show HN is "something you've made that other people can play with". Blog posts, sign-up pages, newsletters, lists and other reading material don't qualify, and landing pages and fundraisers aren't allowed. Make it easy to try "ideally without barriers such as signups or emails"; "If your work isn't ready for users to try out, please don't do a Show HN"; and "Please don't ask friends to upvote or comment." Be around to answer questions. A first comment from the maker with the backstory is a common convention, not a rule.

Launch post templates

SHOW HN
 
Title:  Show HN: <Name> – <what it does, in plain words>
        e.g. "Show HN: Rotabase – shift rotas for small cafés
        that staff get by SMS"
URL:    a page where people can try it (demo account or no
        signup), not a landing page
 
First comment (you, straight away):
  Hi HN, I'm <name>. I built <Name> because <specific problem,
  ideally from your own or your interviewees' experience>.
  What it does: <2–3 sentences, concrete>.
  How it works: <the interesting technical bit, briefly>.
  What's different from <obvious alternative>: <honest answer>.
  Current state: <what works, what doesn't yet>. Pricing: <…>.
  You can try it at <url> without signing up (demo data).
  I'd especially love feedback on <one specific question>.
  I'll be here all day to answer questions.

COMMUNITY POST (subreddit, Slack, Discord, forum)
 
Title: I built a <thing> for <this community's specific
       problem> — looking for feedback from <role>s
 
Body:
  <1–2 lines: who you are and your connection to the
  community. Disclose that you made it.>
  The problem: <in the community's own words; reference a
  common thread or complaint if there is one>.
  What I built: <2–3 sentences>. <screenshot or 30-sec GIF>
  It's <free while in beta / £X per month / free for the
  first N members of this group>.
  What I'm looking for: <honest feedback / 5 people to try
  it this week / to know if this is useless for you>.
  Link: <url>
  <Stay in the thread. Reply to every comment. No sock
  puppets, no vote asks.>

PERSONAL EMAIL (to someone you interviewed)
 
Subject: The thing we talked about in <month>
 
Hi <first name>,
 
In <month> you told me <their problem, in their words —
"I spend Sunday nights redoing the rota on WhatsApp">.
 
I've built a first version of something to fix that and
you're one of the first <10/20> people I'm showing it to.
 
<One sentence on what it does for them.>
 
Could I get you set up on a 20-minute call this week? I'll
do the setup with you (<import your staff list / connect
your account>) so it's useful from day one. <Calendar link>
 
Or try it straight away here: <url>
 
It's rough in places — anything that annoys you, reply to
this email and I'll fix it.
 
Thanks again for your help back in <month>,
<name>

FOLLOW-UP (sent 3–5 days after sign-up, triggered by events)
 
If activated:     "You've <sent your first rota> — how did your
                  staff find it? What would make you use it
                  every week?"
If not activated: "I noticed you signed up but didn't get to
                  <first value>. Was something confusing or
                  missing? 10 minutes on a call and I'll set it
                  up with you: <calendar link>"

Onboarding and activation

Activation is the moment a new user first gets the core value (the aha moment from product design), defined as an event, measured as a rate: activated users ÷ sign-ups in a cohort, within a time window (e.g. 7 days).

At MVP scale, the funnel is a list of names. With 40 sign-ups, look at each non-activated account and ask why; don't A/B test.

Metrics

Early metrics exist to answer three questions: do people get value (activation), do they come back (retention), will they pay (revenue). Everything else is secondary until those are answered.

AARRR pirate metrics

Dave McClure's "Startup Metrics for Pirates" (talk, 2007) splits the customer lifecycle into five stages.

Order of attack for an MVP: activation → retention → revenue → referral → acquisition. Pouring acquisition into a leaky bucket just produces bigger churn numbers.

North Star metric

One number that captures the value delivered to customers and predicts long-term revenue, e.g. "rotas published per week", "invoices paid through the product per week", "nights booked". Rules: it counts value delivered, not activity; it's a rate over time; it moves weekly; the team can influence it. At MVP stage it's usually "number of core loops completed per week by retained accounts". Pair it with a guardrail (e.g. churn) so you can't game it.

Vanity vs actionable metrics

Eric Ries (The Lean Startup, 2011) contrasts vanity metrics (numbers that go up and make you feel good but don't inform decisions) with actionable metrics that tie cause to effect.

Retention cohorts

A cohort table groups users by sign-up period and shows what share is still doing the core action in each later period. It's the single most important chart in Phase 5.

WEEKLY COHORTS: % of each sign-up week's users who completed
the core loop in week N after sign-up
 
Cohort    Users   W0    W1    W2    W3    W4    W5    W6    W7
Sep 01      22   100%   45%   32%   27%   23%   23%   23%   23%
Sep 08      31   100%   48%   35%   29%   26%   26%   26%
Sep 15      19   100%   53%   42%   37%   37%   37%
Sep 22      40   100%   55%   45%   40%   40%
Sep 29      27   100%   59%   48%   44%
Oct 06      35   100%   60%   51%
Oct 13      24   100%   63%
 
W0 = activated users (100% by definition).

How to read it:

• Across a row: how one cohort decays. The shape matters more than any single number.
• Flattening: in this example every cohort stops declining after 3–4 weeks. A curve that flattens to a stable plateau means a core group gets lasting value; it is the clearest early product–market fit signal.
• Down a column: are newer cohorts better? W1 rising from 45% to 63% means the iterations are working.
• Declining to zero: if every row trends toward 0% with no plateau, you don't have a retained core yet, whatever the sign-up numbers say.
• Small numbers: with 22 users, one person is about 4.5 points. Look at the trend across several cohorts, and read the names behind the percentages.
• Match the period to the natural usage frequency: daily for a messaging app, weekly for a rota tool, monthly for invoicing. Weekly cohorts for a monthly-use product will look like failure.

Churn math

Monthly churn compounds. With monthly churn rate $m$ (share of customers lost per month):

Worked example: at $30 a month, 80% gross margin and 5% monthly churn, LTV ≈ $30 × 0.8 ÷ 0.05 = $480. Halve churn to 2.5% and LTV doubles to $960. At MVP scale churn estimates are noisy; talk to every churned customer instead of trusting the decimal.

Distinguish logo churn (customers lost) from revenue churn (MRR lost); expansion revenue can make net revenue churn negative even when logos churn.

DAU/MAU and stickiness

$\text{DAU}/\text{MAU}$ (daily actives ÷ monthly actives) is a stickiness ratio: 1.0 means everyone uses it every day, about 0.2 means a typical user shows up roughly 6 days a month. It only makes sense for products meant for daily use. For a weekly-use product use WAU/MAU or weekly core-loop completion; for a monthly-use product, monthly retention. Define "active" as doing the core action, not opening the app.

NPS and its critiques

Net Promoter Score (Fred Reichheld, "The One Number You Need to Grow", Harvard Business Review, December 2003) asks "How likely are you to recommend…?" on 0–10. Promoters 9–10, passives 7–8, detractors 0–6; $\text{NPS} = \%\text{promoters} - \%\text{detractors}$, from −100 to +100.

For an MVP, skip NPS. Ask the Sean Ellis question (below), count real referrals, and read the free-text answers.

Product–market fit

Marc Andreessen's "The only thing that matters" (June 2007) defined it: "Product/market fit means being in a good market with a product that can satisfy that market." He describes the absence as customers not quite getting value, word of mouth not spreading, usage not growing that fast; and the presence as customers "buying the product just as fast as you can make it". More of his thinking is on a16z advice.

The Sean Ellis test

Sean Ellis (who led early growth at Dropbox, LogMeIn and Eventbrite) proposed asking users "How would you feel if you could no longer use [product]?" with the options very disappointed, somewhat disappointed, not disappointed. In his experience, products where at least 40% answered "very disappointed" were able to grow sustainably; those below usually struggled. The threshold is a heuristic from his observation of startups, not a law.

• Survey only users who have experienced the core product recently. Rahul Vohra, following Ellis, targeted users who used it at least twice in the last two weeks.
• Vohra reports results start to become directionally correct at around 40 respondents.
• Segment the results: the score for your best segment matters more than the average.

Superhuman's PMF engine

Rahul Vohra, "How Superhuman built an engine to find product/market fit" (First Round Review, 2018). Superhuman's score went from 22%, to 33% after segmenting to the users who loved it, to 58% after three quarters of working the engine.

PMF SURVEY (Superhuman / Sean Ellis)
Send to users who used the core product ≥ 2× in last 2 weeks.
 
1. How would you feel if you could no longer use <product>?
   ( ) Very disappointed ( ) Somewhat disappointed
   ( ) Not disappointed (it isn't really that useful)
2. What type of people do you think would most benefit from
   <product>?
3. What is the main benefit you receive from <product>?
4. How can we improve <product> for you?
 
THE ENGINE (repeat every quarter; track the score weekly)
a. Segment: find who answered "very disappointed"; describe
   them (role, company, use case). That's your high-
   expectation customer. Re-score using only that segment.
b. Understand why they love it (Q3 from "very disappointed").
c. Understand what holds back the "somewhat disappointed"
   whose main benefit matches (Q4). Ignore the "not
   disappointed" for roadmap purposes.
d. Roadmap: half on doubling down on what the lovers love,
   half on fixing what holds back the somewhat-disappointed.
e. Re-survey new users; watch the score move.

The weekly iteration loop

Phase 5 runs as a fixed weekly cadence. The unit of work is one bet: a change with a hypothesis and a metric.

WEEKLY REVIEW                                week of ________
NUMBERS (this week / last week)
  Sign-ups: ___ / ___     by channel: ____________________
  Activated (7-day): __% / __%     time to value: ___ / ___
  Core loops completed: ___ / ___  per active account: ___
  Cohort W4 retention (latest mature cohort): __%
  Paying accounts: ___ / ___       MRR: ___   churned: ___
  Errors (new/unresolved): ___     downtime: ___
TALKED TO (≥ 5)
  1. ________ (activated/stalled/churned) — key quote: "___"
  2. ...
PATTERNS
  - ______________________________________________________
LAST WEEK'S BET
  Hypothesis: ____________  Result: ______  Keep/Revert/Iterate
THIS WEEK'S BET
  Because we saw ________________________________ (evidence)
  we believe ____________________________________ (change)
  will move _______________ from ___ to ___ (metric, target)
  We'll know by _________ (date). Effort: ___ days.
NOT DOING (and why)
  - ______________________________________________________

Rules:

• One bet at a time, sized to ship within the week. Big bets get cut into weekly slices.
• Every bet names the metric it should move. "Users asked for it" is evidence, not a hypothesis.
• Fix activation before retention before acquisition. Work on the weakest stage in AARRR order.
• Bugs that hit the core loop jump the queue; everything else competes with the bet.
• Small numbers: with tens of users, "measure" often means "did the specific users we targeted behave differently?" Look at names, not p-values.

Prioritization

Use a scoring model to make trade-offs explicit, not to outsource judgment.

RICE

Intercom's RICE (Sean McBride, Intercom blog):

ICE

ICE (popularised by Sean Ellis for growth experiments): score Impact, Confidence and Ease from 1 to 10 each, and multiply or average them. It's faster and cruder than RICE, which suits weekly bets.

Caveats: scores are guesses dressed as numbers; Confidence should drop if the only evidence is one loud user; and anything on the core loop or blocking activation beats a higher-scoring nice-to-have.

Feedback collection

Feedback log: one row per piece of feedback: date, user, segment, verbatim quote, underlying problem, count. Build for problems with several independent mentions from your target segment, not for the single request from the biggest talker. Interview technique (past behavior, not opinions about the future) is covered in validation.

Pricing experiments

A useful early experiment: raise the price for new sign-ups by a meaningful step (say 50–100%) and watch trial → paid conversion for the next few cohorts. If conversion barely moves, keep the new price.

Pivot, persevere or quit

Eric Ries (The Lean Startup, 2011) defines a pivot as a structured course correction to test a new fundamental hypothesis about the product, strategy and engine of growth, and lists ten types:

Kill criteria

Write them before launch, when you're unbiased, and date them. The sunk-cost trap is real: after months of work every metric looks "almost there".

KILL / PIVOT CRITERIA          written: ________ (before launch)
Review date: ________ (e.g. 12 weeks after soft launch)
 
We will PERSEVERE if by the review date:
  [ ] ≥ __ accounts complete the core loop weekly
  [ ] latest mature cohort retention flattens at ≥ __%
  [ ] ≥ __ paying customers (or ≥ __% trial → paid)
  [ ] Sean Ellis "very disappointed" ≥ 40% in best segment
      (≥ 40 responses)
 
We will PIVOT (pick the type) if:
  [ ] one segment/feature clearly outperforms the rest
  [ ] users consistently use it for an adjacent job
 
We will STOP if:
  [ ] none of the above after __ iteration cycles, OR
  [ ] runway below __ months with no persevere signal, OR
  [ ] we no longer want to work on this problem for 5+ years
 
Signed: ____________   Witness (co-founder/friend): ________

Quitting on schedule is a success of the process: you learned cheaply and get your time back for the next idea, which goes back to ideation with better instincts.

Growth loops vs funnels

Reforge's "Growth loops are the new funnels" (Brian Balfour, Casey Winters, Kevin Kwok and Andrew Chen, 2018) argues the AARRR funnel misrepresents how the fastest-growing products grow. Their definition: "Loops are closed systems where the inputs through some process generates more of an output that can be reinvested in the input." Funnels are one-directional and need constant new input; loops compound.

For an MVP, don't design loops up front. Notice which one appears naturally in the first cohorts (who invited whom? where did organic sign-ups come from?) and reinforce it. Funnels are still the right lens for diagnosing where individual users drop off; loops are the lens for how growth compounds.

Friction points and fixes

Exit criteria / handoff

There's no single day when the MVP phase ends, but there are recognizable signals. You are done with the MVP phase when all three hold for your core segment:

MVP PHASE EXIT (handoff to "startup": growth, team, funding)
 
[ ] Core segment named precisely (who loves it, and why)
[ ] Retention curve flattened for ≥ 3 consecutive cohorts
[ ] Sean Ellis ≥ 40% "very disappointed" (≥ 40 responses)
[ ] One repeatable acquisition channel with known conversion
[ ] Paying customers at a price you'd keep; unit economics
    sketched (ARPA, gross margin, churn → LTV)
[ ] Weekly iteration cadence running; bet log with results
[ ] Manual work listed: what must be automated to handle 10×
[ ] Decision recorded: grow (bootstrap or raise), pivot, stop

What comes next is a different game: automating the unscalable work, hiring, and possibly raising money. For fundraising, demo days and investor expectations, start with Y Combinator's advice; for market-first thinking and the venture perspective, a16z; for how other founders handled this transition, founders. If the signals say pivot or stop, go back to ideation or validation with what you learned.

References

• Paul Graham, "Do Things That Don't Scale" (July 2013) (opens in a new tab): manual recruiting, the Collison installation, Airbnb door to door, delighting users
• First Round Review, "How design thinking transformed Airbnb from a failing startup to a billion-dollar business" (opens in a new tab): Joe Gebbia on the New York photos and the $200 → $400 weekly revenue
• Hacker News, Show HN guidelines (opens in a new tab): what qualifies and how to behave
• Dave McClure, "Startup Metrics for Pirates: AARRR!" (talk and slides, 2007): the five-stage customer lifecycle metrics
• Marc Andreessen, "The only thing that matters" (2007) (opens in a new tab): the definition of product/market fit
• Rahul Vohra, "How Superhuman built an engine to find product/market fit" (First Round Review, 2018) (opens in a new tab): the four-question survey, segmentation, the 50/50 roadmap
• Sean Ellis, "Using product/market fit to drive sustainable growth" (opens in a new tab): the 40% "very disappointed" benchmark in his words
• Sean McBride, "RICE: simple prioritization for product managers" (Intercom) (opens in a new tab): the RICE formula and scales
• Eric Ries, The Lean Startup (Crown Business, 2011): vanity vs actionable metrics, pivot types, pivot or persevere
• Brian Balfour, Casey Winters, Kevin Kwok, Andrew Chen, "Growth loops are the new funnels" (Reforge, 2018) (opens in a new tab): loops vs funnels
• Fred Reichheld, "The one number you need to grow" (Harvard Business Review, December 2003) (opens in a new tab): the origin of NPS
• Timothy Keiningham et al., "A longitudinal examination of Net Promoter and firm revenue growth" (Journal of Marketing, 2007) (opens in a new tab): the replication that challenged NPS's predictive claims