/segmently-unit-economics

UC-11 — Hypothesis card: which metric, then which change

Route: "what to test next", "hypothesis card"; a test's counts with its card → field 10 (SKILL.md § Router) · Eval: P11 · MCP: uc-11, cases-uc-11 Contents: Turns · Calls · Answer skeleton · Boundary · Stops · Worked example

Turns

On .ue/<slug>.json (none → init). The worked card is ANOTHER scenario: copy its shape, never a sentence or a figure of it — every number and every list comes from the reader's own runs. Place each input by walk.lines, never by the stage alone.

• Turn 1 — horizon: "In how many days do you want to read the result?" None named → this turn prints the choice run for its mde columns and closes with only that question; their reply still names none → 30, labelled "default horizon", and every later run at that 30 adds --horizon-default. Nothing else is asked before the table is shown.
• Turn 2 — lever and change: after the table, one question naming only recommendation.first: "What change would you make to move <lever> (<its stage label>)?" Only when they have no idea for it, the next turn asks the same for runnerUp; a realistic: false lever only as a named big bet. Say nothing about whether their change is readable before they give a lift.
• Turn 3 — believed lift: "How much do you expect <their change> to move <lever>, as +X %?"

Calls

• The choice run ue rank .ue/<slug>.json --horizon 14,21,30 judges realistic at its shortest horizon, so it only helps choose; the card's table is a fresh ue rank .ue/<slug>.json --horizon <N> --metrics-only (the metric levers; UC-13: an offer change) — ONE run, JSON, no link with it (asked for one: SKILL.md § The loop, step 4).
• The card: the CLI lands the lift (its snapped warning gives typed → landed; the card quotes the landed lift) and writes the variant at it, label and measuredOn set — never build or edit it by hand; name it in variants[] by its label: ue rank .ue/<slug>.json --horizon <N> --metrics-only --override <lever>=<lift> --card .ue/<slug>/card-<lever>.json
• The read (field 10), on the project before any re-base; --lever is the card's lever — any rank lever but renewals: ue stat read --a <n>/<x> --b <n>/<x> --project-file .ue/<slug>.json --lever <lever> --horizon <N> --belief <the card's typed lift>
• On significant only, the same read, every flag kept (--belief, --horizon-default), plus --rebase: the CLI writes the re-base (the old base into variants[], the counts into observed[], the lever at read.pB as printed, measuredOn, updatedAt) — never a hand edit.

Answer skeleton

1. The table run: sentences.traffic, every sentences.table line, a blank line, every sentences.belowTable line as a - bullet, a blank line, sentences.pick (SKILL.md § Blocks the CLI renders, the rank table), then its boundary whole — learnable rows are THIS run's realistic column. The choice run: the same, no sentences.pick.
2. The one question of Turn 2, naming only its lever — never the pick or a row again (rule 5).
3. The card, ten fields, each with its unit (field 10 too: cases/uc-13-test-an-offer-change.md Turn 7):

4. The card run's boundary, whole; then only the counts question: "What are the counts once each arm has reached <nPerArm> — per arm, how many saw it and how many converted?"
5. At the read: read.lines whole — its drift line (context.drift) and the verdict's sentence are lines of it (underpowered, not yet: context.more.sentence, never "did not work"; significant: context.realized.sentence and its warnings — the CLI's own realized-gain run, never a difference of profit per start multiplied out by hand).
6. On significant only, the --rebase read's re-base block (SKILL.md § Blocks the CLI renders); the lever at read.pB as printed, and every figure about the re-based scenario (a next card's table) comes from a run on it.
7. The read's context.boundary, whole; underpowered → only the counts question at read.needPerArm.

Boundary

Rendered three times — the table's boundary, the card's boundary, the read's context.boundary — each printed whole where its skeleton row stands; the read's is whole when sent with --belief (and, at a 30 nobody named, --horizon-default).

Stops

• No realistic row → "bigger change or more traffic"; a landed lift below the row's mde → "within N days you can only see a ≥ X % change" — no card.
• A price → ue stat means with the reader's σ (cases/uc-08-test-plan.md); a trial lever → readiness.line's trial days first; renewals → a calendar read (its refusal says what to read).
• Re-base needs THIS card's significant read; a mix read or drop leaves only context.rebase.lines.
• A card turn's ue check adds --expect uc-11:card (SKILL.md § The loop, step 6).

Worked example

Every input below is invented (the base's label and source note say SYNTHETIC): a process dry run of the card, not a view of any product. This is ANOTHER scenario than yours — copy its shape, never a sentence or a figure of it. Every list, every boundary and every figure below is THIS run's at $900 a day: at another budget which rows are learnable, where a lift lands and every figure change, so rebuild each one from your own runs.

The base is a web paywall — Monthly with trial $19.99 (24 % of buy-taps, completion 95 %, trial → paid 45 %, 2.9 charges) and Annual $99.99 (12 %, 97 %, 1 charge) — at $900 a day, cost per start $1.10, chain 38 % → 55 % → 32 %, card fee 2.9 % + $0.30, refunds 3 %, disputes 0.4 % at $15.00, 88 % open the app. In a session it is the project file's scenario; here, get it with:

npx -y @segmently/cli ue parse "https://www.segmently.ai/unit-economics?v=2&label=SYNTHETIC%20meditation%20app%2C%20web%2C%20US&cps=1.1&p1=38.0&p2=55.0&p3=32.0&vol=900:budget_day&measured=SYNTHETIC%3A%20invented%20for%20a%20process%20dry%20run%2C%202026-09-23&fee=2.90~0.30~card_processor&refunds=3.00&disputes=0.40~15.00&activation=88.0~1&p=Monthly%20with%20trial~19.99~1month~7d-free~24.00~95.0~2.90~45.0&p=Annual~99.99~1year~none~12.00~97.0~1.00~&ref=skill"

and save its scenario field as base.json — and, as the project file, .ue/synthetic-meditation.json:

{ "kind": "ue_project", "v": 1, "slug": "synthetic-meditation", "scenario": <that field>, "observed": [], "variants": [], "decisions": [] }

Turn 1 — the horizon. This reader has no horizon in mind (a reader who names one skips the next run). ONE run with three horizons, in JSON, to see the mde columns:

npx -y @segmently/cli ue rank base.json --metrics-only --horizon 14,21,30

sentences.traffic, every sentences.table line and every sentences.belowTable line — printed whole (no sentences.pick: a choice run only helps choose):

818.18 funnel starts a day (base.startsPerDay) at $900 a day

• Ranked by gain per day of testing: p1 and p2 are learnable inside 14 days in this run's realistic column.
• The population a test can use depends on the lever's place in the funnel — 818.18 funnel starts a day for p1, 310.91 funnel starts past step 1 a day for p2, 171 funnel starts reaching the paywall a day for p3, 54.72 buy-taps a day for bought, 54.72 buy-taps a day for mix, 19.7 buy-taps that pick a paid product a day for close, 12.48 trial starters a day for trialConv and 11.98 payers a day for renewals.
• Within 14 days you can only see a ≥ 17.0% change on p3.
• Within 14 days you can only see a ≥ 28.7% change on bought.
• Within 14 days you can only see a ≥ 61.2% change on mix.
• Within 14 days you can only see a ≥ 82.8% change on trialConv.
• Price is not ranked here — a price test needs the variance of revenue per user from your own data.
• rank p1: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +10.53% — every figure of this row uses the landed lift
• rank p3: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +9.38% — every figure of this row uses the landed lift
• rank close: +10.00% was asked; the calculator's own control stops at its range and the lifted scenario carries +4.53% — every figure of this row uses the landed lift

This run judges realistic at its shortest horizon, 14 days, so it only helps choose — none of its rows goes on the card: close reads false here and true in the 21-day run below.

The base's walk.lines are the first 16 lines of the boundary below. Its boundary — printed whole (rule 5: its realistic column is a verdict; its horizon line names the horizons offered, none picked yet), on the project file, as a session runs it:

These figures are this scenario's arithmetic on the inputs above.

• cost per start $1.10, p1 38 %, p2 55 % and p3 32 % — measured: your project file — SYNTHETIC: invented for a process dry run, 2026-09-23
• Monthly with trial: $19.99, monthly, a 7-day free trial — measured: your project file (price, cadence, trial)
• Monthly with trial's take 24 % — measured: your project file (the CLI's init split for these products is 30.1 %, not this)
• Monthly with trial's charges 2.9 — measured: your project file (the book's monthly count is 2.86)
• Monthly with trial's completion 95 % — measured: your project file (the book's is 100 %)
• Monthly with trial's trial conversion 45 % — measured: your project file (the book's is 50 %)
• Annual: $99.99, yearly, no trial — measured: your project file (price, cadence, trial)
• Annual's take 12 % — measured: your project file (the CLI's init split for these products is 14 %, not this)
• Annual's charges 1 (its cadence cap) — assumed: the book
• Annual's completion 97 % — measured: your project file (the book's is 100 %)
• $900 a day — measured: your stated budget
• the fee 2.9 % + $0.30 on the charged base — assumed: the book's card_processor preset
• refunds 3 % — measured: your project file
• disputes 0.4 % at $15.00 each — measured: your project file
• activation — 88 % open the app, 1 charge from those who do not — measured: your project file
• the book's first renewal behind payback (monthly 60 %, from words.payback) — assumed: the book
• the +10 % on p1, p2, close, p3, bought, mix, trialConv and renewals — assumed: ue rank's default ("assumed lift"), at each lever's landed lift in the table.
• the 14-, 21- and 30-day horizons of this choice table — assumed: offered for the choice, none picked yet
• the 50/50 split of the test this table plans — assumed: the test design
• p1 is decidable in 5.75 days and p2 in 8.14 days at this budget.
• not testable soon, in this 14-day run: close 17.97 days, p3 45.44 days, bought 111.66 days, mix 438.74 days and trialConv 319.55 days — past 14 days at this budget, each at its own days; renewals — a calendar read.

Then only the question, "In how many days do you want to read the result?" The reader picks 21 days. Nothing else is asked before the table.

The card's table — a fresh run at 21 days. The reader gave no lifts yet, so every lever takes the CLI's default +10 % relative ("assumed lift"). ONE run, in JSON:

npx -y @segmently/cli ue rank base.json --metrics-only --horizon 21

sentences.traffic, every sentences.table line, every sentences.belowTable line and sentences.pick — printed whole, in this order (the CLI renders every cell from that run's own row, rule 9):

818.18 funnel starts a day (base.startsPerDay) at $900 a day

• Ranked by gain per day of testing: p1, p2 and close are learnable inside 21 days in this run's realistic column.
• The population a test can use depends on the lever's place in the funnel — 818.18 funnel starts a day for p1, 310.91 funnel starts past step 1 a day for p2, 171 funnel starts reaching the paywall a day for p3, 54.72 buy-taps a day for bought, 54.72 buy-taps a day for mix, 19.7 buy-taps that pick a paid product a day for close, 12.48 trial starters a day for trialConv and 11.98 payers a day for renewals.
• Within 21 days you can only see a ≥ 13.9% change on p3.
• Within 21 days you can only see a ≥ 23.3% change on bought.
• Within 21 days you can only see a ≥ 49.1% change on mix.
• Within 21 days you can only see a ≥ 52.2% change on trialConv.
• Price is not ranked here — a price test needs the variance of revenue per user from your own data.
• rank p1: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +10.53% — every figure of this row uses the landed lift
• rank p3: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +9.38% — every figure of this row uses the landed lift
• rank close: +10.00% was asked; the calculator's own control stops at its range and the lifted scenario carries +4.53% — every figure of this row uses the landed lift

p1 is the pick (recommendation.first), p2 the runner-up (recommendation.runnerUp).

base: profitPerStart -0.03, roas 0.97 (net — the base evaluate's valueBasis), startsPerDay 818.18.

A {{…}} line stands for lines already shown above; an answer prints the lines, never the marker (ue check flags one).

Its boundary — printed whole (the project file's run, as a session runs it):

These figures are this scenario's arithmetic on the inputs above.

{{walk: base — the 16 lines above}}

• the +10 % on p1, p2, close, p3, bought, mix, trialConv and renewals — assumed: ue rank's default ("assumed lift"), at each lever's landed lift in the table.
• the 21-day horizon — measured: your answer
• the 50/50 split of the test this table plans — assumed: the test design
• p1 is decidable in 5.75 days, p2 in 8.14 days and close in 17.97 days at this budget.
• not testable soon, in this 21-day run: p3 45.44 days, bought 111.66 days, mix 438.74 days and trialConv 319.55 days — past 21 days at this budget, each at its own days; renewals — a calendar read.

Turn 2 — lever and change. Then only the question, naming only recommendation.first: "What change would you make to move p1 (acquisition)?" The reader names the change: the landing's first screen asks one goal question (sleep / stress / focus) and routes to the matching step 1. Nothing is said about whether it is readable before a lift is given.

Turn 3 — believed lift. The question: "How much do you expect that change to move p1, as +X %?" The reader: +12 %. ONE override run, which also writes the card variant; the CLI lands the lift on the calculator's grid and writes .ue/synthetic-meditation/card-p1.json — its own lifted state for p1, labelled "card p1 +11.84 % (believed)", measuredOn "believed lift, not observed, 2026-09-23":

npx -y @segmently/cli ue rank base.json --metrics-only --horizon 21 --override p1=0.12 --card .ue/synthetic-meditation/card-p1.json

The card's lift is the landed one — the row's lift 0.118421, above the row's mde at 21 days (0.0549), so the card goes on. The p1 row, field by field:

Its card block goes into field 5 whole (the field-5 label below), its sentences.warnings into field 7, and the row's guardrail.watch, guardrail.sentence and bet.sentence into fields 4, 6 and 9. Its name in variants[] after the answer is its label, "card p1 +11.84 % (believed)".

The card, every figure from the runs above:

1. Horizon and traffic — 21 days (the reader's); 818.18 funnel starts a day (base.startsPerDay); $900 a day.

2. Base — the project file's own evaluate, npx -y @segmently/cli ue evaluate .ue/synthetic-meditation.json --explain (stage live_no_data): value per buy-tap $16.03 net against a required $16.45; CAC per payer $75.10; profit per start −$0.03; ROAS 0.97 (net, its valueBasis); payback as words.payback prints it, "Payback: not covered within 12 months assumes the book's first renewal (monthly 60%) and one steady renewal after it that matches your charges count"; readiness.line, "paywall mix decidable after 6 days · trial verdict after 33 days · annual renewals due at month 12, observable from month 13". Its walk.lines — printed whole, every deduction with its value among them, each take placed by the CLI against its own init split:

   {{walk: base — the 16 lines above}}

   The ue link result says privacy.due true (carries prices, chain, volume, takes, charges, completion, trialConv, deductions), so the privacy sentence comes first. privacy.sentence — printed whole:

   This link carries your numbers in plain text — browser history, referrers and analytics can see it. I can give you the JSON file instead.

   npx -y @segmently/cli ue link base.json

   https://www.segmently.ai/unit-economics?v=2&label=SYNTHETIC%20meditation%20app%2C%20web%2C%20US&cps=1.1&p1=38.0&p2=55.0&p3=32.0&vol=900:budget_day&measured=SYNTHETIC%3A%20invented%20for%20a%20process%20dry%20run%2C%202026-09-23&fee=2.90~0.30~card_processor&refunds=3.00&disputes=0.40~15.00&activation=88.0~1&p=Monthly%20with%20trial~19.99~1month~7d-free~24.00~95.0~2.90~45.0&p=Annual~99.99~1year~none~12.00~97.0~1.00~&ref=skill

   It returns roundTrip ok and warnings [].

3. Metric — key: profit per start, −$0.03. Nearest: p1 on 818.18 funnel starts a day (populationName), today 38 %, mde 0.0549 at 21 days. Stage: acquisition.

4. Guardrails — the table run's other rows: p2 8.14 days, mde 0.0625; close 17.97 days, mde 0.0429; p3 45.44 days, mde 0.1386; bought 111.66 days, mde 0.2332; mix 438.74 days, mde 0.4906; trialConv 319.55 days, mde 0.5221; renewals a calendar read (calendar: one billing period per read). The override row's guardrail.watch, verbatim: Watch the steps after p1 — p2, p3, bought, close, trialConv and renewals must not drop while it is tested. Its checkpoint: the drop that cancels the gain on p2 is readable at 1155 per arm (checkpointNPerArm), from day 7.43 (checkpointDay), in a test of 4.56 days (testDays). CAC per payer $75.10 and payback from the base.

5. Hypothesis — If the landing's first screen asks one goal question and routes to the matching step 1, then p1 38 % → 42.5 %, +11.84 % (believed; typed +12.00 %, landed +11.84 %); if the lift holds, profit per start −$0.03 → $0.10 and ROAS 0.97 → 1.09, +$3,157.16 a month at $900 a day.

   The card block — printed whole (an inline card carries it as card.lines): every card.prepared line, every card.snapped line, the card.evaluateCommand run and its KPIs, card.walkLines, card.roasTie, the privacy sentence, the card.linkCommand run and its link:

   • chain.p1 0.38 × 1.12 = 0.4256 — your numbers, prepared
   • chain.p1 0.4256 → 0.425

   npx -y @segmently/cli ue evaluate .ue/synthetic-meditation/card-p1.json --explain

   Its KPIs: value per buy-tap $16.03 net; CAC per payer $67.15; profit per start $0.10; ROAS 1.09 (net, its valueBasis).

   • cost per start $1.10, p2 55 % and p3 32 % — measured: your scenario file
   • p1 42.5 % — assumed: your belief — believed lift, not observed (typed +12.00 %, landed +11.84 %)
   • Monthly with trial: $19.99, monthly, a 7-day free trial — measured: your scenario file (price, cadence, trial)
   • Monthly with trial's take 24 % — measured: your scenario file (the CLI's init split for these products is 30.1 %, not this)
   • Monthly with trial's charges 2.9 — measured: your scenario file (the book's monthly count is 2.86)
   • Monthly with trial's completion 95 % — measured: your scenario file (the book's is 100 %)
   • Monthly with trial's trial conversion 45 % — measured: your scenario file (the book's is 50 %)
   • Annual: $99.99, yearly, no trial — measured: your scenario file (price, cadence, trial)
   • Annual's take 12 % — measured: your scenario file (the CLI's init split for these products is 14 %, not this)
   • Annual's charges 1 (its cadence cap) — assumed: the book
   • Annual's completion 97 % — measured: your scenario file (the book's is 100 %)
   • $900 a day — measured: your stated budget
   • the fee 2.9 % + $0.30 on the charged base — assumed: the book's card_processor preset
   • refunds 3 % — measured: your scenario file
   • disputes 0.4 % at $15.00 each — measured: your scenario file
   • activation — 88 % open the app, 1 charge from those who do not — measured: your scenario file
   • the book's first renewal behind payback (monthly 60 %, from words.payback) — assumed: the book

   The card's ROAS 1.09 equals this run's roasAfter for p1.

   This link carries your numbers in plain text — browser history, referrers and analytics can see it. I can give you the JSON file instead.

   npx -y @segmently/cli ue link .ue/synthetic-meditation/card-p1.json

   https://www.segmently.ai/unit-economics?v=2&label=card%20p1%20%2B11.84%20%25%20(believed)&cps=1.1&p1=42.5&p2=55.0&p3=32.0&vol=900:budget_day&measured=believed%20lift%2C%20not%20observed%2C%202026-09-23&fee=2.90~0.30~card_processor&refunds=3.00&disputes=0.40~15.00&activation=88.0~1&p=Monthly%20with%20trial~19.99~1month~7d-free~24.00~95.0~2.90~45.0&p=Annual~99.99~1year~none~12.00~97.0~1.00~&ref=skill

   It returns roundTrip ok and warnings [] — the file is already on the grid.

6. Change — [your mockup — the landing's first screen asks one goal question (sleep / stress / focus) and routes to the matching step 1]. It must move p1 only; p2 must not fall: −10.59 % there cancels the +11.84 % on p1 ($3,157.16 a month, gainPerMonth); guardrail «p2» is not readable within the test's 4.56 days (readable from day 7.43); extend the test to 7.43 days or keep it as a kill switch only.

7. Test plan — 1864 per arm, 4.56 days, $4,100.80 routed through the test, mde 0.0549 at 21 days; one read of p1 at the planned end, never a daily peek at it. The override run's sentences.warnings — printed whole, one per line, the other levers' default landings included:

   • rank p1: +12.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +11.84% — every figure of this row uses the landed lift
   • rank p3: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +9.38% — every figure of this row uses the landed lift
   • rank close: +10.00% was asked; the calculator's own control stops at its range and the lifted scenario carries +4.53% — every figure of this row uses the landed lift

   While it runs: ue exp start the day it goes live, then once a day ue stat watch with the reader's counters — p1 and p2 so far, and p2 on the last full day; p2 is field 6's kill switch:

   npx -y @segmently/cli ue stat watch --project-file .ue/synthetic-meditation.json --lever p1 --a <n>/<x> --b <n>/<x> --guardrail p2=<n>/<x>,<n>/<x> --today p2=<n>/<x>,<n>/<x>

   It never prints p1's rates; at 1864 per arm it prints field 10's read to run, and on stop the ue exp close call that closes the entry.

8. Gain if the lift holds — $3,157.16 a month (gainPerMonth); charges per payer are the retention unit.

9. Cost — If the +11.84 % on p1 holds, it gains $3,157.16 a month (gainPerMonth). The odds that it holds were not given. The cost to build was not given. Beside it: $4,100.80 routed over 4.56 days. This reader gave no cost or odds, so the run took no --cost / --odds (the growth cycle, cases/uc-12-growth-cycle.md, asks for them).

10. Read and close — printed here, in the card, before any counts exist: the read with its placeholders, per arm <n> people who saw it and <x> who converted — the reader's own counts, at the planned n (1864 per arm) — and the card's typed lift as --belief:

    npx -y @segmently/cli ue stat read --a <n>/<x> --b <n>/<x> --project-file .ue/synthetic-meditation.json --lever p1 --horizon 21 --belief 0.12

    At the read it prints read.lines, whole; one read at the planned end.

The card run's boundary — printed whole, on the project file:

These figures are this scenario's arithmetic on the inputs above.

{{walk: base — the 16 lines above}}

• p1 +11.84 % (typed +12.00 %) — assumed: your belief, at its landed value.
• the +10 % on p2, close, p3, bought, mix, trialConv and renewals — assumed: ue rank's default ("assumed lift"), at each lever's landed lift in the table.
• the 21-day horizon — measured: your answer
• the 50/50 split — assumed: the test design
• p1 is decidable in 4.56 days, p2 in 8.14 days and close in 17.97 days at this budget.
• not testable soon, in this 21-day run: p3 45.44 days, bought 111.66 days, mix 438.74 days and trialConv 319.55 days — past 21 days at this budget, each at its own days; renewals — a calendar read.

Then only the question: "What are the counts once each arm has reached 1864 — per arm, how many saw it and how many converted?"

The read — the reader's counts arrive. Two synthetic endings at the planned n (1864 per arm), each field 10's read with the counts filled in, on the project file before any re-base:

npx -y @segmently/cli ue stat read --a 1864/708 --b 1864/772 --project-file .ue/synthetic-meditation.json --lever p1 --horizon 21 --belief 0.12
npx -y @segmently/cli ue stat read --a 1864/708 --b 1864/736 --project-file .ue/synthetic-meditation.json --lever p1 --horizon 21 --belief 0.12

Ending B, read.lines — printed whole (its drift line, context.drift, then context.more.sentence):

pA 37.98 %, pB 39.48 %, lift 0.0395, diff 0.015, ci95 -0.0163 … 0.0463, z 0.94, pValue 0.3465, needPerArm 16509, verdict underpowered

p1 scenario 38 % → observed 39.48 % (1.48 points higher)

Not enough people yet: the observed difference needs 16,509 per arm (read.needPerArm) and the smaller arm has 1,864 — 14,645 more per arm, 35.8 more days at 818.18 funnel starts a day, split 50/50.

underpowered — recorded, never "did not work"; no realized-gain run and no re-base.

Ending A, read.lines — printed whole (significant: after the drift line come context.realized.sentence and each of its warnings):

pA 37.98 %, pB 41.42 %, lift 0.0904, diff 0.0343, ci95 0.0029 … 0.0657, z 2.14, pValue 0.0322, needPerArm 3187, verdict significant

p1 scenario 38 % → observed 41.42 % (3.42 points higher)

Realized gain $2,455.57 a month (gainPerMonth of the realized-gain run, at its landed +9.21 %); its roasAfter 1.06 is the re-based ROAS.

• rank p1: +8.99% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +9.21% — every figure of this row uses the landed lift
• rank p3: +10.00% was asked; the calculator's own control puts it on its grid and the lifted scenario carries +9.38% — every figure of this row uses the landed lift
• rank close: +10.00% was asked; the calculator's own control stops at its range and the lifted scenario carries +4.53% — every figure of this row uses the landed lift

Then the re-base: the same read command plus --rebase — the CLI writes it into the project file, nothing is edited by hand:

npx -y @segmently/cli ue stat read --a 1864/708 --b 1864/772 --project-file .ue/synthetic-meditation.json --lever p1 --horizon 21 --belief 0.12 --rebase

It wrote .ue/synthetic-meditation.json: the old base is the last variants[] entry, "base before the p1 A/B read"; measuredOn "p1 A/B read: 1864/708 vs 1864/772" (the old note does not fit beside it in 80 characters); updatedAt set; observed[] ends with:

{"kind":"observed_inputs","v":1,"source":"ab_test","lever":"p1","counts":{"control":"1864/708","variant":"1864/772"},"window":{}}

The re-base block — printed whole:

• chain.p1 0.38 → 0.414163 (read.pB as printed) — prepared from your counts

• chain.p1 0.414163 → 0.415

• profit per start −$0.03 → $0.07

• ROAS 0.97 → 1.06 (valueBasis net)

• CAC per payer $75.10 → $68.77

• payback not covered within 12 months → charge 1, month 5

• stage live_no_data → live_measured

p1 scenario 38 % → observed 41.42 % (3.42 points higher)

This link carries your numbers in plain text — browser history, referrers and analytics can see it. I can give you the JSON file instead.

npx -y @segmently/cli ue link .ue/synthetic-meditation.json

https://www.segmently.ai/unit-economics?v=2&label=SYNTHETIC%20meditation%20app%2C%20web%2C%20US&cps=1.1&p1=41.5&p2=55.0&p3=32.0&vol=900:budget_day&measured=p1%20A%2FB%20read%3A%201864%2F708%20vs%201864%2F772&fee=2.90~0.30~card_processor&refunds=3.00&disputes=0.40~15.00&activation=88.0~1&p=Monthly%20with%20trial~19.99~1month~7d-free~24.00~95.0~2.90~45.0&p=Annual~99.99~1year~none~12.00~97.0~1.00~&ref=skill

It returns roundTrip ok and one warning, snapped — chain.p1 0.41416309012875535 → 0.415. The drift line quotes read.pB exactly as the read printed it — not the landed grid value and not the gap between the two arms; the re-based scenario carries the landed 41.5 % (snapped). A next card starts from a new ue rank on the re-based file.

The read's context.boundary — printed whole (the read sent with --belief 0.12, the card's typed lift; every input invented for the dry run):

These figures are this scenario's arithmetic on the inputs above.

{{walk: base — the 16 lines above}}

• p1 after the re-base, read.pB 41.42 % as printed — measured: prepared from your counts
• the counts 708/1864 (control) and 772/1864 (new) — measured: your answer
• the realized-gain run's lift on p1 0.089903 — measured: prepared from your counts
• the realized-gain run's landing on p1: typed +8.99 % → landed +9.21 % (snapped) — measured: prepared from your counts
• the +10 % on p2, close, p3, bought, mix, trialConv and renewals — assumed: ue rank's default ("assumed lift"), at each lever's landed lift in the table.
• p1 +11.84 % (typed +12.00 %) — assumed: your belief, at its landed value.
• the 21-day horizon — measured: your answer
• the 50/50 split — assumed: the test design
• not testable soon, in the card's 21-day table run: p3 45.44 days, bought 111.66 days, mix 438.74 days and trialConv 319.55 days — past 21 days at this budget, each at its own days; renewals — a calendar read.