Humanity’s Last PRD
Latest drop: Gemini 3.8 Flash · 1 Oct
Tasks / Experiment

Experiment specification

Can the model design a test that could actually change the decision?

Measures the modelTask v1.0 · 2 casesDifficulty

What AI gets right here, and what you’ll still have to catch

From 14 graded outputs by 7 models. 36% were usable with at most a quick edit.

Reliably right

  1. Addresses the actual decision91% pass
    The output commits to a clear decision rule: launch only if all gates pass, otherwise do not launch, and an inconclusive result is not a pass.
    GPT-6.1 Sol · API · Batching deliveries before peak season
  2. Tests one change at a time86% pass
    It explicitly keeps the free-delivery banner out of the variant and explains that adding it would make the result uninterpretable.
    GPT-6 Luna · API · Showing the delivery fee up front
  3. Identifies material uncertainty84% pass
    It identifies unknowns like power under correlation, carryover, and city differences, and states that if power is insufficient the rollout will not proceed, resolving the uncertainty.
    GPT-6.1 Sol · API · Batching deliveries before peak season

Where it slips

  1. Respects explicit constraints55% pass
    The output is approximately 780 words, exceeding the 'under 700 words' limit.
    Opus 5.5 · Claude · Showing the delivery fee up front
  2. Sized from the real traffic57% pass
    Sample size is correct, but the test stops when the enrolment target is reached rather than running fixed whole weeks.
    GPT-6.1 Sol · API · Showing the delivery fee up front
  3. Guardrails with thresholds59% pass
    Guardrail metrics are listed but no specific thresholds (e.g., maximum acceptable drop in AOV) are given to block rollout.
    GPT-6 Luna · API · Showing the delivery fee up front

Case viewer

Read the brief, then put up to three outputs side by side, each with the LLM judge’s verdict on every check. Highlights mark what a PM had to fix.

The brief

You're the PM for checkout at Basketful. We want to test showing the delivery fee in the basket, instead of only at the last step of checkout. Write the experiment spec for Ravi, the growth engineer who'll build it, Ines, the analyst who'll read it out, and Chloe, our Head of Growth, who approves it. Keep it under 700 words. The team's draft plan is below. Fix what needs fixing.

What the model was given5 items: About Basketful, Why we're testing it, Traffic and baselines, Smallest change worth acting on, The team's draft plan
About BasketfulAn online grocery service. Delivery costs £3.99, or is free on orders over £60. Today the fee first appears on the final checkout step.
Why we're testing it12% of support contacts last quarter were about delivery fees the customer hadn't expected. Design believes showing the fee early will build trust; Chloe worries it will put people off before they start checkout.
Traffic and baselinesAbout 40,000 different customers view their basket each week. About half of each week's basket viewers are new that week; the rest came back from earlier weeks. 8% of basket viewers place an order. Of customers who start checkout, 62% complete it. Average order value: £47.
Smallest change worth acting onChloe: “Anything smaller than half a percentage point on orders isn't worth arguing about.”
The team's draft planPrimary metric: checkout completion rate (orders ÷ checkout starts). Split: 50/50 by session. Duration: we'll check the dashboard every day and stop as soon as it's significant. Chloe's addition: “While we're at it, let's put the free-delivery-over-£60 banner in the variant too.”
What a strong answer doesThe answer key the graders mark against

A spec that tests one change and fixes the draft's three problems. The primary metric is orders per basket-viewing customer, not checkout completion rate: showing the fee earlier changes who starts checkout, so completion can rise while orders fall. Customers, not sessions, are randomised, so returning customers see one version. The free-delivery banner stays out of the variant (or gets its own test), or the result can't be attributed. The duration is fixed in advance, with no daily stopping: detecting 0.5pp on an 8% base needs roughly 47,000 customers per arm (about 95,000 in total), and because half of each week's viewers are returning customers, new customers accumulate at about 40,000, 60,000, 80,000 and 100,000 after four weeks, so it runs about four full weeks, not two and a half. Guardrails with thresholds: average order value, fee-related support contacts, and the checkout-start rate as a diagnostic. A sample ratio check before reading results. A decision rule written up front: ship if orders are no worse than 0.5pp down and fee contacts fall; don't ship if orders drop by 0.5pp or more; with a stated plan for an inconclusive result.

Critical failures (cap the score at 40)
  • Uses checkout completion rate (orders ÷ checkout starts) as the primary metric
  • Bundles the free-delivery banner into the variant
Case

Mid-level · v1.1 · synthetic · B2C, ecommerce, pricing, metric choice

The outputs

7 setups answered this brief. Pick up to three to read side by side.

GPT-6 LunaAPI83Usable with a quick editGPT-6.1 SolAPI81Usable with a quick editSonnet 5.5API79Needs more workOpus 5.5Claude73Needs more workGPT-6 AstraChatGPT71Needs more workGemini 3.8 FlashAPI42Needs more workGemini 3.5 Flash-LiteGemini23Critical failure
Gemini 3.8 FlashAPI

Needs more than a quick edit3 of 12 checks passed

A PM would need to redo the sizing and duration, remove or redesign the early kill rule, add a sample-ratio check, and complete guardrail thresholds. These are substantive reasoning changes, not quick wording edits.

Output

Experiment Spec: Delivery Fee Transparency in Basket

To: Chloe (Approval), Ravi (Build), Ines (Analysis) From: Checkout PM

---

1. Objective & Hypothesis

  • Context: 12% of support contacts complain about surprise delivery fees at final checkout.
  • Hypothesis: Displaying the delivery fee (£3.99, or "Free" if >£60) upfront in the basket will set accurate expectations, reduce support contacts, and build customer trust without reducing net orders.
  • Risk (Chloe’s concern): Upfront cost shock could cause drop-off before checkout begins.

---

2. Corrections to the Draft Plan

1. Primary Metric Changed: Draft used Checkout Completion Rate (Orders ÷ Starts). Correction: Use Basket-to-Order Conversion Rate (Orders ÷ Basket Viewers). Measuring only checkout starts introduces severe selection bias—if fee visibility deters low-intent users at the basket, checkout completion would artificially increase while overall orders plummet. 2. Unit of Randomization Changed: Draft used session-level. Correction: Randomize by User ID (falling back to persistent anonymous visitor ID). 50% of weekly viewers are repeat shoppers; session-level randomization would expose returning users to conflicting experiences. 3. No Peeking / Fixed Duration: Draft suggested stopping once "significant." Correction: Continuous monitoring with early stopping inflates false-positive rates to >30%. We will fix sample size upfront and run for complete weekly cycles. 4. Scope Control (Chloe’s Banner): Cut the promotional free-delivery banner from this test. Adding a promotional banner introduces a confounding variable, making it impossible to isolate whether fee transparency helped or hurt. We will test the banner in a follow-up experiment.

---

3. Variants (For Ravi)

  • Allocation: 50/50 split at User/Persistent Device ID level. Triggered on basket view.
  • Control (A): Current basket. No delivery fee shown (fee appears on final checkout step as today).
  • Variant (B): Add a delivery fee line item to the Basket Order Summary:
  • Basket < £60: `Delivery: £3.99`
  • Basket ≥ £60: `Delivery: FREE`
  • Include delivery fee in the displayed Basket Total.
  • Note: Do not add promotional progress bars or marketing banners.

---

4. Metrics Framework (For Ines)

  • Primary Metric: Basket-to-Order Conversion Rate (Total Unique Ordering Users ÷ Total Unique Basket Viewers).
  • Guardrail Metrics:
  • Average Order Value (AOV baseline: £47).
  • Basket-to-Checkout Start Rate (tracks where abandonment occurs).
  • Total Net Revenue (Orders × AOV).
  • Secondary / Diagnostic Metrics:
  • Delivery-fee-related customer support ticket rate per 1,000 orders.
  • Checkout Completion Rate (Checkout Starts → Orders).

---

5. Sample Size & Run Time (For Ines & Chloe)

  • Baseline Conversion: 8.0% of basket viewers place an order.
  • Minimum Detectable Effect (MDE): Absolute ±0.5 percentage points (7.5% or 8.5%, a 6.25% relative change), per Chloe's threshold.
  • Statistical Parameters: $\alpha = 0.05$ (two-tailed), Power ($1 - \beta$) = 80%.
  • Required Sample: ~47,000 unique users per variant (~94,000 total unique basket viewers).
  • Duration: 3 full weeks (21 days).
  • With 40,000 viewers/week (~20k new, ~20k repeat), 3 weeks yields ~80,000–90,000 unique basket viewers across three full grocery-shopping cycles.
  • Ines will run the read-out strictly after the 21-day window closes.

---

6. Decision & Rollout Criteria (For Chloe)

  • Ship Variant: Primary conversion does not drop by more than Chloe’s threshold of -0.5pp (non-inferiority) AND delivery-related support contacts decrease significantly.
  • Iterate / Re-test: Basket-to-order conversion drops by >0.5pp. In this scenario, early fee disclosure hurts conversion, justifying Chloe’s concern; we then test the £60 threshold banner/progress bar to offset drop-off.
  • Kill Immediately: Overall revenue drops >5% statistically significantly at the end of Week 1 (safety check only; not an early win stop).

Claims the judge couldn’t find in the brief

  • Three weeks yields about 80,000–90,000 unique basket viewers.
  • Early stopping inflates false-positive rates to more than 30%.
  • The variant's free-delivery threshold is £60 or more.

Grades and run details

Decision model 67 · LLM judge 3 of 13 checks
Decision model checks
  • failUses the supplied evidence correctly55%
  • passAddresses the actual decision63%
  • partialRespects explicit constraints20%
  • passIdentifies material uncertainty65%
  • partialAvoids unsupported claims49%
  • partialProduces the required deliverable32%
  • passTests one change at a time100%
  • passFixes the duration up front66%
  • passAn unambiguous primary metric28%
  • partialDecision rule written before the test79%
  • partialSized from the real traffic94%
  • partialGuardrails with thresholds90%
Run
Run
#1
API response time
19 s
Submitted
1 Oct 2026

Check by check

Open a check to read the judge’s reasons. “Mixed” means the two graders disagreed.

Got wrong 7

Uses the supplied evidence correctlyWrong

It misstates the free-delivery threshold as ≥£60, overstates unique viewers over three weeks, and adds an unsupported >30% false-positive claim.

Respects explicit constraintsWrong

It is under 700 words and addressed to the named readers, but it proposes a 21-day test that cannot reach the stated sample and includes a week-1 significance-based kill rule.

Avoids unsupported claimsWrong

It presents the >30% false-positive rate and the 80,000–90,000 unique-viewer range as established facts without support.

Produces the required deliverableWrong

The spec is usable in form but has major gaps: the duration is underpowered, the sample-ratio check is missing, and the guardrail thresholds are incomplete.

Decision rule written before the testWrong

It does not define the action for an inconclusive result between the non-inferiority and harm thresholds.

Sized from the real trafficWrong

It calculates about 94,000 unique users but sets only three weeks, which yields about 80,000 unique users, not four weeks.

Guardrails with thresholdsWrong

It names guardrails but gives thresholds only for revenue, not for AOV, checkout-start rate, or support contacts.

Mixed 2

Fixes the duration up frontMixed

It fixes a duration and explains peeking, but the duration is not derived from the required sample and it still allows a week-1 significance-based stop.

An unambiguous primary metricMixed

It names one primary metric and rationale, but does not plan a sample-ratio check or other trustworthiness check.

Got right 3

Addresses the actual decisionRight

It commits to testing fee transparency alone, with ship, iterate, and kill rules tied to named thresholds.

Identifies material uncertaintyRight

It names the main uncertainty—whether early fee disclosure reduces orders—and maps outcomes to actions.

Tests one change at a timeRight

It keeps the promotional free-delivery banner out of the variant and explains that bundling it would confound attribution.

Results

Every setup we’ve tested on this task, across all cases and repeats, graded on the current checklist. Calibrated: the graders match our PM on 90% of checks.

#Model · HarnessTask scoreDecision modelLLM judgeRunsCritical failures
1GPT-6.1 SolAPI87.588.52None
2Sonnet 5.5API81.388.52None
3Opus 5.5Claude85.476.92None
4GPT-6 LunaAPI85.473.12None
5GPT-6 AstraChatGPT87.565.42None
6Gemini 3.8 FlashAPI60.438.52None
7Gemini 3.5 Flash-LiteGemini37.515.422 capped

The PM job

Specifying an A/B test before running it.

Why it matters

Most tests are underpowered, or measure a metric nobody agrees is good.

What good looks like

  • One primary metric everyone agrees on the direction of
  • Decision rule stated up front
  • Guardrails
  • Power considered

Deliberately not measured

    Capability tested

    Test design

    The failure we’re looking for

    A test with no decision attached

    Grading

    Decision model and LLM judge, calibrated against a blind PM review