---
id: "16.1"
title: "Research plan"
dept: research-and-insights
status: draft
owner: Param
branded: false
last_verified: 2026-10-26
sources:
  - "brand-directions/book/content/d2.js:916-1144, section d-12 what-we-know, what-we-dont, how-we-test"
  - "brand-directions/book/content/b.js:329-349 b-5 alternatives; b.js:502-523 b-8 find-out; b.js:664-685 b-10 partnership-goals"
  - "brand-directions/_research/product-audit.md:7, 322-335"
  - "brew origin/main a559de6: feed-service.js:1262-1275, feed-service.js:2056"
  - "memory: brew-live-stores, brew-model-compare-2026-10-20, ape-consumer-research, no-spend-while-building, treg-usage-rules"
  - "handbook: product/jobs-to-be-done.md, product/personas.md, product/design-partner-program.md, product/partner-feedback-log.md"
---

# Research plan

## Answer

What we must learn from the five design partners before pricing and launch, how and from whom; nothing on this plan has been run yet, so the call it settles is which questions go first.

## Today

| What is true now | Source |
|---|---|
| No research plan has been agreed or run; the old book drafted test scripts for a brand audition | [src: brand-directions/book/content/d2.js:916-1144] |
| No Brew partner call or interview has a dated record | [src: handbook research-and-insights/interview-and-call-library.md] |
| Five real partner prospects: Arata, Nutrova, Mana, Puma India, Scrubroom; Cozy and Zenant are our test brands | [src: memory brew-live-stores] |
| What we know about the build comes from code, not from people: an audit of origin/main b68aa9f on 25 Oct | [src: brand-directions/_research/product-audit.md:7] |
| Eight sizing questions are written and none is answered | [src: brand-directions/book/content/b.js:502-523] |
| Six outcomes the design partnership must answer are written and none is met | [src: brand-directions/book/content/b.js:664-685] |
| Eight usability questions, each with a test, none run | [src: brand-directions/book/content/d2.js d-12 what-we-dont] |
| One model study was run on 20 Oct, 54 pairs of Arata Plays, and its inputs were not equal | [src: memory brew-model-compare-2026-10-20] |
| The product keeps hearts, saves, Not for me and written card notes per brand | [src: brew/feed-service.js:2056] [src: brew/feed-service.js:416-417] |
| An engine view lists time to feed, return days, card action rate, approvals and card feedback per brand | [src: brew/feed-service.js:1262-1275] |
| Our old voice interview product, APE, still runs at oldsu.joinsquareup.com; it has never run a study for Brew | [src: memory ape-consumer-research] |
| Paid probes need Param's yes first, and treg is for data only | [src: memory no-spend-while-building] [src: memory treg-usage-rules] |

## Tomorrow

The questions for this quarter, ranked as a proposal.

| # | Question | Method | From whom | Version | Tag |
|---|---|---|---|---|---|
| 1 | Is the main job right, said in their words? | Two questions on the weekly partner call | One decider at each of the five | v1 | New |
| 2 | Do founders get what it does in five seconds? | The instant understanding test in [Usability tests](/research-and-insights/usability-tests) | Five founders or marketers | v1 | New |
| 3 | Where does a new person stop between paste and first hand off? | Script 1, timed | Five people new to the product | v1 | New |
| 4 | Do brands hire us for the Plays documents or for the feed and drafts? | Count Plays PDF opens against jams for four weeks | Our own logs | v1 | New |
| 5 | What do they use and pay for marketing help now? | One question on the first call | The five | v1 | New |
| 6 | Which picks get handed off, by area and source? | Logs, then ask why on the call | The five | v1 | New |
| 7 | What does one brand cost us per week? | Our own cost logs per brand | Our logs | v1 | New |
| 8 | How many own brand stores in India run ads and show reviews? | Desk counts, see [Market and category research](/research-and-insights/market-and-category-research) | Public sources | v1 | New |
| 9 | Which model writes the best Plays? | Blind bake off: the same five Plays on Opus, Pro, Flash and two open models | Scored by our team | v1 | New |
| 10 | Which country after India? | The same counts for the next candidates | Public sources | v2 | New |

## How we do it

1. Give each question an owner, a method, the partner and a start date before it starts.
2. Ask on the weekly partner call proposed in the [design partner program](/product/design-partner-program); keep the same two or three questions for six weeks so answers compare.
3. Log every call the same day in the [interview and call library](/research-and-insights/interview-and-call-library) and every partner item in the [feedback log](/product/partner-feedback-log).
4. Write each finding in the [insight repository](/research-and-insights/insight-repository) with its evidence and what it changed.
5. Quote a partner only with permission. Never invent a quote, a person or a finding.
6. Ask Param before any paid probe, and check the treg balance live before blaming it.

## Numbers

| Metric | Definition | Where measured | Status |
|---|---|---|---|
| Questions answered | Questions on this plan with a sourced answer | This page | Proposed; 0 of 10 today |
| Partners heard | Partners with a logged call in the week | Interview and call library | Proposed; 0 today |
| Days to answer | Days from a question's start to its finding | Insight repository | Proposed |

## Open

1. **Which three questions go first?** Proposal: 1, 2 and 3, all on the weekly call slot.
2. **May design partners be asked to test, and how often?** Proposal: one 30 minute session each in November, inside the weekly call.
3. **Who runs the calls?** Proposal: Param runs them, Sai or Arvind takes notes into the library the same day.
4. **Run the model bake off?** Proposal: first the fair rerun of five Plays on Pro, about $1, then the bake off, about $5 to $10.
5. **Is India first?** Proposal: run the India counts first and keep global open until they are in.