# Product Garage: operating playbook for Brand Book 3.1

This file contains the forms and decision procedure behind the [edition 3.1 master brand book](index.html). The book sets the current brand standard; this playbook makes it repeatable. Its templates originated in 3.0 and remain in use alongside the [3.1 change log](CHANGELOG-v3.1.md). Where this playbook and the book differ, use the 3.1 book for brand direction and the current event sheet for event facts. Neither replaces legal review or a founder's decision. Updated 24 September 2026.

## 1. What is decided, observed, working or open

Use these exact statuses in the decision ledger, asset register and event sheet.

| Status | Use it when | Current example |
| --- | --- | --- |
| Confirmed | A founder decision or formally approved operating rule exists | Name is Product Garage; working sessions are flagship; premium balances room and work; October has 12 participants |
| Observed | A traceable event, document or direct answer exists | The live October listing describes two founders and four teams of three |
| Working | The team has a clear recommendation it can execute now | Casebook layout, shutter reveal and documentary proof; two-story-unit media pilot |
| Open | A decision or external check is still required | Vector logo, filming policy approval, final business model and validated public descriptor |

These labels are not a permission to hedge ordinary copy. The public should hear a clear offer. Internally, the label prevents preference from masquerading as research.

## 2. Initial source and decision ledger

The ledger is intentionally short. Add the original briefs, session boards and follow-ups before expanding any outcome claim.

| ID | Source | What it supports | What it does not support | Status |
| --- | --- | --- | --- | --- |
| PG-01 | Founder decisions recorded in this project, 23–24 Sep 2026 | Name, flagship offer, October capacity, balance of premium, palette, type, recruitment CTA and media ambition | That the public understands or prefers the chosen descriptor | Confirmed |
| PG-02 | [Five-part brand-engine memo](../product-garage-brand-engine.md) | The Work → Exchange → Standard → Signals → Evidence framework and why it was chosen | Proof that every part is already delivered consistently | Confirmed framework |
| PG-03 | [October public listing](https://luma.com/w8r14u3r), live checked after correction | Two founders, four teams of three and 12 participant seats | A permanent one-founder format, or business impact | Observed event fact |
| PG-04 | [Public first-session account](https://www.linkedin.com/posts/vikasnagprasanna_last-week-we-ran-the-first-product-garage-activity-7450822880315084800-zgk-) | A first-party description of live challenge and visible thinking | Independent causal proof of value | Observed self-report |
| PG-05 | [Claude Brand Book 2.0](https://claude.ai/artifact/Jpg1rV7FX6yMHCcUahhdG1) and [local Book 1.1](../product-garage-brand-book-v1.1/index.html) | Prior creative and operating proposals | Founder ratification of every proposal | Working references |
| PG-06 | Original past briefs, boards, decisions, permissions and 30-day actions | Not yet assembled in this workspace | Any outcome or impact assertion | Open evidence gap |

When a new claim is proposed, give it a source ID and a proof rung. A source may be true and still be insufficient for the claim being made.

## 3. Discovery: collect the messy evidence

### Interview sample and recruiting

Run 10–12 short interviews as a directional first pass: roughly one-third founders or decision owners, one-third relevant operators, and one-third people who consume product content but have never attended. Do not only recruit fans. Interview partners separately if their role becomes central to growth.

Begin with actual behaviour and only then show Product Garage. Use the same order for each interview.

1. “Tell me about the last product or go-to-market decision that stayed unresolved longer than it should have.”
2. “What did you do to get unstuck? What else did you consider?”
3. “What made any outside input useful or useless?”
4. “When would you give up a morning, or give someone else access to a live company problem?”
5. Show the undecorated public descriptor. Ask: “What is this? Who is it for? What would happen?”
6. Ask: “What would make you believe the room or content was worth your attention?”
7. Show the visual synthesis and a simpler control without explaining the concept. Ask what each promises and which is more credible.
8. One day later, ask what visual cue, phrase or offer they remember unaided, if practical.

Record exact language, the person's audience segment, date, consent to retain notes, and relevant context. Code answers for actual alternatives, trigger situation, desired progress, proof threshold, objection and spontaneous wording. Keep disconfirming responses visible. Ten interviews can reveal confusion; they cannot statistically certify a market position.

### Positioning worksheet

Follow [April Dunford's sequence](https://www.aprildunford.com/post/a-quickstart-guide-to-positioning) for each audience, not one generic positioning sentence.

| Step | Question | Evidence bar |
| --- | --- | --- |
| Actual alternative | What did this person do the last time Product Garage did not exist for them? | Recent action, not a competitor brainstorm |
| Difference | Which capability can Product Garage deliver repeatedly that the alternative cannot? | Operating record, not a promise |
| Value | Why does that difference matter in the person's situation? | Interview answer or follow-up, not only internal logic |
| Best fit | Who cares most, and whom should we not target? | Pattern across suitable people |
| Category | Which familiar words create the right expectation? | Unprompted comprehension and contrast with alternatives |

Candidate public descriptor: “Product Garage shows how product and go-to-market decisions get made. Anyone can learn from the public work. Selected operators can join working sessions that examine a company’s live challenge with its decision owner.” Test this against a shorter “product thinking network” version and a “working-session community” version. Keep the wording people can explain back accurately, not the most elegant sentence on a slide.

### Decision record

```text
ID, owner, date, review date:
Decision to make:
Audience and actual situation:
Evidence used, with source IDs:
Options considered:
Strongest case against the preferred option:
Choice and why:
What would disprove it:
Assets, scripts or behaviours that change:
Who approved the change:
```

## 4. Creative testing and distinctive assets

Brief two executions of each touchpoint: the chosen Casebook + Shutter + Documentary synthesis and a deliberately simpler control. Keep the words and information constant so the visual treatment is the main difference. Test a site hero, case carousel, video cover and one private decision record.

Ask four questions without explaining either execution: what Product Garage does, whom it serves, what they would expect to receive, and what they remember the next day. Record confusion as carefully as preference. Separately score production repeatability and legibility with the team. These are judgement scores, not audience results.

Do not call slatted light or the red mark a “distinctive asset” yet. Record recognition without a wordmark and misattribution to other brands after repeated real exposure. [Ehrenberg-Bass](https://marketingscience.info/news-and-insights/brands-of-distinction) distinguishes designed brand elements from assets that are actually associated with one brand.

## 5. One event sheet feeds every public surface

Create an event sheet before Luma, a site page or a social post goes live. Each field has one named owner and a last-checked date.

```text
Event ID / revision / owner / checked on:
Public title and exact brand name:
Format: working session / partner working session / open event
Date, start, end, arrival, venue, accessibility contact:
Partner names and roles:
Decision owners, companies and naming permissions:
Challenge question(s) and restrictions:
Participant capacity, team count and team size:
Application open and close dates, decision date:
Participant expectations and what they receive:
Capture notice URL, approval status, audio/video plan:
Primary operator CTA and form URL:
Canonical description and shortened variants:
Source links and approvals:
```

October's verified public facts are 12 participants, four teams of three and two founders. This does not define every future room. Reconcile any listing against the sheet before publication and after each material change.

## 6. Accept, prepare and run a working session

### Challenge acceptance gate

All must be true, whether there is one founder or more than one:

1. There is a named decision owner for each challenge.
2. Each question can be stated in one sentence and bounded to a morning's useful work.
3. At least two plausible options or strategic paths exist.
4. Participants can receive enough safe context to reason responsibly.
5. Owners will answer hard questions and consider an unwelcome recommendation.
6. Private and public disclosure boundaries can be set in advance.
7. A follow-up can ask what the owner did, without obliging them to adopt the advice.

If a challenge is too broad, narrow it before selecting operators. If it is a disguised pitch, recruiting exercise, request for execution labour or demand for endorsement, decline or offer a separate service. A session with insufficient context should be rescheduled, not improvised.

### Brief and facilitation board

Send a 2–4 page brief about five days before the session. It should contain owner, exact decision, why now, known facts with sources, assumptions, constraints, options considered, what useful contribution looks like and what may not be recorded or published. Ask operators to bring two questions and one assumption to challenge, not completed unpaid strategy work.

The board has seven areas: **question, facts, assumptions, options, trade-offs, recommendation, what would change the recommendation**. Mark significant statements fact, inference or opinion. Let teams disagree in the final record. If new information invalidates the brief, pause and reframe rather than pushing through the planned agenda.

### Decision record within 72 hours

```text
PRIVATE / event ID / decision owner / date
Original question:
Facts the teams relied on, with sources:
Important assumptions not verified:
Options compared:
Preferred option and the reasons for it:
Strongest objection or dissent:
Evidence that would change the recommendation:
Next test or action, owner and date:
Information restricted from public use:
Founder approval of factual accuracy:
```

Create the public case separately. It may omit names, numbers and sensitive reasoning. At about 30 days, record adopted, adapted, rejected or deferred, plus the owner's reason. No follow-up answer means “unknown”, not “implemented”.

### Service recovery

| Failure | Response |
| --- | --- |
| Brief late or thin | Narrow or reschedule. Inform operators promptly. |
| Wrong room mix | Recruit the missing perspective or reduce parallel work. |
| Founder changes the decision mid-room | Pause, reframe and record what changed. |
| A participant dominates or sells | Facilitator redirects to the question and invites others. |
| Privacy boundary crossed | Stop the relevant recording and isolate the material; consult the capture owner. |
| Founder refuses public disclosure | Deliver the private record; publish only independent teaching content. |

## 7. Media production that four volunteers can actually run

The 30-day starting recommendation is **two core story units per week**: one researched case/teaching piece and one operator perspective. Adapt these for LinkedIn and Instagram only when appropriate. Send an email edition every other week. If the team's reliable capacity is under eight hours weekly, reduce to one core story unit. Other channels have no mandatory floor in this pilot.

Edition 3.1 distinguishes this pilot from the confirmed direction of two LinkedIn articles per week: the pilot supplies one written article, and the second is the first addition if the expansion rule is met. Track filming, editing and approval time separately from writing time when making that decision.

Assign actual names and hours before committing to the schedule:

| Function | Minimum weekly task | Time to log |
| --- | --- | --- |
| Research and case source | Verify the question, facts and permission boundary | Actual hours |
| Writer or interviewer | Create one clear narrative or recorded conversation | Actual hours |
| Editor/designer | Produce native LinkedIn and Instagram versions where useful | Actual hours |
| Approval owner | Check claims, rights, brand and scheduling | Actual hours |

One person may cover more than one function, but the hours must still exist. Keep a two-week buffer before increasing frequency. Review qualified audience response, not only reach. Do not add YouTube long-form as a deadline until there is approved footage, an owner, an edit estimate and a repeatable release process.

### Editorial case brief

```text
Audience and entry point:
Decision question:
Why the obvious answer is insufficient:
Known facts, each with a source:
Assumptions and what would test them:
Options, including the rejected one:
Trade-off and recommendation:
What happened next, if known:
What is safe to publish:
One reader-useful takeaway that does not require attendance:
Claim rung and approvers:
Primary format, adaptation formats and measured hours:
```

### Message test record

Test two genuinely different headlines or hooks on comparable distribution when possible. Log the premise, versions, audience, dates, reach conditions, comprehension, saves or relevant replies, and what changes next. Never infer a creative winner from an unbalanced one-post comparison. This is the practical Ogilvy layer.

## 8. Capture and consent: the production brief for legal review

The [Irish Data Protection Commission's event photography guidance](https://www.dataprotection.ie/en/faqs/topical-data-protection-issues/when-i-am-attending-public-event-can-organisers-take-promotional-photographs-me-without-my-consent) says identifiable images are personal data and recommends particular care at smaller events, including consent or a way to identify consent. It does not settle all questions about video, audio, company information, lawful basis or retention. Irish counsel should approve the actual process before a filmed session.

The capture owner must be able to answer the following before cameras enter the room:

```text
What will be captured: stills, silent video, speaker audio, interviews, board footage?
For what purposes and on which platforms?
What is the legal basis for each type of processing?
What does the participant know before applying and before entering the room?
How can someone participate meaningfully without being filmed or recorded?
What do coloured lanyards or other signals mean to the crew? Do they expose a preference publicly?
Where may cameras point? Which tables, screens or materials are restricted?
Can microphones capture people who chose not to be recorded? If yes, redesign the audio plan.
Who approves clips showing company material or identifiable speech?
Who may access raw files, for how long, and how are objections handled?
```

Prefer controlled angles, approved close-ups and separately permissioned interviews over blanket full-session recording. A lanyard can help a camera operator, but it cannot make a mixed room camera-safe or keep someone out of an ambient microphone. If the planned shot or sound cannot honour a person's choice, do not take it.

## 9. Claims and proof ledger

| Rung | Claim type | Evidence required | Example of overreach to reject |
| --- | --- | --- | --- |
| 1 Fact | What happened | Event sheet, register, approved output | Calling 12 seats “16 participants” |
| 2 Response | What someone said | Exact quote, context and permission | Turning “I enjoyed it” into “the company improved” |
| 3 Action | What changed afterward | Owner follow-up with date and permission | Claiming adoption because a recommendation was presented |
| 4 Impact | What it caused | Timeframe, metric, credible attribution and permission | Claiming revenue impact from a good room without evidence |

For each public claim, record: exact words, source ID, rung, owner, evidence path, who may be named, approval date, channels, review/expiry date and any later correction. If a case is anonymised, check that details do not still identify the company or person.

### Monthly scorecard

| Dimension | Measure | Decision it informs |
| --- | --- | --- |
| Experience | Founder utility, operator value, brief readiness, 30-day action | Improve the room or narrow the promise |
| Acquisition quality | Suitable applications and challenge submissions by source | Adjust distribution and descriptor |
| Media usefulness | Saves, relevant replies, return use, email conversions by series | Keep or stop a format |
| Memory | Unprompted offer comprehension and cue attribution | Change or invest in identity cues |
| Capacity | Hours per story unit, approval lag, backlog | Set the next publishing floor |

Set a baseline before numerical targets. The long-term million-follower ambition is a reach north star, not a proxy for all five dimensions.

## 10. Asset register and release gate

Use one register for logo masters, concept imagery, photographs, clips, case templates and event artwork. Each row needs asset ID, path, source, owner, version, rights, people depicted, company material, approval, expiry/review and archive status. Label all three moodboards **AI-generated concept direction**. Never use them as a “from the room” image.

Before an asset ships, check six questions:

1. Does it serve one audience and one clear job?
2. Is the headline in Product Garage's direct, complete-sentence voice when body copy is needed?
3. Are the name, event facts and CTA drawn from current approved sources?
4. Is every claim on an evidenced rung and every person/material cleared?
5. Is the visual system legible, accessible and recognisable without pretending a cue is already famous?
6. Does it leave something reusable: a tool, case, subscriber, proof point or improved room standard?

Routine copy and design can be approved by the editor/designer against this gate. Position and identity changes go to the founder group. Company cases require challenge-owner approval. Capture requires the additional legal and operational gate. Archive replaced assets; do not silently overwrite history.

## 11. Open business decisions

Brand guidance cannot invent a business model. The company still needs a deliberate decision on who pays, what is free, how partners participate, what access costs or sponsorship changes, and which income stream supports the public media ambition. Until then, do not equate premium with high price or promise a permanent free format. Test any model against the founder, operator and public-learner exchange, and protect the independence of recommendations.

## 12. Review sequence

1. Before the next filmed session: approve the capture and audio plan.
2. Before a permanent identity rollout: approve the vector mark, red match, fonts and accessible templates.
3. Within 30 days: gather the original case sources, run the interview and comprehension sprint, and compare the chosen creative system against a simpler control.
4. At day 30 of publishing: set the real media floor from measured hours and qualified response.
5. At day 90: review whether the Door, Community and Room each deliver their own promise, and update the master book with recorded decisions rather than a cosmetic version bump.