Explainer Randy's PM Frameworks

Project Framework — Product Management

The six D's applied to PM work: product and feature steps, the 2018 originals, the measure-and-learn cycle then and now, scripts vs. user stories, and proposed revisions. Companion to "Project Framework — Car is IE".

Generated
November 18, 2026, 3:22 PM MT
Created
November 18, 2026, 2:24 PM MT
Built from
  • wiki/concepts/six-d-framework.md
  • wiki/concepts/script-writing-process.md
  • wiki/sources/project-framework-product-management.md
  • wiki/sources/project-framework-car-is-ie.md
  • wiki/concepts/circles.md
  • wiki/concepts/measurement-plan-rubric.md

Restates your 2018 "Car is IE" one-pager for product work. It was written with Claude on 2026-11-14/15 at your request ("using the same methodology… what a PM would do"), revised after a self-critique the same night, and published as a web page.

How you see it now: the 2018 version is a quick mnemonic for framing a project. It's outdated as an artifact, but the concept holds, and this 2026 PM version is the current form.

Start at Zero, end documented.

Where "Car is IE" comes from

PPlanning / info gathering
CCommitment
AAnalysis
RResearch
IIncubation
SSelection
IImplementation
EEvaluation

The project steps after planning spell CARIS IE, read as "Car is IE" (confirmed 2026-11-16). Traced to an email thread dated 2018-03-16, when you were a Customer Success Manager at Eggplant. The file itself doesn't explain the acronym.

Product management at a glance

1 DiscoverProblem statement + evidence
2 DefineSuccess metric + scope
3 DesignRequirements + prioritized backlog
4 DevelopAccepted stories
5 DeployRelease + go-to-market
6 DocumentRelease notes + decision log

Use this sequence for any product work. Each step ends with the output shown. The six D's date to at least 2018-03-16. The sections below apply them to a whole product, a single feature, the measure-and-learn loop, and user stories.

Managing a product · Launching a feature · The 2018 originals

Managing a product (2026)

  1. Market and customer research (Discover)
  2. Opportunity analysis (Discover)
  3. Vision and strategy (Define)
  4. Stakeholder commitment (Define)
  5. Roadmap — outcomes, not features (Design)
  6. Delivery — backlog to shipped (Develop)
  7. Launch and enablement (Deploy)
  8. Measure and document — back to 1 (Document)

Launching a new feature (2026)

  1. Understand the current experience (Discover)
  2. Document the problem and workflow (Discover)
  3. Validate with customers and the team (Define)
  4. Define success — the metric that moves (Define)
  5. Design and build (Design, Develop)
  6. Prepare users — help, support, sales (Deploy)
  7. Release in stages — beta, then all (Deploy)
  8. Review and document — notes + retro (Document)

Implementing a project (2018)

  1. Planning / information gathering
  2. Commitment
  3. Analysis
  4. Research
  5. Incubation
  6. Selection
  7. Implementation
  8. Evaluation

New business workflow (2018)

Understand and document the current process → verify it by interviewing the team → suggest improvements → build and test → train users → roll out in steps → review.

2018 vs. 2026

The 2018 lists aren't tagged to the D's and have no Document step. The 2026 lists tag every step and end in Document.

Product states: the measure-and-learn cycle, then and now

2018 · System states of an automation cycle Zero Alter / Configuration Data-driven: gather all info before running Testing Diagnostic & Reporting Zero / Cleanup cleanup returns the system to Zero 2026 · Product states: every product bet starts and ends at Zero Zero Baseline metrics captured Hypothesis Bet + success metric written Experiment Flag, beta or design partners Measure Usage data and customer input Learn / Reset Keep, kill or iterate results become the new baseline
Both cycles start and end at Zero. The 2018 cleanup returns the system to Zero, and in 2026 the results become the new baseline.

Never ship without a baseline, and never close a bet without resetting it. In B2B, design partners are often the only slice big enough to learn from, so Experiment includes them and Measure uses customer feedback alongside usage data.

The 2018 automation cycle is the earliest dated version of your automation repeatability pattern (Zero → … → Zero). The 2026 product states are a PM reading of it.

Writing scripts (2018) → writing user stories (2026)

#Scripts (2018)User stories (2026)
1Review the use caseReview the customer problem — evidence before solution
2Run through the manual workflowWalk the user's workflow by hand
3Define prerequisites, so anyone unfamiliar can follow itDefine prerequisites — so anyone new can follow them
4Automate procedurallySlice it — thin stories that ship on their own
5Make the automated test run consistentlyWrite each story plainly — plus acceptance criteria
6Optimize: conditionals and loopsMake it testable — same result every time
7Variablize: make it data-drivenOptimize — edge cases and error states
8ModularizeVariablize — tie it to a metric or event
9Parameterize, so others can use your codeParameterize — a template others can reuse
10—Document — link to decision, release notes, results

The pattern

Get it working once by hand, then make it consistent, then make it reusable.

Key difference

Code is refactored into modules after it works, but stories are sliced before they're written. Otherwise the acceptance criteria and tests get redone.

"Define variation upfront" in the automation repeatability pattern is the same instinct as Variablize.

Proposed revisions — 2026-11-18 review

Added at your request; proposed, not yet worked into the six D's above.

  1. Give measurement its own step. The six D's end at Document; measuring lives only in the Zero cycle or inside "measure and document". Outcomes over outputs argues for a Measure step, or Document read as "measure, then document" (see Measurement plan rubric).
  2. Add a go/no-go gate between Define and Design. Prioritization is buried in Design ("prioritized backlog"). An explicit kill point is where RICE, WSJF, Kano and MoSCoW plug in.
  3. Pick one customer segment in Discover. The one step CIRCLES (Lewis Lin's, not yours) adds that the six D's don't spell out.
  4. One public name. The 2026 content is the six D's; keep CARIS IE ("Car is IE") as its 2018 origin story.
  5. Canonical copy. The wiki is the source of truth; the web page, editable doc and Word file are views of it.
Read together

One missing step (Measure) and one missing gate (go/no-go). The structure itself holds.

Revision history · Open questions · Related

2026-11-15 self-critique

  1. Opportunity analysis moved before vision, so the D-tags run in order.
  2. "Slice it" moved from step 8 to step 4.
  3. Experiment and Measure reworded for B2B (design partners, customer feedback).
? Open
  • Which copy is canonical? Proposed answer (revision 5): the wiki; the other copies are views.
  • The 2018 project, workflow and scripting lists aren't tagged with the six D's, and none has a Document step (noted in the 2026-11-14 review).

Related

PM decision frameworkMaking a single product call
Problem definition templateThe Discover output in depth
Automation repeatability patternZero → … → Zero
Script writing processScripts and user stories
CIRCLESEstablished design framework (Lin); fits inside Discover → Design
Established FrameworksRICE, WSJF, MoSCoW, Kano and more: their own group on the home page