rampstackco/claude-skills / skills/discovery-research-synthesis
discovery-research-synthesis
VerifiedTurning research artifacts into actionable PM insight. Customer interviews, user research notes, support ticket reviews, sales call transcripts, survey data, in-app feedback, all synthesized into the decisions they are meant to inform. The discipline of moving from raw discovery data to clear product direction without losing signal in the synthesis or fabricating insight that was not actually there. Triggers on research synthesis, customer interview synthesis, user research analysis, discovery readout, research insights, sales call analysis, support ticket analysis, qualitative data analysis. Also triggers when a team has done research but cannot turn it into decisions, when synthesis is producing pretty decks but no roadmap movement, or when an upcoming PM decision needs to be grounded in research already conducted.
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 369 B |
| SKILL.md | skill md | 24.6 KB |
| references/common-discovery-synthesis-failures.md | markdown | 11.3 KB |
| references/from-pattern-to-product-implication.md | markdown | 9.8 KB |
| references/pattern-naming-patterns.md | markdown | 8.8 KB |
| references/research-types-and-when-each-fits.md | markdown | 10.4 KB |
| references/synthesis-review-and-validation.md | markdown | 10.7 KB |
| references/synthesis-sequence-walkthrough.md | markdown | 11.2 KB |
| references/tagging-and-clustering-discipline.md | markdown | 9.3 KB |
| references/when-to-gather-more-data.md | markdown | 11.0 KB |
| references/writing-for-decisions-not-decks.md | markdown | 10.4 KB |
references/writing-for-decisions-not-decks.md
View on GitHub# Writing for decisions, not decks
Synthesis output should be optimized for decision use, not for presentation. The format choice (document vs deck), the structure (patterns lead, evidence appendix), the voice (opinionated), and the closing (decision input section) compose to produce synthesis the team actually references.
---
## Document, not deck
**The principle.** Deck format is poor for synthesis.
**Why decks fail.**
- Slides cannot carry argument density. A pattern, its evidence summary, its implications, and its decision input do not fit on one slide; spreading them across slides loses coherence.
- Readers cannot reference slides later. A pattern needs to be findable; decks decay into icons and bullet points that are hard to search and cite.
- Decks are presentation artifacts; synthesis needs to be a reference artifact that the team uses for months.
- The deck format encourages summarization that loses precision.
**Why documents work.**
- A 4-12 page document can carry the pattern, evidence, implications, and decision input in connected prose.
- Documents are searchable, linkable, and citable.
- Documents accommodate density without overwhelming; readers can scan headings and dive into sections.
- Documents read as load-bearing analytical work; decks read as presentation.
**The exception.** A short executive summary deck (5-8 slides) can pair with a substantive document. The deck summarizes; the document carries the work.
**The discipline.** If the synthesis is being optimized for a single 30-minute readout meeting, the format will fail when the team needs to reference the synthesis in week 6 of execution. Optimize for the reference use, not the presentation moment.
---
## Lead with patterns and implications; relegate evidence to appendix
**The principle.** Readers scanning for decision input want patterns and implications first.
**The reader's question.** "What does this synthesis tell me to do?" Patterns and implications answer it. Evidence supports the answer but is not the answer.
**The structure that works.**
- Executive summary (1 page): the patterns, the recommendations, the decision input.
- Pattern sections (1-3 pages per pattern): pattern name, evidence summary in 2-3 sentences, implications, decision input.
- Recommendations section (1 page): overall synthesis recommendation; what should change.
- Appendix: detailed evidence, quote pulls, methodology notes, full transcripts or links to them.
**The structure that fails.**
- Decks that lead with quote walls. Readers admire the quotes; they do not extract decisions.
- Documents that bury the patterns under methodology and background. Readers stop reading before reaching the patterns.
- Synthesis that mixes evidence and analysis throughout. Readers cannot tell where observation ends and recommendation begins.
**The discipline.** Each pattern section is short (2-4 paragraphs of synthesis text); evidence linked or summarized briefly; the appendix is where evidence depth lives. Readers who want depth follow the link; readers who want decisions stay in the body.
---
## One section per pattern, parallel structure
**The principle.** Synthesis with 4-8 patterns has 4-8 parallel sections, each with the same structure.
**Per-section structure.**
- Pattern name (the heading).
- Evidence summary (2-3 sentences naming the artifacts and the dominant signals).
- Implication (1-3 implications with cost notes).
- Decision input (specific decision the implication informs).
**Why parallel structure works.**
- Readers learn the structure from the first section and can scan subsequent ones.
- Comparison across patterns is easier when each is presented the same way.
- Editing for completeness is easier (each section either has the elements or it does not).
**The variation.** Some patterns warrant longer treatment (richer evidence, more complex implications); some are shorter. Length variation is fine; structural variation is not.
---
## Be opinionated
**The principle.** Synthesis that hedges every pattern reads as a researcher avoiding accountability.
**Hedged synthesis.**
- "Users may sometimes occasionally seem to indicate possible difficulty."
- "It might be considered to potentially explore some adjustments."
- "There could be benefits to investigating opportunities."
These commit nothing. Readers cannot debate or decide on hedged synthesis.
**Opinionated synthesis.**
- "Users abandon configuration step 3 in 40% of signup attempts because credit card details are not accessible at signup."
- "The Q2 roadmap should prioritize the onboarding redesign over the dashboard expansion."
- "The not-act recommendation is to defer enterprise admin controls until small-team segment growth stabilizes."
These commit positions. Readers can disagree; disagreement produces decisions.
**The discipline.** Write synthesis as a recommendation. The PM doing synthesis is making analytical claims; readers may push back, but pushback is the point. Hedging avoids pushback by avoiding claim.
**The exception for honest uncertainty.** When the data genuinely does not support a position, name the uncertainty explicitly: "The evidence supports two possible interpretations; we recommend gathering more data before deciding." This is honest uncertainty, not hedging. Hedging is dressing up uncertainty as if it were synthesis.
---
## Include a decision-input section
**The principle.** What does this synthesis recommend the team do?
**The section's role.** The decision-input section connects the synthesis to the team's prioritization work. Without it, the synthesis presents patterns and implications but does not aggregate them into recommendations.
**What the section addresses.**
- Where should priorities shift based on this synthesis?
- Which roadmap items does the synthesis validate? Which does it challenge?
- Which decisions can the team make now with this synthesis as input?
- Which decisions need more research before making?
- What is the synthesis's overall view of where the product should focus?
**The section's voice.** Direct, specific, opinionated. The team can disagree, but the section should commit a recommendation.
**The discipline.** Without this section, synthesis is a collection of patterns; with this section, synthesis is decision input. The section is what makes the synthesis load-bearing for the team's strategic work.
---
## Length calibration
**The principle.** Synthesis should be as short as possible without losing the load-bearing content.
**Typical lengths.**
- Discovery cycle synthesis from a meaningful research batch: 6-12 pages.
- Smaller-scale synthesis from a single research type (e.g., support audit alone): 3-6 pages.
- Quick-turn synthesis from a focused interview batch: 2-4 pages.
**Length anti-patterns.**
- Synthesis padded to 30+ pages. Readers stop reading; the load-bearing content gets lost.
- Synthesis compressed to 1 page. Density too high; specifics get lost.
- Synthesis stretched across multiple documents without clear primacy. Readers do not know which document is the synthesis.
**The discipline.** Edit for length. Each pattern section should be tight; each implication should be specific; the recommendation should be concise. Readers will engage with a 6-page synthesis they can finish in 20 minutes; readers will skim a 30-page synthesis and miss the recommendations.
---
## The reference-after-shipping test
**The principle.** Synthesis that gets referenced months after publication has earned its keep.
**The test.** Six months after a synthesis ships, can the team:
- Name the patterns the synthesis surfaced?
- Cite the synthesis in roadmap discussions?
- Use the synthesis as input for related decisions that came up after publication?
**Failures the test reveals.**
- Synthesis nobody can name suggests insight-theater (presented but not absorbed).
- Synthesis nobody references suggests document-shelf-syndrome (filed away after readout).
- Synthesis the team contradicts in later decisions suggests the synthesis was not load-bearing in their actual prioritization work.
**Recovery.** When prior synthesis fails the test, the next synthesis should adjust: shorter, more opinionated, more directly tied to specific decisions. Each round of synthesis should be more usable than the prior.
---
## Visual elements in synthesis documents
**The discipline.** Visuals should carry argumentative load the prose alone cannot.
**Visuals that work.**
- A pattern-to-implication-to-decision diagram showing how the synthesis connects to product work.
- A frequency chart for patterns supported by quantitative data.
- A segment comparison table where patterns differ across segments.
- An evidence map showing artifact-source distribution per pattern.
**Visuals that fail.**
- Stock images decorating section breaks.
- Quote walls without context.
- Word clouds (rarely informative).
- Persona illustrations (decorative; not synthesis output).
**The frequency.** Visual elements every 2-3 pages of synthesis is typical. More than that risks decoration; fewer is fine if the prose carries the work.
---
## Common synthesis-writing failures
**Format choice.** Deck instead of document; the synthesis decays into bullets that lose precision.
**Buried recommendations.** Patterns are clear; the recommendation is one sentence on page 11. Readers miss it.
**Hedge-stacking.** Every claim caveated; readers cannot extract a position.
**No decision-input section.** Synthesis presents patterns; team has to aggregate them into decisions on their own; usually does not.
**Length excess.** 30 pages of synthesis where 8 would have served. Readers skim; load-bearing content gets lost.
**Length compression.** 1 page of synthesis where 6 were needed. Density too high; specifics lost.
**Quote-wall presentation.** Evidence dominates; analysis recedes; readers admire the quotes and miss the synthesis.
---
## Methodology-level choices that stay in the public skill
The document-not-deck principle. The patterns-lead-evidence-appendix structure. The parallel-section discipline. The opinionated voice. The decision-input section. Length calibration. The reference-after-shipping test. Visual element discipline. Common failures.
## Implementation choices that stay internal
Specific document templates and styles. Specific document hosting (Notion, Coda, Google Docs, Confluence). Specific link conventions for evidence. Specific reviewer workflows. Specific archive practices. The team's own conventions for synthesis length within the bands. These vary by team and tooling.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |