The discovery research synthesis skill.

What this skill is for

The PM suite, grouped by where work happens.

The PM catalog now spans 15 skills across four phases: upstream discovery, strategy and planning, execution, and measurement. This skill is upstream: the synthesis discipline that turns research into the decisions strategy and execution depend on.

The synthesis flow

From research artifacts to decisions, in six stages.

The sequence is non-negotiable. Teams that skip stages produce findings that look like synthesis but are not: untagged transcripts, unclustered tags, unnamed patterns, patterns without implications, implications without decisions.

Total synthesis time for a typical discovery cycle (12 interviews + 250 support tickets + 20 sales calls): 60-80 hours of focused work over 1.5-2 weeks. Teams that underestimate synthesis time at the start of discovery cycles often discover the gap after research ships and synthesis stalls.

What is in the skill

Thirteen sections covered in the body.

1. 01

   What this skill is for

   One-off research synthesis work. Distinction from user-feedback-aggregation (ongoing streams), jtbd-framing (a framing technique within synthesis), pm-spec-writing (downstream consumer).

2. 02

   Data-dump vs insight-theater vs actionable-synthesis

   The keystone framing. Litmus test: three months after synthesis ships, can team members name the decisions it informed?

3. 03

   Discovery research types

   Five types: customer interviews (depth), support tickets (volume), sales call analysis (prospect view), survey data (validation), in-app feedback (in-the-moment friction).

4. 04

   The synthesis sequence

   Six stages: transcribe, tag, cluster, name patterns, infer implications, name the so-what. Non-negotiable; skipping stages breaks the chain.

5. 05

   Tagging and clustering discipline

   Five principles: tag at artifact level, allow tag proliferation, cluster on patterns not overlap, name after clustering, tag dissent and contradictions.

6. 06

   Pattern naming

   Patterns are not category labels. Names commit specific positions. Avoid bloat (4-8 patterns typical, not 30). Avoid truisms. Name with conviction.

7. 07

   From pattern to product implication

   The bridge from observation to decision. Implications are the writer's analytical work, not user prescriptions. Each pattern can have multiple implications. Each implication acknowledges cost.

8. 08

   Writing for decisions, not deck performance

   Document over deck. Patterns lead, evidence in appendix. Opinionated voice. Decision-input section. Optimized for reference use, not for the readout meeting.

9. 09

   The synthesis review and validation loop

   Participant review where possible. Adjacent-team review. Quantitative validation. The challenge-the-synthesis session. Iteration before publishing.

10. 10

    When to halt and gather more data

    Pattern thinness, segment under-representation, contradictory clusters, decision time pressure. The 'we have enough' trap.

11. 11

    Common failure modes

    11+ patterns: data-dump, insight-theater, category bloat, truism patterns, misrecognition, vague implications, hedged synthesis, quote walls, wishlist synthesis.

12. 12

    The framework: 12 considerations

    Actionable not data-dump or theater, sequence non-negotiable, tags from data, patterns named with conviction, implications falsifiable, decision-driving so-whats, document not deck, opinionated, reviewed, gaps acknowledged.

13. 13

    Closing: synthesis is where research earns its keep

    Discovery research is expensive; synthesis is where the investment converts into decisions. Programs that take synthesis seriously earn returns; programs that data-dump or theater do not.

Pairs with these platforms

Three platforms with synthesis-relevant workflows.

The skill is platform-agnostic. These platforms ship workflows that fit synthesis programs: Notion (synthesis docs and tagging at scale), AirOps (workflow-driven synthesis pipelines with AI assistance), Mixpanel (quantitative validation of qualitative patterns).

• Notion-centric teams

  Notion

  Briefs as a queryable database

  Open the page

• Content teams that prefer managed workflow builders to build-it-yourself pipelines

  AirOps

  AirOps's official MCP and Claude Connector for AEO data and Brand Kits

  Open the page

• Product teams and analysts asking questions of product event data

  Mixpanel

  Mixpanel's official hosted MCP for product analytics

  Open the page

Direction 7 closes

The first of five PM skills closing Direction 7.

Frequently asked questions.

What does 'actionable synthesis' actually require?

The six-stage synthesis sequence: transcribe and prepare, tag at the artifact level, cluster across artifacts, name patterns, infer product implications, name the so-what tied to a specific decision. Each stage is non-negotiable. Teams that skip stages produce findings that look like synthesis but do not drive decisions: untagged transcripts, unclustered tags, unnamed patterns, patterns without implications, implications without decisions. The litmus test: three months after a synthesis ships, can team members name the decisions the synthesis informed?

How is this skill different from jtbd-framing?

JTBD-framing is one specific framing technique often applied within synthesis (situation, motivation, outcome statements; struggling moments; hire/fire criteria). This skill is broader: it covers the synthesis discipline regardless of which framing techniques are applied. JTBD is a tool used within synthesis; this skill is the synthesis itself. The two compose: discovery synthesis often uses JTBD as one of several framing approaches.

How is this skill different from user-feedback-aggregation?

Discovery research synthesis covers one-off research projects: a defined batch of artifacts (12 customer interviews + 250 support tickets + a sales call audit), a defined synthesis output, a defined timeline. User-feedback-aggregation covers always-on feedback streams: support, NPS, in-app, sales calls, social, councils, all flowing continuously. Different cadences, different tooling needs, different synthesis discipline. The two compose: discovery research often draws from feedback streams plus commissions targeted research.

What does 'data-dump vs insight-theater vs actionable-synthesis' mean?

Data-dump: research artifacts handed to product team raw. 'Here are 47 transcripts; figure it out.' No synthesis, no signal. Insight-theater: overpolished synthesis dressed as insight. Pretty quote walls, designed personas, no decision-driving conclusions. The deck that gets a 30-minute review and is never referenced again. Actionable synthesis: synthesis that drives decisions. Each pattern has a so-what attached; each finding leads to a product implication; the document gets referenced in roadmap discussions for months.

How long does substantive synthesis take?

60-80 hours of focused synthesis work for a typical discovery cycle (12 interviews + 250 support tickets + 20 sales calls). Roughly 1.5-2 weeks of dedicated PM time across the six stages: transcribe and prep (15-20 hours), tagging (25-40 hours), clustering (4-8 hours), pattern naming (2-4 hours), implications (3-6 hours), so-what (2-4 hours). Most teams underestimate synthesis time at the start of discovery cycles; budgeting it explicitly is the discipline.

What if the synthesis reveals the research did not collect enough data?

Name the gap explicitly. Do not pretend the data is sufficient when it is not. Honest options: synthesize for the represented segments and flag missing ones; pause to gather more research if a load-bearing decision warrants it; ship synthesis with named gaps and post-launch instrumentation that catches wrong decisions. The 'we have enough' trap is the failure mode: sunk-cost reasoning suppressing honest gap-naming.