# Research Synthesis: [Study title]

> A record of how raw research data became insights: the sources analysed, atomic observations,
> the affinity clusters they formed, insights with links to their evidence and a stated confidence,
> and the opportunities and open questions that follow. It follows common qualitative analysis
> practice (affinity diagramming and thematic analysis). Replace every `[bracketed]` value and
> every row marked *Example*. Worked examples continue the Quillmere Libraries interviews from
> design/research-interview-guide (Quillmere Libraries is a fictional city library service).

<!--
Guidance: How to use this template
- Work bottom up: observations first (section 2), clusters second (section 3), insights last
  (section 4). Writing insights first and then looking for quotes is confirmation bias.
- Every insight links to observation IDs, and every observation links to a participant and a
  source. A reader must be able to walk from an insight back to what someone actually said or did.
- State confidence honestly with the rubric in section 5. "Low" is a useful result: it tells the
  team what to research next.
- Common pitfall: counting mentions as if a qualitative sample were a survey. "7 of 10" describes
  this sample; it does not estimate a population share.
-->

## Document control

<!-- Guidance: Name who analysed the data. Synthesis done by one person alone is weaker; record the second analyst. -->

| Field | Value |
|---|---|
| Document ID | [RSY-001] |
| Version | [1.0] |
| Status | [In analysis / In review / Final] |
| Owner (lead analyst) | [@researcher] |
| Second analyst | [@name] |
| Reviewers | [@product-owner], [@designer] |
| Last updated | YYYY-MM-DD |
| Classification | [Internal; pseudonymised] |
| Studies synthesised | [*Example:* RIG-001 interviews, 10 participants, YYYY-MM-DD to YYYY-MM-DD] |

### Revision history

| Version | Date | Author | Change |
|---|---|---|---|
| 0.1 | YYYY-MM-DD | [@name] | Observations coded |
| 0.2 | YYYY-MM-DD | [@name] | Clusters and draft insights after team workshop |
| 1.0 | YYYY-MM-DD | [@name] | Reviewed by second analyst |

## 1. Inputs

<!-- Guidance: List every source analysed, including what was excluded and why. A synthesis that silently drops sessions is not trustworthy. -->

| Source ID | Type | Participants | Date | Used | Notes |
|---|---|---|---|---|---|
| SRC-01 *(Example)* | Interviews (RIG-001) | P01 to P10 | YYYY-MM-DD | Yes | Transcripts and notes |
| SRC-02 *(Example)* | Support tickets, last 90 days | n/a | YYYY-MM-DD | Yes | 214 tickets, filtered to reservations |
| SRC-03 *(Example)* | Usability test UTR-001 debrief | U01 to U08 (different people, own codes) | YYYY-MM-DD | Yes | Debrief answers only |
| SRC-04 *(Example)* | Pilot interview | Colleague | YYYY-MM-DD | No | Pilot only, not a participant |

## 2. Observations

<!-- Guidance: One observation per row: a single fact about what a participant said or did, in neutral words, with a source. Keep interpretation out of this table. -->

| ID | Participant | Source and time | Observation |
|---|---|---|---|
| OBS-001 *(Example)* | P03 | SRC-01, 12:30 | Saves screenshots of book posts and rarely returns to them |
| OBS-002 *(Example)* | P07 | SRC-01, 09:15 | Keeps a paper list on the fridge of books to look for |
| OBS-003 *(Example)* | P01 | SRC-01, 18:40 | Gave up on a reservation when the wait was 9 weeks and bought the e-book |
| OBS-004 *(Example)* | P05 | SRC-01, 21:05 | Did not know reservations could be collected at any branch |
| OBS-005 *(Example)* | P09 | SRC-01, 15:20 | Trusts a friend's recommendation more than any list |
| OBS-006 *(Example)* | n/a | SRC-02 | 38 of 214 tickets ask where a reserved book will arrive |

The example shows one or two observations per cluster for brevity; in a real synthesis this table
holds every coded observation, and the participant counts in section 3 come from it.

Coding: [*Example:* two analysts coded transcripts P01 to P04 independently, compared codes,
agreed a codebook, then split the remaining six].

## 3. Affinity clusters

<!-- Guidance: Group observations by similarity without a predefined frame, then name each cluster with a sentence that says what it means, not a one-word label. -->

| Cluster ID | Cluster statement | Observations | Participants |
|---|---|---|---|
| CL-1 *(Example)* | Book ideas are captured in many places and then lost | OBS-001, OBS-002 | 7 of 10 |
| CL-2 *(Example)* | Long or unknown waits push people to buy instead | OBS-003 | 5 of 10 |
| CL-3 *(Example)* | Collection options are poorly understood | OBS-004, OBS-006 | 4 of 10, plus tickets |
| CL-4 *(Example)* | Personal recommendations outweigh lists and algorithms | OBS-005 | 6 of 10 |

```mermaid
flowchart TD
  O1["OBS-001"] --> C1["CL-1 Ideas captured then lost"]
  O2["OBS-002"] --> C1
  O3["OBS-003"] --> C2["CL-2 Waits push people to buy"]
  O4["OBS-004"] --> C3["CL-3 Collection options unclear"]
  O6["OBS-006"] --> C3
  O5["OBS-005"] --> C4["CL-4 Personal recommendations win"]
  C1 --> I1["INS-01"]
  C2 --> I2["INS-02"]
  C3 --> I2
  C4 --> I3["INS-03"]
```

## 4. Insights

<!-- Guidance: An insight explains why something happens and matters for the product. Link evidence, state confidence, and say what would change your mind. -->

| ID | Insight | Evidence | Confidence | What would change it |
|---|---|---|---|---|
| INS-01 *(Example)* | People do not lack ideas for what to read; they lack one place that holds them until they are ready to borrow | CL-1; OBS-001, OBS-002 | Medium | A survey showing most borrowers already use a wishlist |
| INS-02 *(Example)* | Uncertainty about when and where a book will arrive drives abandonment more than the wait itself | CL-2, CL-3; OBS-003, OBS-004, OBS-006 | Medium | Abandonment rates that do not change when wait and branch are shown |
| INS-03 *(Example)* | Recommendations from people the borrower knows carry more weight than library lists | CL-4; OBS-005 | Low | More interviews across age groups; only urban participants so far |

## 5. Confidence rubric

<!-- Guidance: Use one rubric for every insight so that "High" means the same thing across studies. -->

| Level | Criteria |
|---|---|
| High | Seen across most participants and at least two independent sources or methods; no strong contrary evidence |
| Medium | Seen across several participants in one method, or supported by a second source with gaps |
| Low | Seen in a few participants, or in one segment only, or contradicted by some evidence |

## 6. Opportunities

<!-- Guidance: Turn insights into "How might we" questions or opportunity statements. Do not jump to one solution; that is the next design step. -->

| ID | Opportunity | From insight | Next step |
|---|---|---|---|
| OPP-01 *(Example)* | How might we help borrowers keep book ideas in one place until they are ready? | INS-01 | Concept test of a "Want to read" list |
| OPP-02 *(Example)* | How might we show when and where a reserved book will arrive before people commit? | INS-02 | Add to the reservation redesign; check in a usability test |
| OPP-03 *(Example)* | How might we bring personal recommendations into the app? | INS-03 | Further interviews first, confidence is low |

## 7. Gaps and next questions

<!-- Guidance: Say who was not heard and what remains unknown. This guides the next study. -->

- *Example:* no participants under 25 or over 70; no screen reader users in this round.
- *Example:* how many borrowers already keep wishlists elsewhere (needs a survey).
- *Example:* whether showing estimated wait reduces abandonment (needs an experiment, see product/ab-test-plan).

## Standard and licence note

Based on common qualitative analysis practice: affinity diagramming (the KJ method attributed to
Kawakita Jiro) and thematic analysis as described by Braun and Clarke (2006, "Using
thematic analysis in psychology", *Qualitative Research in Psychology*, 3(2), 77 to 101). The
text is original. Related: design/research-interview-guide, design/usability-test-report,
product/persona, product/journey-map. Template licence: MIT.