--- title: 'Research Opportunity: Qual Research Process — Stephen Rust' source: notion interviewee: Stephen Rust date: '2026-01-09' type: opportunity-analysis status: active tags: [research, opportunities, qual, reporting, stephen] --- # Stephen - Desi Catch-up - Qual Research Process ## 1. INTERVIEW BRIEF (use the template from Interview Processing Guide) ## Interview Brief — *Stephen Rust / Desi catch-up on Qual research* - **Date:** 9 Jan 2026 - **Interviewee role:** Research lead (does qual + quant; moderates focus groups; designs qual studies) - **Research maturity (low/medium/high):** High - **Current tool ecosystem (systems of record vs tools of convenience):** - Systems of record: Focus groups + transcripts + debrief deck (PowerPoint) - Tools of convenience: ChatGPT for transcript querying and drafting; Movie Maker for clipping videos; manual web search for images ### A) What they are trying to do (JTBD) - **Job 1: Design qual research that avoids priming and gets to the real category context** - **Desired outcome:** respondents talk naturally about the category before being led to the client brand. - **Context / trigger:** brand discovery / relaunch briefs (example: Shaw’s premium ham). - **Job 2: Run an iterative qual process that adapts based on early learnings** - **Desired outcome:** adjust what to probe and how much time to spend where, based on early groups and client reaction. - **Context / trigger:** multi-group focus group programmes; live client observation. - **Job 3: Turn messy qual evidence into decision-grade outputs that “hit between the eyes”** - **Desired outcome:** a small number of clear attributes/territories/maps/personas with evidence that guides creatives and brand teams. - **Context / trigger:** debrief and next-step workshopping with creative teams. ### B) Where it breaks (Breakpoints) - **Breakpoint 1** - **Stage:** Before - **Failure mode:** discussion guide ordering mistakes can prime respondents and distort findings. - **Frequency:** often (design rigor described as “even more important” than quant) - **Who is impacted:** researcher + client decision makers - **Severity (1–5):** 4 (risk of invalid insight for expensive decisions) - **Workaround:** careful sequencing (category first, brands later), blind stimulus/taste tests, experienced moderation - **Cost:** risk of misleading conclusions; rework; bad downstream decisions (explicit “spend millions” risk later) - **Evidence quote:** “If I did brands first… I would completely have primed them.” - **Breakpoint 2** - **Stage:** During - **Failure mode:** discussion forums lack live cues (tone, emotion) that clients and creatives rely on to internalise insight. - **Frequency:** sometimes (depends on brief) - **Who is impacted:** client, creative agency, researcher (delivery effectiveness) - **Severity (1–5):** 3 (can weaken understanding/conviction) - **Workaround:** use focus groups for high-stakes briefs; keep forums for lighter exploration - **Cost:** weaker buy-in; less “felt” understanding (**implied**) - **Evidence quote:** “You see it… in the tonality of the voice… which you won’t get in discussion groups.” - **Breakpoint 3** - **Stage:** After - **Failure mode:** analysis-to-deck is slow because insights require clustering, naming, and compressing into a compelling slide with evidence. - **Frequency:** often - **Who is impacted:** researcher + client team - **Severity (1–5):** 4 (major time sink; delivery risk) - **Workaround:** iterative clustering; rewatch video; query transcripts with ChatGPT; manual slide building - **Cost:** significant thinking time + slide production time - **Evidence quote:** “Creating the presentation… that is the timely bit.” ### C) Switching and willingness-to-pay signals - **Switching triggers:** - Faster path from raw qual evidence to a compelling, defensible debrief (implied by “I’d love to have someone…”). - Better support for forums as “quick qual” product (explicit omnibus-style pricing analogy). - **What must be true to trust/standardize:** - Outputs must stay grounded in evidence and not “make up” claims (implied by emphasis on empirical clips across groups). - Ability to validate hypotheses across groups (“test that hypothesis down in group 3 and 4”). - **Constraints (security, procurement, bandwidth):** OPEN (not discussed) - **Where they would pay for help (self-serve vs do-it-together vs managed):** - “Quick qual” discussion group product with AI report-back at a fixed price is positioned as viable (but no explicit WTP number beyond analogy). ### D) Fast success loop (earliest win) - **Earliest “win” definition:** produce one or two “killer” slides (e.g., defining attributes, territory map, persona) with supporting evidence clips that the client can act on. - **What metric/proxy would show it:** - **Proxy metric:** *Time from transcript available → first decision-grade slide created (with evidence attached)* (Threshold **OPEN**). - Closest existing library metric: **M-AI-001 Time from data lock to deck-ready output** (adapted for qual). - **Time-to-win expectation (if stated):** OPEN (no explicit target) ### E) Architecture mapping ### Fits (map to existing nodes) - **Product:** Qual Discussion Groups *(exists as a product concept in project context; exact Notion node names OPEN)* - **Core Workflow:** Qual Design - **Job Area:** Sample Frame & Recruiting - **Feature Family:** Screener & Profile Definition - **Feature(s) implied or validated:** Define sample frame from objectives; generate screener survey - **Core Workflow:** Qual Design - **Job Area:** Discussion Guide & Stimulus Design - **Feature Family:** Order & Priming Guardrails - **Feature(s) implied or validated:** Prevent early brand priming; plan category-to-brand sequencing - **Core Workflow:** Qual Deployment - **Job Area:** Live Session Facilitation (Focus Groups) / Async Facilitation (Forums) - **Feature Family:** Client Observation & Debrief Loop - **Feature(s) implied or validated:** Observer mode; post-session “stop and think” checkpoints - **Core Workflow:** Analysis & Insight - **Job Area:** Synthesis & Evidence Packaging - **Feature Family:** Insight Clustering & Slide Formation - **Feature(s) implied or validated:** Cluster attributes; generate matrices; attach evidence clips ### Gaps / opportunities (doesn’t fit cleanly) - **Proposed new node(s) (if needed):** - **“Evidence-to-Slide Builder”** (Feature Family) — taking rough notes/buckets and producing a slide with compelling evidence. - **Why it matters:** slide creation is described as “pain” even after insights are known; evidence assembly is manual. - **Where it would ladder:** Qual → Analysis & Insight → Synthesis & Evidence Packaging - **Status:** **PROPOSED** ### F) Top evidence (quotes) 1. “If I did brands first… I would completely have primed them.” 2. “Creating the presentation… that is the timely bit.” 3. “I’d love to have someone… I’ve written out this main theme… Is there any way to get that into a PowerPoint slide?” 4. “You see it… in the tonality of the voice… which you won’t get in discussion groups.” ### G) What this changes (required) - **CONFIRMS:** qual follows a comparable lifecycle to quant (design → fieldwork → analysis), but is more iterative and client-interactive in-field. - **REFINES:** “design confidence” matters in qual too, but manifests as *anti-priming sequencing + iterative guide evolution*, not just question wording. - **ADDS:** evidence-to-slide creation support is a real gap; current process is manual and time-heavy. --- ## 2. CODED INSIGHTS (5–10 max) ### Insight 1 - **Insight statement:** Qual design requires rigorous sequencing to avoid priming, often more than quant. - **Stage:** Before - **Severity (1–5) + why:** 4 — wrong order can invalidate findings for high-stakes decisions. - **Workaround summary:** Start with category context; delay brand mention; blind exercises. - **Cost signal summary:** Risk of misleading conclusions; downstream wasted investment. - **Evidence quote:** “If I did brands first… I would completely have primed them.” ### Insight 2 - **Insight statement:** Qual is inherently iterative: the guide and emphasis shift after early groups and client “stop and think” sessions. - **Stage:** During - **Severity (1–5) + why:** 3 — not failure, but core workflow need; unsupported iteration wastes time. - **Workaround summary:** Adjust guide after 1–2 groups; reallocate time across topics. - **Cost signal summary:** Time lost if iteration is clumsy; missed value if not adjusted (**implied**). - **Evidence quote:** “Absolutely iterative… we even have a stop and think session with the client.” ### Insight 3 - **Insight statement:** The biggest time sink is compressing qual evidence into a small set of compelling, decision-grade slides. - **Stage:** After - **Severity (1–5) + why:** 4 — major drag; affects delivery quality and speed. - **Workaround summary:** Manual clustering; repeated rewatching; trial-and-error naming. - **Cost signal summary:** Significant “thinking time” plus slide production time. - **Evidence quote:** “Creating the presentation… that is the timely bit.” ### Insight 4 - **Insight statement:** Evidence packaging is manual: finding and clipping video moments across groups is laborious but critical for trust. - **Stage:** After - **Severity (1–5) + why:** 4 — required for credibility; takes time. - **Workaround summary:** Ask for timestamps via ChatGPT; manually clip/edit in Movie Maker. - **Cost signal summary:** “Huge amount of work” to find and assemble proof. - **Evidence quote:** “Finding that video, huge amount of work… then we clip it and edit it.” ### Insight 5 - **Insight statement:** Discussion forums work best as “quick qual” for lighter decisions, but focus groups win when clients need live emotional conviction. - **Stage:** Cross - **Severity (1–5) + why:** 3 — modality choice affects insight conviction and stakeholder alignment. - **Workaround summary:** Use forums for lightweight exploration; focus groups for high-stakes brand territory work. - **Cost signal summary:** Wrong modality risks weak buy-in or missed nuance (**implied**). - **Evidence quote:** “You see it… in the tonality… which you won’t get in discussion groups.” ### Insight 6 - **Insight statement:** AI is already used as a query and drafting partner, not just summarisation, to test hypotheses and accelerate output writing. - **Stage:** After - **Severity (1–5) + why:** 3 — meaningful speed gain; adoption already happening. - **Workaround summary:** Use ChatGPT to validate themes, extract moments, draft positioning territory language. - **Cost signal summary:** Reduces time vs “long time to write that insight” (still manual finishing). - **Evidence quote:** “I frame that then ask AI to fill it in… that used to take… a long time.” --- ## 3. ARCHITECTURE MAPPING > Note: Exact existing node names in Notion are OPEN; mapping uses our canonical hierarchy and aligns to the implied Qual product structure. > ### Insight 1 (REFINES) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Qual Design - **Job Area:** Discussion Guide & Stimulus Design - **Feature Family:** Order & Priming Guardrails - **Feature:** Flag priming risk in guide order (e.g., brand too early) ### Insight 2 (CONFIRMS + REFINES) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Qual Deployment - **Job Area:** Iterative Moderation & Guide Evolution - **Feature Family:** Stop-and-Think Checkpoints - **Feature:** Capture and apply learnings after early sessions - **Status:** **OPEN** whether Job Area exists today; may need **ADD** if absent. ### Insight 3 (REFINES) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Analysis & Insight - **Job Area:** Synthesis & Story - **Feature Family:** Insight Clustering & Narrative Compression - **Feature:** Collapse themes into a small set of decision-grade takeaways ### Insight 4 (ADD — PROPOSED if not present) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Analysis & Insight - **Job Area:** Evidence Packaging - **Feature Family (NEW):** Evidence-to-Clip Workflow (**PROPOSED**) - **Feature:** Find, clip, and attach supporting moments across sessions ### Insight 5 (CONFIRMS) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Qual Deployment - **Job Area:** Client Observation & Participation - **Feature Family:** Observer Experience - **Feature:** Observer mode for clients/creatives (watch, take notes) - **Status:** Node existence **OPEN**; if absent, treat as **ADD** candidate. ### Insight 6 (REFINES) - **Portfolio:** EmpathyIQ - **Product:** Qual Discussion Groups - **Core Workflow:** Analysis & Insight - **Job Area:** Synthesis & Story - **Feature Family:** Hypothesis Testing & Drafting Support - **Feature:** Query transcripts to validate themes and draft outputs - **Note:** Must be phrased as outcome, not “AI”. --- ## 4. PROPOSED ARCHITECTURE ACTIONS (required) - **CONFIRM:** Qual work can be structured into Design → Deployment → Analysis, similar to quant, but with heavier iteration and stakeholder involvement. - **REFINE:** *Qual Design* should explicitly include priming/ordering rigor as a first-class outcome (not an implicit craft skill). - Suggested revised Outcome/Promise (if relevant to a node): **OPEN** (we don’t have the current text, so can’t safely propose a rewrite beyond this directional refinement). - **SPLIT:** None (no evidence of an over-broad node from this single interview). - **ADD:** 1. **Name:** Evidence-to-Slide Builder - **Type:** Feature Family - **Outcome/Promise:** Help researchers turn rough themes and supporting evidence into a slide that is clear enough to act on. - **Ladders to:** Qual → Analysis & Insight → Synthesis & Evidence Packaging - **Status:** PROPOSED 2. **Name:** Evidence Clip Assembly - **Type:** Feature - **Outcome/Promise:** Let researchers find and assemble short proof moments from recordings to support key claims. - **Ladders to:** Qual → Analysis & Insight → Evidence Packaging → Evidence-to-Clip Workflow - **Status:** PROPOSED 3. **Name:** Stop-and-Think Checkpoints - **Type:** Feature Family (or Feature, depending on existing structure) - **Outcome/Promise:** Enable teams to adjust what to probe based on early sessions without losing momentum. - **Ladders to:** Qual → Deployment (and Design) - **Status:** OPEN (needs broader validation across more qual practitioners) - **REJECT / Not now:** “Discussion forum as mini focus group replacement” for high-stakes briefs where live emotion and client conviction matter. It’s not positioned as a substitute here. --- ## 5. PATTERN UPDATES (for Insights Registry) ### Which existing patterns this strengthens (if any) - **P-103 — Reporting is a time sink; reuse and “deck continuity” is a major opportunity** Strengthened for **qual** specifically: the bottleneck is “compelling slide + compelling evidence,” not just summarising transcripts. - **P-003 — “Confidence before/during/after” is strong framing, but could be hygiene** Adds nuance: “confidence” in qual design shows up as avoiding priming + iteration discipline; WTP signal still **OPEN**. ### Any new pattern candidates 1. **Pattern name:** Evidence packaging is the real reporting bottleneck - **Category:** Pain / Workaround - **Stage:** After - **Severity:** 4 - **Evidence quotes:** - “Creating the presentation… that is the timely bit.” - “Finding that video, huge amount of work… then we clip it and edit it.” 2. **Pattern name:** Qual design confidence is about priming control and sequencing - **Category:** Trust / Pain - **Stage:** Before - **Severity:** 4 - **Evidence quote:** - “If I did brands first… I would completely have primed them.” --- ## 6. OPEN ITEMS / ASSUMPTIONS UPDATED - **OPEN:** Is “quick qual discussion forum + AI report back at fixed price” a real commercial SKU, or just a thought experiment? (He analogised to omnibus pricing but did not commit.) - **OPEN:** How broadly “stop-and-think checkpoints” generalise across qual teams (could be Rust-specific practice vs standard). - **UPDATED ASSUMPTION (supports A-002):** “Confidence” applies to qual, but is less about correctness of a static guide and more about controlling priming + adapting through fieldwork.