Marketer's Cowork discovery research · Zeotap CDP · Garvita Gupta
Start a session
Step 1 — what are you running today?
Step 2 — which persona are you interviewing?
Step 3 — participant info
0 / 0 answered
Reference only — for your own understanding of what each section is testing. Don't read these aloud or paraphrase them to the participant; they exist so you can ask sharper follow-ups, not so you can steer toward a "right" answer.
Ground rules for the interviewerDo
Ask about the last specific instance, not a general pattern or hypothetical.
Ask what they've already tried, built, or paid for — not just what they'd want in theory.
Dig past the first answer to the underlying job — a feature request is a clue, not the finding.
Write down verbatim phrases, not paraphrases.
Note the participant's industry via the Industry field above, and probe for vertical-specific color inside JTBD, planning, and channel questions as it comes up — retail tends to be catalogue/inventory-driven, telecom tends to be churn/contract-driven — rather than relying on separate vertical-only questions.
Tell participants upfront that any quotes used in the executive readout will be anonymized — this earns more candor, not less.
Where possible, run interviews with two people — one interviewing, one taking notes — especially for the ~65–80 min Track A session. Freeing the interviewer from note-taking keeps them focused on listening.
Don't
Ask leading questions — offer open framing rather than a menu of pre-defined answers, and let the participant's own words reveal the pattern.
Mention the product concept by name, "agentic AI," or "autonomous decisioning" before the prior-tool-experience section of that track's interview. Naming the concept early anchors every answer that follows.
Ask "would you use a tool that does X" — ask how they handle X today and what breaks when they do.
Let one strong voice in the focus group set the tone — use silent-write-first and round-robin.
If interviewing more than one person at the same company, don't reference what an earlier participant there told you — that's feeding one person's answer into another's interview.
Recruiting criteria (reference only, not asked live)Track A — Strategic Lifecycle Lead (primary recruit, 8–12 total)
Include: senior lifecycle marketer with real accountability for a lifecycle stage (retention, reactivation, win-back, upsell); currently dependent on analytics, marketing ops, or content teams to act on an idea.
Deliberately mix in: 2–3 on a largely manual stack, 2–3 on an advanced CDP/ESP (Braze, Iterable, Bloomreach, Insider, CleverTap, MoEngage), 2–3 who've piloted, evaluated, or churned off an AI/agentic marketing tool.
Guarantee industry representation: at least 4–5 retail/e-commerce and 4–5 telecom out of the 8–12 — Zeotap's majority client base. Optionally 1–2 from an adjacent vertical (BFSI, travel) as an outlier check.
Exclude from Track A specifically: marketers without decision authority over targeting, timing, channel, or offer (that's the Track B profile, not Track A); pure analysts with no campaign ownership.
Track B — Operational/Granular Executor (a handful, ideally same companies as Track A recruits)
Runs journeys, builds segments, or executes tests day to day — hands-on with the tooling, even without final decision authority. Skip if the Track A recruit does this work themselves.
Track C — Content/Creative Lead (a handful, separate recruit if content isn't owned by the lifecycle marketer)
Owns content/brand/copy decisions for lifecycle campaigns. This can be a dedicated content or brand team member with no lifecycle-stage ownership at all — that's expected for this track, not a disqualifier.
Track D — Marketing Ops/Governance Lead (a couple, short targeted conversations)
Owns audit trail, compliance review, or override capability for campaigns.
Recruiting source (applies across all tracks)
Mix existing Zeotap customers, active prospects, and — if possible — a few marketers with no Zeotap relationship at all. Recruiting only from Zeotap's own install base risks under-discovering pain points the current product doesn't address, and introduces social-desirability pressure to be diplomatic with your own vendor even in a 1:1.
Offer an incentive (gift card, early access, or similar) — an unpaid ask for 65–80 minutes from senior practitioners carries real no-show and decline risk.
Hypothesis map — what this research must decide
Discovery that can't lose an argument isn't discovery — it's a listening tour. These are the six falsifiable bets behind the concept; every question in the flow exists to move one of them. At readout, each bet gets a verdict — Confirmed / Killed / Unclear — backed by verbatim quotes.
H1 — The bottleneck is production, not decisioning
Marketers broadly know who to target and what to say — what's slow is producing the creative variants and getting them through review.
Confirmed if: "last time" stories stall at creative/review stages, and variant counts are low because production is expensive. Killed if: content ships same-day and the stall is analysis or targeting instead.
H2 — The data ↔ execution seam is a felt pain
The boundary between the data platform (CDP/warehouse) and the sending tool causes lived pain: mismatched audiences, sync delays, the same segment defined in four places.
Confirmed if: unprompted seam incidents with real consequences. Killed if: nobody raises it and the probes reveal a non-event.
H3 — Insight-to-live takes weeks, and it hurts
From "I noticed something off in the base" to "a campaign is live against it" is measured in weeks — and marketers feel that lag as a top-3 problem, not an accepted fact of life.
Confirmed if: reported timelines run 2+ weeks and the forced-ranking pain relates to speed. Killed if: it's days, and they're content with it.
H4 — Autonomy is acceptable for creative & timing, not audience & offer
Marketers will let an agent draft, test, and time messages — but who gets targeted, what offer goes out, and what money is spent stay human decisions until proven otherwise.
Confirmed if: trust boundaries consistently land there across tracks. Killed if: the boundary lands elsewhere.
H5 — Trust is earned through reversible wins and visibility, not claims
Where marketers already delegate today, trust grew through small, checkable, reversible wins.
Confirmed if: "stopped checking" stories cite accumulated evidence, visibility, and an easy undo. Killed if: delegation happened by mandate, or genuinely never happens.
H6 — The lifecycle lead champions; ops/governance and a budget owner decide
The person we're building for can start the purchase but rarely finish it — adoption dies in governance, IT, or budget unless audit and override are first-class.
Confirmed if: buying stories show multi-party sign-off, and governance can point to real veto incidents. Killed if: the lead buys autonomously.
Decision gate (leadership-agreed, not asked live)
Track A — Strategic Lifecycle Lead (primary persona · core ≈ 45–50 min · full ≈ 65–75 min)
Senior lifecycle leads realistically give you 45–60 minutes, not 80 — run the core path (Warm-up → JTBD → Prior AI tools → Buying reality → Accountability → Close) and add "if time" sections only in a genuinely full slot.
Warm-up (5–6 min)core
Role, team size, which lifecycle stage(s) they own, how their own performance is measured.
why we askBaseline classification — used to segment findings by seniority and KPI ownership across interviews.
"What does a typical day in your role look like?"
why we askLow-stakes opener that gets them talking before the harder questions.
"How many campaigns or journeys are you actively running right now? How do you decide which one gets your attention today?"
why we askChecks whether a single-goal tool addresses most of their real day, or a narrower slice of it.
Team structure & ownership (5–6 min)if time
If squeezed, still ask the "who's involved" question below — it feeds H6 (buying reality) together with the Buying reality section.
"Walk me through your team's structure — who reports to whom, and where do you sit in it?"
why we askMaps the org chart around this decision — feeds H6 (who champions vs. who actually decides), alongside the Buying reality section.
"Who's involved in a typical campaign, and what does each person actually do? What happens when one of them is slow or unavailable?"
why we askTests the coordination-dependency bottleneck directly — this is the pain the product's "self-sufficiency" pitch is built on.
"Is there a separate content or brand team you depend on, or do you write/select copy yourself?"
why we askDetermines whether Track C needs a genuinely different recruit at this company.
"How many people, across all teams, would need to touch a single campaign from idea to live?"
why we askQuantifies coordination overhead — a high number is direct evidence for the product's core promise of reducing cross-team dependency.
Plan — decision-making process (2–3 min)if time
"Walk me through how a campaign plan actually gets built today — a doc someone drafts alone, a live meeting, something else?"
why we askEstablishes what the Plan stage of the workflow looks like today, to compare against the product's human-in-the-loop Plan step.
JTBD — the last time something needed fixing (12–15 min)core
"Tell me about the last time you noticed something was off in your customer base — a segment underperforming, a cohort shrinking, a drop-off. Walk me through it start to finish."
why we askThe core JTBD anchor question — everything else in this section unpacks this one story in more analytical detail (H1, H2, H3).
Probe: How did you first notice it? What told you — a dashboard, a colleague, a gut feeling?
why we askDistinguishes proactive dashboard-driven detection from a reactive tip-off — informs what "Explore" should surface unprompted.
Probe: What did you do in the first hour? The first day? The first week?
why we askMaps today's actual response timeline — the baseline H3 measures against.
Probe: How long from "I noticed it" to "something was live"?
why we askThe literal H3 number, in their own words, before the synthesis step asks for it again.
Probe: Where in that process did you get stuck, or annoyed, or give up on doing it the "right" way?
why we askPinpoints exactly where in the journey the bottleneck sits (H1).
"What does a good outcome look like for a situation like that? How do you know it worked?"
why we askDefines their own success criteria — informs what the product's Measure stage should report back.
Channel & targeting strategy (5–6 min)if time
Policy-level channel questions — the hands-on percentage breakdown lives in Track B.
"Walk me through the last multi-channel campaign you planned — how did the channel mix come together?" channel
why we askTests channel-decision logic today — informs whether the product's channel orchestration should be human-authored policy or agent-suggested.
"Tell me about the last time you added or dropped a channel from the mix. What triggered it?" channel
why we askSurfaces what actually triggers a channel-mix change — informs how reactive the product's channel logic needs to be.
"What metrics do you personally track daily, weekly, and monthly to know how your campaigns are performing?"
why we askDirectly informs which metrics the product's performance dashboard should surface by default.
Prior experience with AI / automated decisioning tools (10–12 min)core
Answers "have they used any other such tool, what did they think, what gaps did they see." Concept naming is safe from here on.
"Have you used any tool that automatically decided who to target, when, or through which channel — rather than you configuring the rules yourself?"
why we askOnly name examples if they draw a blank: Braze Fusion/Canvas AI, Iterable, Bloomreach, Insider, CleverTap, MoEngage AI, Movable Ink. Establishes the competitive/prior-tool landscape (H4, H5).
If yes — "Tell me about the last time it made a decision you disagreed with. What did you do?"
why we askTests the trust boundary in a live, negative-outcome scenario (H4).
If yes — "What did you have to give up or trust in order to use it?"
why we askDirect language for H5 — the trust cost of delegating today.
If yes — "What made you stop using it — or what would make you stop?"
why we askChurn/trust-collapse triggers — informs what would make a marketer abandon this product too.
If yes — "What's the one thing it got wrong that you still remember?"
why we askA memorable-failure anchor — the kind of error the product's guardrails must specifically prevent.
If no — "What's stopped you from trying one?"
why we askSurfaces adoption barriers before the product is even evaluated — procurement, trust, budget, etc.
Buying reality (4–5 min)core
Pain that can't travel a purchase path never becomes revenue. Past purchases are facts; "I'd definitely buy this" is theatre — so we ask about the last purchase, not the next one.
"Walk me through the last marketing tool your team actually bought — who first pushed for it, who evaluated it, who signed, and how long from first conversation to it being live?"
why we askThe real buying committee and cycle length (H6) — the difference between a 6-week champion-led sale and a 9-month procurement grind changes how the concept must be packaged, priced, and piloted.
"And the last tool your team abandoned or churned off — what killed it?"
why we askChurn stories are the adoption failure modes a new tool must design around — integration debt, low usage, the champion leaving.
"If you had to cut one tool from your stack tomorrow, which goes first — and why?"
why we askA value ranking of the incumbent stack — whether a new entrant would replace a resented line item (easier sale) or has to win net-new budget (harder).
Accountability & vision (6–7 min)core
"Tell me about something in your workflow you used to review manually every single time — and now you don't. What made you stop checking?"
why we askTests how trust is actually earned stepwise (H5) — the real mechanics: how many wins it took, what visibility they needed, whether an easy undo mattered.
"Tell me about the last campaign that went wrong — walk me through what happened, who ended up accountable, and what changed afterward."
why we askTests where accountability sits today, ahead of the autonomous/approval/org-enforced governance tiers.
Probe: What did you have to pull together to explain it upward — to your boss or leadership — and where did it come from?
why we askUpward-reporting angle on Measure — pairs with Track B's hands-on diagnosis version of the same question.
"Tell me about the worst time a message went out to the wrong audience — what happened, and what was the fallout: a complaint, a regulator, an unsubscribe spike, something else?"
why we askGrounds the cost-of-error assumption behind the concept's complexity-based tiering logic.
The next question is opinion about a hypothetical future, which carries less weight than the reported incidents above (Mom Test) — treat as a hypothesis to pressure-test in Phase 2, not a validated requirement.
"Which part of your job would you least want an agent involved in, even if it did it well?"
why we askSeparates identity/control concerns from pure accuracy trust (H4). This same question is asked again after the prototype demo below — deliberately, to see if seeing guardrails moves the boundary.
Close — forced ranking & next steps (3–4 min)core
Without a forced trade-off, every pain in the interview sounds important and none of them is rankable. Never skip this.
"Of everything we've discussed today, which one problem would you pay to make disappear tomorrow — and what is it costing you right now, in hours, headcount, or money?"
why we askOne forced choice plus one cost number per interview — across 8–12 interviews this becomes a pain league table with an attached economic size.
"Who else — inside or outside your company — should we be talking to about this?"
why we askRecruiting multiplier — warm intros to exactly-profiled peers at zero recruiting cost.
"We'll be building against what we hear in these conversations. Would you be open to a 30-minute look at an early prototype in a few weeks?"
why we askBuilds the Phase 2 recruiting pipeline inside Phase 1 — an eager yes versus a polite dodge is itself data on how much the pain resonated.
Phase 1.5 — Concept reaction: show the prototype (last 10–12 min, 1:1)
Sequencing rule: run this ONLY after the discovery sections and the forced-ranking close are done. Script the handover: "Before we wrap — can I show you something we're exploring? It's rough, it's a concept, and a negative reaction is genuinely useful." Say "concept", never "product". Record their first ten seconds verbatim.
(Stay silent after the demo. Then:) "What did you just see? Walk me back through what you think it does."
why we askComprehension before opinion — if a senior marketer can't retell it, the narrative is broken no matter how much they smiled.
"Where in your current week would this actually slot in — and what, if anything, would it replace?"
why we askWorkflow fit and displacement reality — does it substitute something they already pay for, or add a new thing to babysit.
"What's the first thing you'd never let it do?"
why we askThe post-stimulus trust boundary — compare against their pre-demo answer in Accountability & vision. H4 tested twice per participant, deliberately.
"What's missing that would make this unusable for your team tomorrow?"
why we askAsks for objections, not compliments — the real v1 backlog in priority order.
"If we ran a six-week pilot with your team — one journey, one channel, a proper control group — would you take it? What would you need to say yes?"
why we askCommitment beats compliments — a countable pilot-intent signal plus their explicit preconditions.
Track B — Operational/Granular Executor (~35–42 min)
Runs the journeys, builds segments, executes tests day to day.
Warm-up (2–3 min)
"Walk me through a typical day in your role — what do you personally operate day to day?"
why we askBaseline of what this persona actually touches hands-on day to day — sets context for every mechanic question that follows.
Tools & channel mechanics (9–11 min)
"Walk me through every tool you touch to go from noticing a problem to a live campaign." (list them out, in order)
why we askMaps the actual tool chain — informs integration requirements and where the data↔execution seam (H2) might live.
"Which of those steps are manual that you wish weren't?"
why we askDirect automation-opportunity signal — the most literal "what should this product automate" question in the guide.
"What percentage of your lifecycle sends go through each channel today — email, SMS, push, WhatsApp, in-app? How was that split arrived at?" channel
why we askQuantifies real channel usage and how the split is decided — answers whether email is still the dominant channel.
"How do you actually apply a suppression or exclusion rule day to day — is it built into the segment logic, a separate list you maintain, or something you check manually before a send?"
why we askHands-on mechanics version of the SUPPRESS decisioning primitive — informs whether suppression logic should live inside segment definitions or as a separate layer in the product.
"Walk me through how send timing or cadence across campaigns actually gets set today — a fixed rule, chosen manually each time, or something else?"
why we askMechanics-level version of the WHEN decisioning primitive — parallels the channel question above, and grounds the "send-time optimization" example raised in the AI-tools section below.
Prior experience with AI/automated tools (6–7 min)
Establishes hands-on exposure to existing automation — this persona is the one most likely to have actually touched these features day to day. Concept naming is safe from here on for this track.
"Do any of the tools you use day to day already do something automatically for you — send-time optimization, auto-segmentation, subject-line testing, AI-picked channel per user, AI-suggested audiences — rather than you configuring it by hand?"
why we askEstablishes hands-on exposure to existing automation, parallel to Track A's and C's prior-tool-experience sections — this persona is the one most likely to have actually touched these features day to day.
If yes — "Tell me about the last time it did something you had to go back and override or fix. What happened?"
why we askTests trust boundary in a live, negative-outcome scenario (H4), from the hands-on operator's view.
If yes — "How much do you trust its output today — walk me through what you always double-check before it goes out, versus what you're comfortable letting through without review?"
why we askMaps the granular trust boundary by category (H5) — informs exactly which parts of a tool's output need human-in-the-loop review versus full autonomy, not just whether trust exists in general.
If yes — "What's the one thing it got wrong that you still remember?"
why we askMemorable-failure anchor — the kind of error the product's guardrails must specifically prevent.
If no — "What's stopped you from turning one of these features on, even though it exists in your tools?"
why we askSurfaces adoption barriers at the point closest to the actual toggle — often different from Track A's strategic-level barriers (e.g. "I don't trust it" vs. "we were never told about it" vs. "IT locked it down").
"Tell me about the last time you did something manually that felt like it shouldn't have needed a human — walk me through it."
why we askGrounded, specific-instance version of "what should be automated" — pairs with the earlier "which steps are manual that you wish weren't" question but forces a concrete story instead of a general list.
Testing & experimentation mechanics (6–7 min)
Written open-ended on purpose — don't ask "do you run A/B tests," ask how the process works and let a "we don't really" answer emerge naturally if that's the truth.
"Walk me through how a test or experiment typically gets set up in your workflow — from idea to running it."
why we askMaps today's test/iterate loop mechanics — informs how a testing/optimization feature should fit into existing habits.
"How do you decide which segments or audiences go into a test?"
why we askTies test design to the WHO decisioning primitive — informs whether audience selection for tests should be agent-suggested.
"Tell me about the last test you ran — how long did it take to get a readable result, and who decided to ship the winner?"
why we askSpecific-instance version covering test duration and decision authority together.
Explore — attention & signal tracking (5–6 min)
"Where do you get your analytics from today — a dashboard you check yourself, someone you ask, something else? What metrics do you track daily, weekly, and monthly?"
why we askMaps today's Explore-stage habits and which metrics matter most.
"How many dashboards, alerts, or automated reports currently compete for your attention in a given week?"
why we askQuantifies signal overload — supports the thesis that a single agent view reduces cognitive load versus juggling many dashboards.
"Tell me about the last time you went looking for an answer in your data without already knowing what you were looking for. How did you do it?"
why we askTests the undirected "Explore" stage specifically — whether discovery happens proactively or only when a dashboard flags something.
Measure & Iterate — diagnosis and what changes next (6–7 min)
"When a campaign misses its number, what's your actual hands-on process for figuring out why?"
why we askPairs with Track A's "explaining to your boss" probe — this is the working-level diagnosis, that's the upward-reporting version.
"Tell me about the last time something you learned from one campaign actually changed how you ran the next one. What changed, and how did that change happen?"
why we askTests the Iterate stage directly, anchored on a real specific instance rather than a general habit. "How did that change happen" also surfaces whether iteration today means rebuilding from scratch or reusing/cloning a prior campaign's setup — relevant to whether the product needs a clone-and-modify capability.
Track C — Content/Creative Lead (~34–40 min)
Content generation, selection, performance analysis, and creative/testing optimization are confirmed in scope.
Warm-up (3–4 min)
Role, team, which content/brand areas they own.
why we askBaseline classification for this persona.
"What does a typical day in your role look like?"
why we askLow-stakes opener — same purpose as Track A's version.
Content creation & workflow (7–9 min)
"Walk me through how content/copy gets created for a lifecycle send today — in-house, agency, templates, an AI writing tool?"
why we askTests H1 (is the bottleneck production, not decisioning) directly.
"Which AI writing or creative tools has your team actually adopted in the last 12 months — and what happened after?"
why we askAdoption facts — what was tried, what stuck, what was quietly dropped and why (H1, H5).
"How many content variants do you typically need per campaign or segment, and how long does producing each one take?"
why we askQuantifies production cost — core evidence for H1.
"Tell me about the last time this workflow felt like the biggest drag on your week — what part was it, and why?"
why we askSpecific-instance pain-point question — forces a concrete, comparable answer instead of a general complaint.
Personalization, metrics & governance (5–6 min)
"Walk me through how content personalization actually works today — what tools are involved, from deciding a variant to it going live?"
why we askMaps today's personalization mechanics — informs how much of that loop the product should own end to end.
"What governance or sign-off is required before a personalized variant, or new content, goes live — brand, legal, compliance?" cross-check: also Track D
why we askGovernance model design input — compared against Track D's Content governance sign-off answer.
Prior experience with content/creative optimization tools (8–10 min)
"Have you used a tool that automatically generated, selected, or optimized content/creative — rather than your team producing it?"
why we askOnly name examples if blank: Persado, Jacquard/Phrasee, Movable Ink Da Vinci, Braze/Iterable's built-in AI content features.
If yes — "Tell me about the last time it produced or picked something you wouldn't have chosen yourself. What did you do?"
why we askTests creative trust boundary in a live scenario (H4, content-specific).
If yes — "What did you have to give up or trust to use it — brand voice control, review time, something else?"
why we askH5 — trust cost of delegating creative decisions specifically.
If yes — "What made you stop using it, or what would make you stop?"
why we askChurn trigger for content-specific tools.
If no — "What's stopped you from trying one?"
why we askAdoption barrier for content-tools specifically.
Content performance & attribution (4–5 min)
"How do you measure whether a piece of content — not just a campaign — actually worked? What metrics tell you a variant is winning?"
why we askEstablishes what "working" means for a piece of content, distinct from campaign-level performance.
"Can you tell which specific creative element drove a result today, or only how the whole campaign performed?" cross-check: also Track D
why we askTests attribution granularity — informs whether the product's reporting needs to isolate creative-element-level performance.
Content ↔ decisioning integration (4 min)
Deciding an offer/audience and producing the content for it usually happen together in practice. Checks whether splitting them removes the bottleneck or just relocates it.
"When an audience or offer decision changes, how much content has to change with it? Walk me through the last time that happened."
why we askTests the core architecture assumption that content and targeting decisions are tightly coupled.
Track D — Marketing Ops/Governance Lead (~15–18 min)
Invisible in the product's UX but gates enterprise approval — owns audit trail and override capability. Doesn't need a full session, just these targeted questions.
Warm-up (2 min)
"Walk me through your role and how long you've been in it — what exactly do you audit or govern?"
why we askBaseline classification — establishes the scope of this persona's authority.
Audit & compliance ownership (4–5 min)
"Walk me through what your audit or compliance review actually covers for a campaign, and how often it happens."
why we askMaps today's audit/compliance process — the concrete shape the product's audit trail must match.
"What metrics or reports do you personally review daily, weekly, or monthly to monitor compliance and campaign health?"
why we askInforms what the product's governance dashboard should surface by default.
"Tell me about the last time you blocked a campaign or a tool from going live. What triggered the block, and what happened next?"
why we askThe actual block triggers in this org — the concrete checklist an agent's guardrails must satisfy to survive review (H6).
Comfort with automation & approval boundaries (4–5 min)
"Overall, how comfortable are you with automated or AI-driven tools making decisions inside your campaigns?"
why we askGeneral autonomy-comfort baseline, ahead of the specific decision-level boundary question below (H4).
"What decisions do you think should always have a human approver before a campaign goes live — for example, who gets targeted, what offer goes out, which channel is used, or the content itself?"
why we askDirectly defines the approval-tier boundary (H4) — parallels the focus group's governance round-robin for cross-check.
Content governance sign-off (2–3 min)
"What do you personally review or sign off on before a content variant goes live, and what would make you block one?" cross-check: also Track C
why we askCompared against Track C's Personalization/governance answer — same question, two vantage points.
Attribution visibility (2–3 min)
"Does your attribution or analytics setup let you isolate which creative element drove a result, or only campaign-level performance?" cross-check: also Track C
why we askCompared against Track C's Content performance & attribution answer.
Phase 2 — Focus Group: Concept Reaction (~65–75 min, 4–6 Track A marketers, same vertical)
Not to re-litigate JTBD or pain — pressure-tests the concept itself and surfaces where the group agrees vs. splits on trust and control. Recruit from Track A participants, all from the same vertical. Run with two people — one facilitating, one taking notes.
Intro & ground rules (5 min) — no wrong answers; disagreement is the useful outcome; individual view matters even if the group leans one way.
Demo / product walkthrough (10–15 min) — show the concept end to end, stage by stage.
why we askTailor this using what Phase 1 interviews surfaced as top pain points for this vertical — don't run it cold/generic.
Silent written reaction (5 min) — before anyone speaks, everyone privately writes down their first reaction and one concern.
why we askPrevents the first person to speak from anchoring the whole room.
Round-robin — first reactions, positives & negatives (15 min): "What was your first reaction? What's the strongest and weakest part of this?"
why we askRotate who answers first each round so the same voice doesn't set the tone every time.
Round-robin — open questions & improvements (10 min): "What questions do you still have? What would you change or add?"
Round-robin — governance & human approval (15 min): "Which of these should always need a human's sign-off before a campaign goes live, and which should never need one — targeting, timing, channel, offer, content?"
why we askDirectly parallels Track D's approval-boundary question — compares the group's collective view against the Governance Lead's individual view.
Wrap-up (5 min): "If we offered your team a pilot next quarter, would you take it — what would you need to see first, and what's the one thing that would make you say no?"
why we askCommitment beats compliments — a countable pilot-intent signal per participant plus their explicit preconditions.
Synthesis summary (fill after the interview)
Fill this in right after the session while it's fresh. Be stingy with "Confirmed" — one enthusiastic anecdote is "leaning", not confirmed.
