Run external export-trust interviews

[tonyketcham]

Run the export/ownership demo with at least two target users, record whether JSON/CSV exports improve adoption trust, capture objections, and update the keep/kill/iterate recommendation from participant feedback.

[tonyketcham]

Proposed interview plan for this issue.

Interview definition

Goal: test whether Flatbread's export / ownership story materially increases trust for potential adopters.

We want to learn:

• whether JSON / CSV exports reduce fear of lock-in
• whether the "raw files + Git + exports + typed read surfaces" story feels credible
• what objections still block adoption even after seeing the export path
• whether exports change the keep / kill / iterate recommendation for this PMF slice

Target participants

Aim for at least 2 people who match one of these profiles:

• a TypeScript app developer who owns content-heavy product surfaces
• a tech lead / staff engineer evaluating content architecture tradeoffs
• a developer with prior negative experience around CMS or data lock-in

Nice-to-have variation:

• 1 participant already comfortable with GraphQL
• 1 participant who is skeptical of adding another data/content layer

Session format

• Length: 25-30 minutes
• Format: live screenshare / call
• Artifacts shown:
  • canonical Flatbread posts/authors/tags example
  • data ownership docs / explanation
  • JSON export example
  • CSV export example
• Capture:
  • quotes
  • objections
  • trust signals
  • whether exports changed willingness to try / adopt

Moderator script

1. Intro (2 min)

"Thanks for doing this. I'm evaluating whether Flatbread's export and ownership story actually makes the product feel safer to adopt. I want your honest reaction, especially anything that feels incomplete, confusing, or unconvincing. I'm testing the product story, not you."

2. Context warm-up (3 min)

Ask:

1. How do you currently manage app content or relational content outside your main database?
2. What usually makes you hesitate before adopting a content tool or content layer?
3. How much do portability and data ownership matter in that decision?

3. Baseline trust question, before demo (2 min)

Ask before showing exports:

1. If I said "Git-backed relational content for TypeScript apps," what would your first trust or lock-in concern be?
2. On a scale of 1-5, how safe would this feel to adopt before seeing any export story?

Record the baseline number.

4. Product walkthrough (8-10 min)

Show, in order:

1. Raw content files as source of truth
2. Relations in the posts/authors/tags example
3. Generated read/query surfaces
4. JSON export output
5. CSV export output
6. Brief explanation of ownership / exit story

While showing, narrate succinctly:

• "The source of truth is still plain files in the repo."
• "Exports are meant to prove you can leave with your data in usable formats."
• "GraphQL / generated APIs are read surfaces over the same model, not the only way to access it."

5. Reaction questions, during/after walkthrough (8 min)

Ask:

1. What, if anything, about the JSON / CSV export story increased your confidence?
2. What still feels risky or locked-in even after seeing this?
3. Does the export path feel like a real exit hatch, or just a convenience feature?
4. If your team needed to migrate away later, what evidence would you want beyond what you've seen here?
5. Which mattered more for trust: raw files in Git, JSON export, CSV export, or generated typed access?
6. Did anything in the demo feel hand-wavy or incomplete?
7. Would this export/ownership story make you more likely to trial the product? Why or why not?

6. Forced choice + recommendation (3 min)

Ask:

1. After seeing this, has your trust score changed from the original 1-5? Why?
2. Which of these best fits your reaction?
   • keep pushing this story as a meaningful advantage
   • iterate, but it is not yet persuasive
   • kill/de-emphasize it; it doesn't meaningfully change adoption confidence
3. What is the single most important next improvement to make this more convincing?

Notes to observe during the interview

Watch for:

• whether the participant spontaneously mentions lock-in, migration, reversibility, or auditability
• whether JSON export or CSV export lands as more useful
• whether they trust the export story because of the raw files, not because of the export itself
• whether they ask for proof of re-import / migration workflows
• whether they care more about operational ownership than format portability
• whether they treat CSV as real utility or just stakeholder/reporting sugar
• whether they confuse GraphQL with product lock-in
• whether they perceive the export path as complete enough for a real exit

Specifically capture:

• exact phrases that indicate trust increased
• exact phrases that indicate trust did not move
• what evidence they expected but did not get
• whether their objections are product gaps, documentation gaps, or proof gaps

Note-taking template

Participant

• Role:
• Team/company type:
• Prior content-tool experience:
• Initial trust score (1-5):
• Final trust score (1-5):

What increased trust

What still blocked trust

Most useful ownership signal

• raw files in Git / JSON export / CSV export / typed API / other

Key quotes

Recommendation from participant

• keep / iterate / kill

Best next improvement

Synthesis questions after both interviews

After running at least two sessions, summarize:

1. Did exports materially improve adoption trust, or were they secondary to the Git-native source-of-truth story?
2. Were objections mostly about missing proof, missing workflow, or missing value?
3. Is the right next move:
   • keep: this is clearly a trust amplifier
   • iterate: promising, but needs stronger proof / messaging
   • kill: not a meaningful PMF lever
4. What is the minimum next experiment needed to resolve uncertainty?

Success criteria for closing this issue

I’d consider this issue well-served once we have:

• at least 2 external participant conversations
• baseline + post-demo trust movement recorded
• concrete objections captured verbatim
• a keep / kill / iterate recommendation updated from participant evidence, not self-review