Pre-Build Research Process (CQ)¶
Owner: Vi | Updated: 2026-02-05
Purpose¶
Before building anything, devs must research what already exists in the market and understand what "great" looks like. This prevents building the wrong thing and ensures we imitate the best, not invent from scratch.
The dev is a mini product manager. They own finding and studying references—not waiting to be told.
Why This Matters¶
| Without Research | With Research |
|---|---|
| Build from imagination | Build from proven examples |
| Reinvent the wheel | Imitate the best |
| "I thought it should work this way" | "ChatGPT does it this way, so we do too" |
| Rework after Erik says "this isn't good enough" | Get it right the first time |
Reality: Our Vietnamese dev team has limited exposure to US/international products. This process gives them that exposure systematically.
When to Do This¶
Before starting ANY Alpha work.
| Situation | Research Required? |
|---|---|
| New feature (Alpha) | Yes - mandatory |
| Beta refinement | Only if direction unclear |
| Final polish | No - scope is locked |
| Bug fix | No |
| Small tweak | No |
Rule: If you don't know what "great" looks like for this feature, you haven't researched enough.
The Research Process¶
Step 1: Define What You're Building¶
Before searching, write down:
Feature: [What am I building?]
Problem: [What user problem does this solve?]
Keywords: [What would I search to find similar products?]
Example:
Feature: Email sequence builder
Problem: Users want to send automated follow-up emails
Keywords: "email sequence tool", "drip campaign builder", "cold email automation"
Step 2: Use AI to Find Products¶
Use Claude/ChatGPT to discover what exists:
Prompt template:
I'm building [feature description] for a B2B SaaS product.
1. What are the best products in the market that do this?
2. What makes each one great?
3. Which one should I study most closely and why?
4. What are the must-have features users expect?
Follow-up prompts: - "Show me how [Product X] handles [specific flow]" - "What do users complain about with [Product X]?" - "What's the simplest version of this that still works?"
Step 3: Study the Best Examples¶
Pick 2-3 products to study deeply.
Minimum study requirements:
| Activity | Required |
|---|---|
| Actually use the product | Yes - sign up, try it |
| Watch demo/tutorial on YouTube | Yes - see how experts use it |
| Take screenshots of key flows | Yes - reference for building |
| Note what makes it great | Yes - don't just copy blindly |
What to capture: - How does the user flow work? - What's the UI pattern? - What happens on edge cases? - What's surprisingly simple? - What would I do differently?
Step 4: Document Your Findings¶
In the ClickUp task, add a comment:
## Pre-Build Research
**Products studied:**
1. [Product A] - [link]
2. [Product B] - [link]
3. [Product C] - [link]
**Best example to imitate:** [Product X] because [reason]
**Key learnings:**
- [What makes it great]
- [Flow I'll imitate]
- [What I'll do differently and why]
**Screenshots:** [attached or linked]
**Ready for Alpha:** Yes
Step 5: Get Feedback Before Building¶
Share your research with Vi before writing any code.
Vi checks: - [ ] Dev studied real products (not just imagined) - [ ] Chosen reference makes sense - [ ] Dev can explain what they'll imitate - [ ] Dev understands why it's good, not just what it looks like
Only after Vi approves: Start Alpha.
Research Checklist¶
Before starting Alpha, dev must check all boxes:
- I defined what I'm building and the problem it solves
- I used AI to find best-in-market products
- I actually used at least one competitor product
- I watched a demo/tutorial
- I took screenshots of key flows
- I documented my findings in the task
- I shared with Vi and got approval
If any box is unchecked: You're not ready to build.
What "Studying" Looks Like¶
Good Study¶
- Signed up for Mailchimp, created a test sequence
- Watched 10-min YouTube tutorial on "How to create drip campaigns"
- Took 5 screenshots of the sequence builder UI
- Noted: "They show a visual timeline, very clear"
- Can explain to Vi exactly what they'll build
Bad Study¶
- "I looked at screenshots on Google"
- "I know how email works"
- "I've seen this before"
- Can't explain what makes the reference product good
Consequences for Skipping¶
If a dev skips research and builds from imagination:
| Offense | Consequence |
|---|---|
| First time | Stop work. Go back to research. Deadline adjusts. |
| Second time | Conversation with Vi about why this keeps happening |
| Pattern | Escalate to Ryan |
This is not optional. Building without research wastes everyone's time.
Vi's Role¶
Vi ensures devs do research before building:
- Check that research is documented in task before Alpha starts
- Ask "What did you study? What are you imitating?"
- Push back if research is thin ("Actually use the product, then come back")
- Don't approve Alpha start without research complete
Vi is the gate. No research = no Alpha approval.
Examples¶
Good Research Documentation¶
## Pre-Build Research - Email Sequence Feature
**Products studied:**
1. Mailchimp - https://mailchimp.com (market leader)
2. Lemlist - https://lemlist.com (cold email focused)
3. Apollo - https://apollo.io (B2B sales)
**Best example to imitate:** Lemlist because:
- Clean UI, not overwhelming
- Visual sequence builder (drag and drop)
- Clear preview of what recipient sees
**Key learnings:**
- Users expect visual timeline view
- Must show "Day 1, Day 3, Day 7" delays clearly
- Preview email is essential before sending
- Simple version: linear sequence, no branching
**What I'll do differently:**
- Skip branching logic for Alpha (Lemlist has it, but complex)
- Focus on core: create sequence → set delays → preview → send
**Screenshots:** [5 images attached]
**Ready for Alpha:** Yes - Vi approved 2026-02-05
Bad Research Documentation¶
## Research
I looked at some email tools. They have sequence builders. I'll build something similar.
This is not research. This is guessing.
Integration with Other SOPs¶
| SOP | Connection |
|---|---|
| Alpha-Beta-Final | Research happens BEFORE Alpha |
| Task Creation Standards | Task should include reference links if known |
| Definition of Done | Alpha checklist includes "research complete" |
FAQ¶
Q: What if there's no competitor for this feature? A: There's always something related. AI chat → study ChatGPT. Dashboard → study Notion/Linear. Search harder.
Q: How long should research take? A: 2-4 hours typically. If it's taking longer, you're overcomplicating it. If it's taking 30 minutes, you're not going deep enough.
Q: Can I skip research for small features? A: If it's truly small (button, minor tweak), yes. If you're building a new flow or capability, no.
Q: What if Erik already told me exactly what to build? A: Still research. Erik's vision + market examples = better outcome. You might find something Erik didn't think of.
Q: What if I already know this space well? A: Prove it. Document what you know. Show Vi. If you can't articulate what you're imitating, you don't know it well enough.
Summary¶
| Principle | Why |
|---|---|
| Research before building | Build right thing first time |
| Use AI to find products | Efficient discovery |
| Actually use competitor products | Screenshots aren't enough |
| Document findings | Creates accountability |
| Get Vi approval before Alpha | Gate ensures quality |
| Consequences for skipping | This is not optional |
Build from examples, not imagination. Imitate the best.