Customer Discovery + Competitive Intelligence

A combined workflow for validating that a problem is real AND mapping who else is solving it. Built for B2B SaaS, managed services, and platform products targeting SMBs or prosumers.

When to Use

• User has a product/startup and wants to validate before building or selling
• User wants to know if competitors exist and who they are
• User asks for "customer discovery for [niche]" or "competitive analysis for [product]"
• Pre-sales research before approaching a market segment

PART A — VALIDATE THE PROBLEM

Deliverables (always produce all 6)

1. Specific pain per niche — not generic "they're busy". Name the task, the time cost, the dollar cost of inaction.
2. Early adopter profile — a specific person with a name, company size, current tool stack, pain moment, and WTP.
3. 5 customer discovery questions — open-ended, non-leading (Mom Test compliant — no "would you use this?").
4. Validation criteria — concrete signals (green/yellow/red light framework).
5. Vitamin or painkiller verdict — per niche, not globally.
6. Cobbled-together solutions today — actual tools, actual prices, actual time cost.

Mom Test Rules (always apply)

• Ask about their life, not your idea
• Ask about past behavior, not hypothetical future behavior
• Never ask "would you pay for X?" — ask "what did you last pay for that does X?"
• A compliment is not a signal. "That's interesting!" = yellow light.
• A commitment (time, money, referral) is a signal.

PART B — MAP THE COMPETITION

Deliverables (always produce all 6)

1. Current behavior — what the target niche actually does today, not what they should do.
2. Direct competitors — companies with the same solution (names, pricing, target customer, overlap %).
3. Indirect competitors — alternatives solving the same job-to-be-done (VAs, freelancers, DIY stacks, doing nothing).
4. The real enemy — the specific behavior or belief WiseChef/product must replace. Usually inertia + a trusted existing solution.
5. Differentiation assessment — honest, feature-by-feature, not marketing claims.
6. Moat assessment — weak moats vs. defensible moats, with specific reasoning.

Research Tool Fallback Chain

Use this order. If one fails, move to the next immediately.

1. web_research (SearXNG) — fastest for broad queries
   ↓ fails if SEARXNG_URL not configured
2. delegate_parallel with gemini-2.0-flash / nemotron-free — good for synthesis
   ↓ fails if LiteLLM routing is broken (unknown url type error)
3. web_extract on specific pricing/blog URLs — most reliable, zero dependencies
   ↓ some URLs 404/403 — have alternates ready
4. Synthesize from domain knowledge — last resort, label clearly as "from domain knowledge, not verified"

Key URLs to try directly for competitive research:

• Pricing pages: https://[competitor].com/pricing or /plans
• Industry blogs: agencyanalytics.com/blog, podcastinsights.com, etc.
• Reddit BLOCKS extraction — skip Reddit, go to industry blogs instead
• Lindy.ai, Relevance AI, HighLevel, Copy.ai, Jasper, Zapier, Make.com, n8n are stable and have scrapeable pricing pages

Pricing Research Targets by Category

AI Agent / Automation Platforms

Social / Content Tools

Report Structure (use this template)

## PART A — VALIDATE THE PROBLEM

### 1. Specific Pain by Niche
#### [Niche 1 Name]
[3-5 specific pains with dollar/time costs]

#### [Niche 2 Name]
[3-5 specific pains with dollar/time costs]

### 2. Early Adopter Profiles
#### [Niche 1]: Meet [Name], [Age] — [Title at Company Type]
- Context: [size, revenue, role]
- Current stack: [real tool names + prices]
- Pain moment: [specific incident]
- Motivation: [what they want to achieve]
- Why early adopter: [specific reason]
- WTP: [price range and condition]

### 3. Five Customer Discovery Questions
1. [Question about life/workflow]
2. [Question about last time they paid/hired]
3. [Question framed around their own narrative]
4. [Question about a specific past failure]
5. [Question about existing tools/automation]

### 4. Validation Criteria
**Green lights (strong signals):**
- [ ] [Specific observable behavior]

**Yellow lights (need more digging):**
- [Vague positive response pattern]

**Red lights:**
- [Specific disqualifying signal]

### 5. Vitamin or Painkiller Verdict
[Per niche judgment with reasoning]

### 6. Cobbled-Together Solutions Today
| Task | Tool | Cost/mo |
|------|------|---------|
| ... | ... | ... |

---

## PART B — MAP THE COMPETITION

### 1. Current Behavior
[What they actually do, not what they should do]

### 2. Direct Competitors
| Company | What They Do | Pricing | Target | Overlap |
|---------|-------------|---------|--------|---------|

### 3. Indirect Competitors
| Alternative | Description | Cost | Why People Choose This |
|-------------|-------------|------|----------------------|

### 4. The Real Enemy
[Behavior or belief that must be replaced. Usually NOT a software tool.]

### 5. Differentiation Assessment
| Claim | Competitive Reality | Honest Assessment |
|-------|--------------------|--------------------|

### 6. Moat Assessment
**Weak moats:**
- ❌ [commoditized factor]

**Stronger moats:**
- ✅ [defensible factor + reasoning]

Pitfalls

1. Reddit is gated — web_extract on Reddit URLs returns "Please wait for verification". Use industry blogs or G2/Capterra reviews instead.
2. delegate_parallel free models fail silently — if you get unknown url type: '/chat/completions' errors, all free model routes are broken. Skip straight to web_extract.
3. SearXNG may not be configured — web_research errors with SEARXNG_URL not configured. Skip to web_extract.
4. Pricing pages often require JS — if web_extract returns navigation/menu text only, the pricing is dynamically loaded. Try the page directly or accept that the free tier + tier names are in the nav text.
5. "Interesting concept" ≠ validation — be strict with the green/yellow/red framework. Most founders hear what they want to hear.
6. Vitamin/painkiller is niche-specific — the same product can be a painkiller for one segment and a vitamin for another. Always assess per niche.
7. The real enemy is almost never software — it's inertia, trusted VAs, or sunk-cost commitment to an existing system. Name it specifically.

Quality Check Before Delivering

• Real competitor names and pricing sourced (not guessed)
• Each niche has a specific named persona (not "a marketing agency")
• Discovery questions pass Mom Test (no "would you use X?")
• Validation criteria include red lights (not just green)
• Differentiation is honest (no marketing spin)
• Moat assessment distinguishes weak from strong moats
• The "real enemy" is behavioral, not technological