Competitive Analysis With Codex: The 2026 AI Agent Guide (Free Prompts)

mkdir competitive-analysis && cd competitive-analysis

codex

## Task: Discover My Direct Competitors

I run [yourcompany.com], which is [one-sentence description of what you sell and to whom].

### Part A: Organic search overlap
1. Visit my homepage. Extract what my product does, who it targets, and 3-5 key features.
2. Search the web for "[my product category] alternatives", "[my product category] vs", and "[my product category] software". Collect domains that appear across multiple searches.
3. Check G2, Capterra, and Reddit for my product category. Note domains that appear in comparisons and discussions alongside products like mine.

### Part B: Paid search signals
Search for "[my brand name] alternatives" and "[my brand name] vs". Note which domains run ads or appear in review roundups.

### Part C: Classify and rank
Categorize each as:
- **Direct** — same solution, same audience
- **Indirect** — different solution, same audience
- **Aspirational** — same solution, larger or different audience

### Output
Save to `competitors.md`:
- Ranked table of direct competitors (name, domain, why they compete, confidence level)
- Secondary list of indirect competitors
- Any data gaps (e.g., no paid search API configured)

# CloudDesk — Direct Competitors

| # | Competitor | Domain | Why They Compete | Confidence |
|---|-----------|--------|-----------------|------------|
| 1 | HelpScout | helpscout.com | Shared "help desk software" queries; SMB support teams | High |
| 2 | Freshdesk | freshdesk.com | Overlaps on "customer support software" and SMB segment | High |
| 3 | Zendesk | zendesk.com | Dominates enterprise; overlaps on mid-market | High |
| 4 | Intercom | intercom.com | Competing on "customer messaging platform" queries | Medium |
| 5 | Zoho Desk | zoho.com/desk | Price-competitive; appears in "cheap help desk" searches | Medium |

## Indirect
- **Front** (front.com) — shared inbox, not full helpdesk; overlapping audience
- **HubSpot Service Hub** (hubspot.com) — CRM-first; competes when buyer uses HubSpot

## Data Gaps
- Paid search overlap: not verified (no SEMrush/Ahrefs API configured)

## Task: Compare Market Reach

Use the competitor list in `competitors.md`.

### Part A: Organic search visibility
For each competitor (and my site):
1. Search "[competitor name]" — note SERP ownership, knowledge panels, sitelinks
2. Search "[product category] software" — note who appears on page 1
3. Search "[product category] vs" — note comparison page rankings
4. Estimate relative organic footprint: Largest / Large / Medium / Small

Label everything [ESTIMATE] unless confirmed by API data.

### Part B: Social presence
For each competitor, check public profiles:
- LinkedIn company page: followers and recent posting frequency
- Twitter/X: followers and engagement on last 5 posts
- YouTube: subscriber count and last upload date (if relevant)

### Part C: AI search visibility (2026 dimension)
Test these queries on ChatGPT, Perplexity, and Google AI Overviews:
1. "best [product category] software"
2. "[product category] for [target audience]"
3. "[competitor A] vs [competitor B]"
4. "alternatives to [my brand]"
5. "[specific use case] tool"

For each query + platform, record which brands are mentioned, whether my brand appears, and the tone. If you can't access an AI platform programmatically, test manually and note it.

### Output
Save to `market-reach.md`:
1. Organic SOV comparison table (label all estimates)
2. Social presence comparison table
3. AI visibility heatmap: competitors × platforms × queries
4. Gap analysis with priority ratings (🔴 High / 🟡 Medium / 🟢 Low)

# CloudDesk — Market Reach

## Organic Search (Estimated from SERP observation, 2026-07-30)

| Competitor | Brand SERP | Category Page 1 | Comparison Queries | Footprint |
|-----------|-----------|----------------|-------------------|----------|
| Zendesk | Strong (full SERP control) | Yes | High | Largest |
| Freshdesk | Strong | Yes | Medium | Large |
| HelpScout | Good (no knowledge panel) | Partial | Medium | Medium |
| CloudDesk (us) | Weak | Low (page 2-3) | Very low | Small |

⚠️ All estimates from manual observation. No API data.

## AI Search Visibility (Manual Sample, 5 Queries, 2026-07-30)

| Query | ChatGPT | Perplexity | Google AIO |
|-------|---------|------------|------------|
| "best help desk software" | Zendesk, Freshdesk, HelpScout | Zendesk, Freshdesk, Zoho Desk | Zendesk, Freshdesk |
| "help desk for small teams" | HelpScout, Freshdesk | HelpScout, Zoho Desk | HelpScout, Freshdesk |
| "alternatives to Zendesk" | Freshdesk, HelpScout, Intercom | Freshdesk, HelpScout, Zoho Desk | Freshdesk, HelpScout |

⚠️ CloudDesk did not appear in any AI answer for these queries.

## Gap Analysis

| Gap | Who's Winning | Priority |
|-----|--------------|----------|
| Category page 1 presence | Zendesk, Freshdesk | 🔴 High |
| AI answer visibility | Zendesk, HelpScout, Freshdesk | 🔴 High |
| Social presence (LinkedIn) | All competitors | 🟡 Medium |
| Comparison query coverage | Zendesk, Freshdesk | 🟡 Medium |

## Task: Extract and Compare Pricing and Features

Use the competitor list in `competitors.md`.

### Part A: Extract pricing
For each competitor, visit their public pricing page. For each plan:
- Plan name, monthly price, annual price (if shown)
- User/seat limit
- Key features included at this tier
- Features locked behind higher tiers
- Free trial? Free plan? Length and limits?

### Part B: Build feature matrix
From each competitor's feature page, collect all named features. Mark each:
- ✅ Has it
- ❌ Doesn't have it
- ⚠️ Partial or add-on
- ❓ Could not verify

### Part C: Structure the comparison

**Pricing Table:**
| Competitor | Cheapest Plan | Mid-Tier | Full Plan | Free Tier? | Free Trial? |
|-----------|--------------|----------|-----------|------------|-------------|

**Feature Matrix:**
| Feature | My Product | Competitor A | Competitor B | Competitor C |
|---------|-----------|-------------|-------------|-------------|

### Output
Save to `pricing-features.md` with source URLs and access dates for every price.

### Rules
- Only record prices visible on public pages. Login-gated pricing → mark "Login-gated"
- Never guess a price
- Be as honest about your own product as competitors — don't inflate your checkmarks

## Task: Analyze Competitor User Reviews

For each competitor in `competitors.md`, analyze 2–4 star reviews from G2, Capterra, and Trustpilot.

### Part A: Collect reviews
Find each competitor's G2, Capterra, and Trustpilot profile pages. Collect up to 30 reviews per competitor, prioritizing:
- Most recent first
- 2–4 star ratings (most actionable feedback)
- Reviews with substantial written text

### Part B: Analyze
For the collected reviews, identify:
1. Recurring praise themes — what do users consistently like?
2. Recurring complaints — what do users consistently criticize?
3. Feature requests — what do users wish existed?
4. Switching reasons — why did users leave or consider leaving?
5. Onboarding experience — how do users describe their first weeks?
6. Support quality — how do users describe support interactions?

### Part C: Per-competitor summaries
Top 3 strengths (with short illustrative fragments, not full quotes)
Top 3 weaknesses (with short illustrative fragments)
One sentence: "Users choose [Competitor] because _____, but they complain about _____."

### Part D: Cross-competitor patterns
What does the market consistently want? What seems unsolved across all competitors?

### Output
Save to `review-analysis.md`.

### Rules
- Only use review text you can access. If a platform blocks you, note it and skip.
- Quote fragments must be short (fair use).
- Do not make up reviews or sentiments.

## Task: Benchmark Purchasing Experience

For each competitor, evaluate as a logged-out prospective buyer. Score each dimension 1–5 (5 = best) with a specific observation:

1. **Pricing transparency** — visible without login? clear plan differences?
2. **Free trial / freemium** — can you try without talking to sales?
3. **Signup friction** — steps from landing page to working product? credit card required?
4. **Demo availability** — screenshots, videos, interactive demo, sandbox?
5. **Sales contact** — how easy to find contact info or start a chat?
6. **Cancellation clarity** — check help center: is cancel/downgrade clear?
7. **Documentation quality** — public knowledge base? setup guides clear?

Also flag any dark patterns: hidden prices, forced sales calls, confusing tiers, hard-to-find cancel buttons.

### Output
Save to `purchasing-experience.md`:
- Scorecard table (all competitors, all dimensions)
- Best practices observed
- Friction points found
- Your own gaps compared to best-in-class

### Important
Stop before entering real personal information when testing signup flows.

## Task: Synthesize My Competitive Analysis

I have completed my competitive analysis. Findings are in:
- `competitors.md`
- `market-reach.md`
- `pricing-features.md`
- `review-analysis.md`
- `purchasing-experience.md`

### Step A: Executive Summary
Write 3–4 paragraphs covering:
- Top 2–3 competitive threats and why
- Biggest visibility gap (where competitors reach customers that we don't)
- Biggest product or pricing gap
- Most urgent single recommendation

### Step B: Priority Matrix

| | Easy to Fix | Hard to Fix |
|---|---|---|
| **High Impact** | Do this week | Plan this quarter |
| **Low Impact** | When convenient | Deprioritize |

Place at least 3 specific actions into the matrix.

### Step C: Recommendations by Team
- **Product team:** 1–3 specific features to build, fix, or improve (name the competitor and gap)
- **Marketing team:** 1–3 content, campaign, or positioning actions
- **Sales team:** 1–3 competitive objections to prepare battlecards for

### Step D: What We Don't Know
List every question the analysis couldn't answer: missing API data, login-gated pages, estimates instead of measurements, assumptions needing human verification.

### Output
Save to `competitive-analysis-report.md`.

### Style Rules
- Every recommendation names a specific competitor and a specific gap
- Don't say "improve SEO" — say "build 5 comparison pages for queries where Zendesk ranks #1–3"
- Don't say "fix pricing" — say "our mid-tier costs $X more than HelpScout's while missing feature Y"
- Mark uncertain claims with [NEEDS VERIFICATION]

mkdir -p .agents/skills/competitive-analysis

---
name: competitive-analysis
description: Run a complete competitive analysis. Use when user says "analyze my competitors", "competitive analysis for my site", "compare us against competitors", "who are my competitors", or "competitive research".
---

# Competitive Analysis Agent

You run a structured 6-phase competitive analysis. Save all results to organized markdown files. Flag every data gap and assumption. Never guess a metric.

## Phase 0: Confirm Scope

Before any research, ask:
1. Target website/domain?
2. What does the product do and who is the audience? (one sentence each)
3. How many competitors? (default: 5, max: 8)
4. Any API keys configured? (DataForSEO, Firecrawl, Semrush, Ahrefs) — if none, use visible web data and label everything [ESTIMATE].

Wait for answers before proceeding.

## Phase 1: Competitor Discovery → `competitors.md`
1. Visit the target homepage. Extract product description, audience, key features.
2. Search for "[product category] alternatives", "[product category] vs", "[product category] software". Collect domains.
3. Search G2, Capterra, Reddit for the product category. Note domains that appear alongside the target.
4. Classify as Direct / Indirect / Aspirational. Rank by competitive closeness.
5. Save with confidence levels and data gaps.

## Phase 2: Market Reach → `market-reach.md`
1. Organic: Search "[competitor name]" and "[product category] software". Note SERP presence. Label [ESTIMATE].
2. Social: LinkedIn, Twitter/X, YouTube follower counts and posting frequency from public profiles.
3. AI visibility: Test 5 category queries across ChatGPT, Perplexity, Google AI Overviews. Record which brands are cited, sentiment, and whether the target appears.
4. Build gap analysis with priority ratings.

## Phase 3: Pricing and Features → `pricing-features.md`
1. Visit each competitor's public pricing page. Extract plans, prices, user limits, features per tier, trial/free tier info.
2. Build pricing comparison table and feature matrix (✅/❌/⚠️/❓).
3. Record source URLs and access dates. Login-gated pages → mark "Login-gated".

## Phase 4: Review Analysis → `review-analysis.md`
1. Find G2, Capterra, Trustpilot profiles. Collect 2–4 star reviews (up to 30 per competitor).
2. Analyze: praise themes, complaints, feature requests, switching reasons, onboarding, support.
3. Per-competitor: top 3 strengths, top 3 weaknesses, one-sentence value proposition from user perspective.
4. Cross-competitor: unsolved problems across the market.

## Phase 5: Purchasing Experience → `purchasing-experience.md`
1. Score each competitor 1–5 on: pricing transparency, free trial, signup friction, demo availability, sales contact, cancellation clarity, documentation quality.
2. Note dark patterns.
3. Stop before entering real personal information during signup tests.

## Phase 6: Synthesize Report → `competitive-analysis-report.md`
1. Read all five saved files.
2. Executive summary (3–4 paragraphs).
3. Priority matrix (Impact × Difficulty) with ≥3 specific actions.
4. Recommendations for product, marketing, and sales teams.
5. "What We Don't Know" section.

## Data Integrity Rules (Never Violate)
1. Never guess a metric. If you can't see it, record "Not publicly available" or "Login-gated."
2. Label every estimate with [ESTIMATE].
3. Record source URL and access date for every data point.
4. Only use review text you can access. Short fragments only (fair use).
5. Never write API keys, passwords, or tokens into any output file.
6. Score your own product with the same honest criteria as competitors.

## Output Files
- `competitors.md`
- `market-reach.md`
- `pricing-features.md`
- `review-analysis.md`
- `purchasing-experience.md`
- `competitive-analysis-report.md`

## Limits
- Max 8 competitors per run (default: 5)
- Max 30 reviews per competitor
- Max 10 AI search queries per run

$competitive-analysis yourdomain.com

Run a competitive analysis for my site [yourdomain.com]