competitive-analysis

name	competitive-analysis
description	Perform competitive market analysis with comparisons and strategic recommendations.

Competitive Analysis

Identify competitors, analyze positioning, and deliver actionable recommendations. Skip textbook frameworks (Porter's, PESTLE) unless specifically requested — they're MBA artifacts, not operator tools.

When to Use

• "Who are my competitors?" / "How do we compare to X?"

• Feature comparison matrix or positioning map needed

• Fundraising deck competition slide

• Finding market gaps

When NOT to Use

• General market sizing (use deep-research)
• SEO-specific competitor keyword analysis (use seo-auditor)

What Practitioners Actually Use

Skip Porter's Five Forces. Operators use these four:

1. April Dunford's Positioning (from "Obviously Awesome") — the most-used positioning method in B2B SaaS. Five inputs in strict order:

1. Competitive alternatives (what customers would do if you didn't exist — including "spreadsheets" and "nothing")

2. Unique attributes you have that alternatives lack

3. Value those attributes deliver (with proof)

4. Best-fit customer characteristics

5. Market category you win in

Key insight: positioning starts from alternatives, not features. Your "competitor" might be Excel.

2. Wardley Mapping (Simon Wardley, free book at medium.com/wardleymaps) — plot components on two axes: visibility-to-user (y) vs evolution Genesis → Custom → Product → Commodity (x). Reveals: where competitors overinvest in commoditizing components, where to build vs buy, what's about to become table stakes. Tool: onlinewardleymaps.com (free). Best for platform/infra competition.

3. Feature comparison matrix — the unglamorous workhorse. Rows = capabilities, columns = competitors, cells = ✓/✗/partial. Battlecards for sales teams are this + "trap-setting questions." Key: weight features by how often they appear in lost-deal notes, not by what engineering thinks matters.

4. Kano mapping applied to competitors — categorize each competitor feature as Basic (expected, table stakes), Performance (more = better), or Delighter (unexpected). Kano's insight: today's delighters become tomorrow's basics. Competitors' delighters tell you where the bar is moving.

Research Toolchain

| Need | Tool | How to use |

|---|---|---|

| Find competitors | webSearch("[product] alternatives site:g2.com") | G2's "alternatives" pages are crowdsourced competitor lists |

| Verified user complaints | webSearch("[competitor] site:g2.com"), Capterra, TrustRadius | Filter reviews to 1-3 stars. Look for repeated phrases — those are exploitable weaknesses |

| Enterprise IT buyers | PeerSpot (formerly IT Central Station) | More technical, less marketing-gamed than G2 |

| Pricing (often hidden) | webFetchcompetitor /pricing page, Wayback Machine for historical,webSearch("[competitor] pricing reddit") for leaked enterprise quotes | |

| Tech stack | webFetch("https://builtwith.com/[domain]") — 673M+ sites, 85k+ technologies. Wappalyzer similar. | Reveals: are they on legacy stack? What vendors? Switching cost signals |

| Traffic/channel mix | SimilarWeb (reliable for large sites, unreliable <50k visits/mo) | See which channels drive competitor traffic |

| Funding/team size | Crunchbase free tier, webSearch("[competitor] raises TechCrunch") | |

| Strategic direction | webSearch("[competitor] site:linkedin.com/jobs") — hiring = roadmap. 5 ML engineers = AI features in 6mo. | |

| Historical messaging | webFetch("https://web.archive.org/web/2024*/[competitor].com") | Shows positioning pivots — what they tried and abandoned |

| SEO/content strategy | Ahrefs (paid, $129+/mo) or webSearch("site:[competitor].com") to map content | |

| Regional competitors (non-US) | App Store: curl -sL "https://apps.apple.com/us/search?term=[name]"· Play Store:webSearch("[name] site:play.google.com") · Search in local language | G2/Product Hunt have near-zero coverage outside US/EU |

Analyzing JS-Rendered Competitor Sites (SPAs)

Many early-stage competitors build their site as a React/Vue/Svelte SPA. webFetchandwebSearchreturn empty results for these — the HTML shell has nothing but<div id="root"></div>. Always trycurl via bash first:

Step 1 — Get meta tags from the HTML shell (works even on SPAs)

curl -sL --max-time 10 "https://competitor.com/" | head -100

# → Reveals: title, meta description, OG tags, and the JS bundle filename

Step 2 — Extract readable strings from their JS bundle

# Get the bundle filename from Step 1 output (e.g. /assets/index-AbCd1234.js), then:

curl -sL "https://competitor.com/assets/index-[hash].js" \\

| grep -oE '"[^"]{10,200}"' \\

| grep -iE "(price|plan|free|gratis|feature|bank|sync|categor|budget|presupuest)" \\

| sort -u | head -80

Step 3 — Check static files for site structure

curl -sL "https://competitor.com/robots.txt"

curl -sL "https://competitor.com/sitemap.xml"

curl -sL "https://competitor.com/manifest.json"

This approach can reveal pricing plans, feature descriptions, FAQ content, legal disclaimers, and marketing copy — all embedded in the compiled JS. For example, this technique on novia.com.do surfaced: their two pricing tiers ("Plan Cita" = free, "Plan Compromiso" = premium coming soon), their exact value proposition copy, a disclaimer that they have no real-time bank connection, and their full FAQ structure.

Critical Rule: When the User Names a Specific Competitor

If the user tells you a competitor exists by name, believe them immediately. Do not attempt to verify through web searches first — that costs multiple failed rounds. Instead:

1. Ask for the URL right away if you don't already have it: "Can you share their website or App Store link?"

2. Once you have a URL, go directly to the SPA research chain (curl → bundle extraction) — do not webFetch or webSearch the name first.

3. If you have no URL and searches return nothing, ask the user for 3–5 key facts: what it does, price, platform, 1–2 differentiators. User knowledge > empty search results every time.

The user naming a competitor is the highest-confidence intelligence you will receive. Don't waste rounds trying to independently confirm what they already know.

Zero Search Results Protocol

When webSearchandwebFetch both return empty results for a competitor, follow this decision tree in order — do not repeat the same failed approach with different keywords:

1. Try curl on the homepage (catches SPA meta tags — see SPA section below)

2. Try curl on their JS bundle (catches pricing, features, copy — see SPA section)

3. Try App Store / Play Store — search apps.apple.comandplay.google.com directly for the app name; the store listing contains descriptions, screenshots text, and reviews

4. Try social proof search — webSearch("[name] instagram OR twitter OR linkedin") — even tiny startups have social profiles

5. Ask the user — if all above fail, ask for 3–5 facts and build the profile from that. State clearly: "I can't find [name] through any of my research tools. Can you tell me what they do, their price, and their main features?"

Empty results ≠ no competition. In niche, regional, or non-English markets (Latin America, Caribbean, Southeast Asia, MENA), G2/Capterra/Product Hunt have near-zero coverage. An app can have 10,000 active users and zero English web presence.

Niche and Regional Market Research

Standard tools (G2, Capterra, Product Hunt) are US/English-centric. For non-US markets use these instead:

| Market type | Where to look |

|---|---|

| Latin America | App Store regional search (apps.apple.com/[country-code]/), Google Play withhl=es, local tech blogs (Contxto, Startup Genome LatAm), AppFollow for store rankings |

| Caribbean / Emerging | Instagram/Facebook pages (startups in these markets often exist only on social), local chambers of commerce, fintech-specific databases (Finnovista, IDB Fintech census) |

| Any niche market | Search in the local language — always try Spanish/Portuguese/French/etc. queries, not just English. A Dominican app will not appear in English results. |

| App stores (universal) | curl -sL "https://apps.apple.com/[country]/search?term=[query]" and Play Store search — app descriptions contain features and positioning that web searches miss |

Rule: Always run at least one search in the target market's primary language before concluding no local competitors exist.

Methodology

Step 1: Frame — Get from user: their product, target customer, and who THEY think competes. Their list is always incomplete — but treat every name they give you as a confirmed lead, not a hypothesis to verify.

Step 2: Expand the competitor set — Run webSearch("[known competitor] alternatives")andwebSearch("[category] vs"). Check G2 category pages. Add indirect competitors (different product, same job) and the "do nothing" option. Also search in the local market language if the product targets a non-English market.

Step 3: Per-competitor dossier — For each (limit to 5-7 for depth):

• Positioning one-liner (their homepage H1 or title tag)

• Pricing model + tiers (try /pricing; if SPA use curl; if nothing found, note it explicitly)

• Top 3 strengths (from 5-star reviews, or inferred from their own copy if no reviews exist)

• Top 3 weaknesses (from 1-2 star reviews — use exact customer language; if no reviews, infer from what they don't mention)

• Funding stage + headcount (Crunchbase/LinkedIn)

• Recent product launches (changelog, blog, Product Hunt)

If a competitor is a direct local rival (same market, same core use case, same pricing model): flag them separately as "Direct Competitor" and add a dedicated head-to-head comparison section — don't just include them as one row in the matrix. Make clear: what does the user's product do that this one doesn't, and vice versa?

Step 4: Synthesize — Build the feature matrix. Plot on a 2×2 (pick the two axes the buyer cares about, not the ones that make user look good). Identify white space.

Step 5: Recommend — Not "monitor the threat." Specific: "Competitor X's reviews mention slow support 23 times — lead with your SLA in sales calls."

Ongoing Monitoring

Competitive landscapes shift constantly — pricing changes, features launch, funding rounds close. The analysis you just delivered is a snapshot. This section defines monitoring as a second deliverable the agent offers immediately after the report, not optional advice.

Always Offer Monitoring Activation

At the end of every competitive analysis, always ask:

"Would you like me to set up ongoing monitoring for these competitors? I'll generate your alert links, a bookmark bundle, and a monthly ritual checklist — takes about 2 minutes to activate. Where do you want alerts to land?"

Do not skip this offer. Present it as a distinct, activatable package — not a footnote.

Channel Decision Tree

Ask the user which channel they prefer, then follow the matching setup path. Priority order: Slack → Teams → Email → Manual file.

Does the user have Slack?

YES → Set up Slack RSS feeds + optional webhook

NO → Does the user work in a Microsoft org (Teams)?

YES → Set up Teams RSS connector

NO → Set up Google Alerts (email)

Always: also save competitor-monitoring.md to the project regardless of channel chosen.

If the user is unsure or doesn't answer, default to Google Alerts (zero-friction, universally available) plus the project file.

Per-Channel Setup Instructions

Slack

Option A — RSS feeds (native, free, no admin required)

/feed subscribe https://[competitor].com/changelog/rss

/feed subscribe https://[competitor].com/blog/rss

Run these commands in a dedicated #competitive-intel channel. The agent should generate the exact feed URLs for each competitor and give the user the ready-to-paste commands.

Option B — Google Alerts → Slack (via Zapier or Make, free tier)

1. Create a Gmail label called "Competitor Alerts" (for each Google Alert)

2. In Zapier/Make: Trigger = new email with label "Competitor Alerts" → Action = post message to #competitive-intel

3. The agent provides exact Zapier/Make setup steps when this option is chosen

Option C — Incoming webhook (advanced)

When the user runs the monthly ritual on demand, the agent can post a summary to a Slack webhook URL. The user provides the webhook URL; the agent stores it in the monitoring brief for future use.

Microsoft Teams

RSS connector (native, free)

1. Open the target channel → click ... → Connectors → RSS

2. Paste the feed URL and set a display name

3. The agent generates the exact feed URLs for each competitor's changelog and blog

Power Automate (for Microsoft 365 orgs)

Trigger: new email matching subject pattern → Post message in Teams channel. The agent describes the exact flow when this option is chosen.

Google Alerts (Email — universal fallback)

For each competitor, generate a ready-to-click alert URL:

https://www.google.com/alerts?q=[COMPETITOR+NAME]&hl=en&gl=us&source=web&hl=en

Replace [COMPETITOR+NAME] with the URL-encoded competitor name. The user clicks once to activate. Recommend setting delivery to "As it happens" for major competitors, "At most once a week" for minor ones.

Also generate alerts for: "[competitor] pricing","[competitor] raises","[competitor] launches".

Manual Project File (always generated)

Regardless of which channel is chosen, always save a competitor-monitoring.md file to the project root. This file is the source of truth and last-checked log. Structure:

# Competitor Monitoring Brief

Last updated: [date]

## Competitors tracked

- [Name] — [URL] — Alert channel: [Slack/#channel | Teams/#channel | Email]

## Bookmark bundle

Per competitor: pricing page, changelog/blog, App Store listing, LinkedIn jobs, G2 reviews page

## Change log

| Date | Competitor | What changed | Action taken |

|------|------------|-------------|--------------|

The Monitoring Package Deliverable

When the user activates monitoring, the agent generates and delivers four items:

1. Alert bundle — ready-to-click Google Alert URLs (or paste-ready Slack/Teams commands), one set per competitor

2. Bookmark list — 5 URLs per competitor: pricing page, changelog or blog, App Store listing, LinkedIn jobs, G2/Capterra reviews page

3. Ritual card — the 30-minute monthly checklist (see below), saved to competitor-monitoring.md

4. On-demand command — a natural language prompt the user can paste back at any time: "Run my monthly competitive check for [competitor name]" — the agent then runs the ritual and updates the monitoring brief

30-Minute Monthly Ritual

A structured, timed sequence the user (or agent, on demand) runs once per month per competitor:

| Minutes | Task | What to look for |

|---|---|---|

| 0–5 | Check Google Alert digest (or Slack/Teams channel) | Any press coverage, blog posts, or product announcements since last month? |

| 5–10 | Open App Store listing | Any version updates? Read the 3 most recent reviews for new complaint patterns |

| 10–15 | Open their changelog or blog | Any feature launches? Note the category (pricing, UX, integrations, AI) |

| 15–20 | Scan their jobs page | Any new roles? 3+ ML/AI roles = AI feature coming. Enterprise Sales hire = moving upmarket |

| 20–25 | Check pricing page (compare to Wayback Machine if needed) | Any tier changes, price increases, or new plan names? |

| 25–30 | Update competitor-monitoring.md | Log date + what changed. Flag anything that warrants a response |

Output of each ritual: one row added to the change log in competitor-monitoring.md. If a change is significant (new pricing tier, major feature launch, funding round), escalate: "[Competitor] launched [X] — this affects [specific recommendation from the original report]. Consider [specific response]."

On-Demand Ritual Invocation

The user can trigger the ritual at any time by telling the agent:

"Run my monthly competitive check for [competitor name]"

"What's changed with [competitor] since last month?"

"Update my competitive monitoring"

The agent reads competitor-monitoring.md to get the last-checked date and bookmark list, runs through the ritual checklist, and appends findings to the change log.

Output — Ask the User First

Before building any deliverable, ask the user how they want the analysis presented using the query tool:

"How would you like your competitive analysis presented — as a slide deckor awritten report?"

Then follow the appropriate path below. Do not default to one format without asking.

---

Option A: Slide Deck

Load the slides skill and build a Replit slide deck. Follow the slides skill's conventions for manifest, components, and design. Structure the deck as:

1. Title slide — Product name, category, date

2. Executive summary — Positioning statement (Dunford format) + top 3 recommendations

3. Competitive landscape — Table: Company, Stage, Pricing, Strength, Weakness

4. Feature matrix — Rows = capabilities, columns = competitors, cells = checkmark/x/partial, color-coded

5. Positioning map — 2×2 chart (matplotlib/plotly image)

6. White space & opportunities — Gaps + Kano analysis

7. Action plan — Top 3 specific actions + battlecard trap-setting questions

8. Sources — Numbered URLs for every claim

---

Option B: Written Report (PDF + Web Preview)

Do not output a markdown summary.Build a polished competitive analysis report as a professional PDF usingjsPDF, with a React web preview that visually matches page-by-page. The report should look like a strategy consulting deliverable.

Build order: Generate the PDF first and present it to the user. Then build the web preview. The PDF is the primary deliverable — the web app is a visual complement.

Report Structure

1. Page 1 — Executive Summary: Product name, category, date. Positioning statement (Dunford format): For [target customer] who [need], [product] is a [category] that [key benefit]. Unlike [primary alternative], we [key differentiator]. Top 3 strategic recommendations (the "so what").

2. Page 2 — Competitive Landscape: Table with Company, Stage, Pricing, Strength (from reviews), Weakness (from reviews). Funding/headcount context for each competitor.

3. Page 3 — Feature Matrix: Rows = capabilities, columns = competitors, cells = checkmark/x/partial. Weight column (1-5) based on buyer conversation frequency. Color-code: green where the user's product wins, red where it loses.

4. Page 4 — Positioning Map: 2x2 chart with axes based on buyer decision criteria (not vanity metrics). Each competitor plotted with logo or labeled dot. Generated via matplotlib or plotly, embedded as image.

5. Page 5 — White Space & Opportunities: Gaps no one serves well, with evidence from reviews and market data. Kano analysis: which competitor features are Basics vs Performance vs Delighters.

6. Page 6 — Action Plan: Top 3 specific actions with source citations. Battlecard-style "trap-setting questions" for sales calls.

7. Final Page — Sources: Numbered URLs for every claim.

PDF Generation (jsPDF)

Use jsPDF to generate the PDF with explicit point-based layout:

• new jsPDF({ unit: "pt", format: "letter" }) — US Letter: 612×792pt

• Use 36pt margins (0.5in). Content area: 540w × 720h points.

• Track Y position as you render each element. When the next element would exceed PAGE_H - MARGIN, calldoc.addPage() and reset Y to the top margin. Never let content silently overflow — always check before rendering.

• Embed charts as images via doc.addImage() — scale to fit content width while respecting remaining page height.

• Add a headerandfooteron each page.Footer must save/restore Y position — do not let footer drawing move the content cursor, or subsequent content will force blank pages.

• Before any manual page break, check whether a fresh page was already added (track an isNewPage flag). Only add a page if you're not already on a fresh one.

• Required before presenting: After generating the PDF, verify there are no blank pages. If any page is blank, fix the page-break logic and regenerate.

Web Preview

The React web artifact renders the same report data as an HTML version that visually mirrors the PDF page-by-page. Each "page" should be a fixed-size container (816×1056px — US Letter at 96dpi) with the same margins, typography, and chart placement as the PDF.

Honesty Rules

• If the user's product loses on most dimensions, say so — then find the niche where they win

• "No competitors" is never true. The competitor is always at least "build it yourself" or "do nothing"

• Flag when data is thin (e.g., "SimilarWeb shows <50k visits — estimate is low-confidence")

• Cite every claim to a URL the user can verify

• Flag inferred data: if a competitor profile was built from JS bundle strings or user-provided facts (not reviews or public docs), say so: "Profile based on homepage copy and compiled JS — no independent reviews available."

• Zero search results is not the same as no competitor. If searches return nothing for a named competitor, say: "I couldn't find [name] in any public database — this likely means they have low SEO footprint, not that they don't exist."

Limitations

• G2/Capterra reviews skew toward mid-market SaaS; thin for enterprise and consumer

• SimilarWeb is inaccurate for sites under ~50k monthly visits

• Cannot access paid CI tools (Klue, Crayon, Kompyte) or PitchBook

• Pricing pages lie — enterprise pricing is almost never public

• webFetch and webSearch fail for: JS-rendered SPAs (use curl), social media URLs (LinkedIn, Instagram, Twitter — blocked by robots), sites behind Cloudflare or login walls, and any product with <1k English web mentions

• Regional/emerging market blind spot: G2, Product Hunt, Crunchbase have minimal coverage of LatAm, Caribbean, SEA, MENA, and African markets. Local competitors in these markets will often have zero results in standard research tools — always fall back to local-language search, App Store regional search, and direct curl analysis

• Monitoring channel assumptions: Slack RSS setup requires the native Slack RSS app to be installed in the workspace (free, but needs a workspace admin in some orgs). Teams RSS connector requires a Teams channel with Connectors enabled. If neither is available, fall back to Google Alerts → email without requiring any third-party tools or admin permissions.