/ Article
A practical guide to using LLMs for competitive analysis, with ready-to-use prompts.
What LLM Competitive Analysis Means
LLM competitive analysis is the practice of using large language models to process, analyze, and synthesize competitive data. Competitor websites, pricing pages, reviews, earnings calls, job postings, product documentation, ad copy: LLMs handle what humans do poorly at scale. Reading 500 reviews across three competitors. Comparing 10 pricing pages side by side. Tracking how a competitor's messaging shifted over six months.
The output is structured competitive intelligence: positioning teardowns, pricing comparisons, objection patterns, hiring signals, and messaging drift analysis. A PMM feeds in raw competitive data and gets back analysis that would have taken hours to assemble manually.
This is the analysis engine behind AI competitive intelligence systems. The Practitioner's Guide to AI Competitive Intelligence covers the full system: collection, processing, analysis, and distribution. This guide covers the specific techniques for using LLMs as the analysis layer.
What LLMs Handle Well (and What They Don't)
Strong applications:
- Summarizing large volumes of unstructured text: earnings calls, blog posts, documentation
- Extracting structured data from unstructured sources, like pulling pricing tiers from a competitor's pricing page
- Classifying competitive signals by type and urgency
- Comparing messaging across multiple competitors simultaneously
- Identifying patterns across hundreds of data points: review complaints, job posting themes
- Generating first-draft competitive briefs and battlecard content
Limitations:
- No real-time web access without tool integration. LLMs analyze the data you provide; they don't monitor competitors autonomously.
- Cannot judge strategic intent. An LLM can identify that a competitor removed their enterprise tier. It cannot tell you whether that decision is permanent or experimental.
- Cannot replace market context. A PMM who has spent two years in a market understands competitive dynamics that no model captures from text alone.
- Outputs require review. LLM analysis ranges from excellent to subtly wrong. The PMM is the quality filter.
Six Techniques With Ready-to-Use Prompts
1. Competitor Messaging Teardown
Feed a competitor's homepage copy, pricing page, and key landing pages. Get a structured positioning analysis.
Analyze this competitor's public-facing content and produce
a structured positioning teardown:
[PASTE HOMEPAGE, PRICING PAGE, AND KEY LANDING PAGE COPY]
Extract:
1. Target audience (who are they speaking to?)
2. Market category (how do they position themselves?)
3. Core value propositions (top 3 claims)
4. Differentiation strategy (what do they claim is unique?)
5. Proof points (what evidence do they provide?)
6. Tone and language patterns (technical vs. accessible,
enterprise vs. SMB, feature-led vs. outcome-led)
7. Notable gaps (what's missing from their messaging?)
2. Review Mining at Scale
Feed 50+ competitor reviews from G2 or Capterra. Extract patterns no human would catch by reading individually.
Analyze these [NUMBER] reviews of [COMPETITOR NAME]
from [G2/Capterra]:
[PASTE REVIEWS]
Extract:
1. Top 5 complaints, ranked by frequency, with exact quotes
2. Top 5 praise points, ranked by frequency, with exact quotes
3. Buyer language: the 10 most common phrases used to describe
the product's value
4. Competitor mentions: which alternatives do reviewers reference
and in what context?
5. Segment patterns: do different reviewer types (by role or
company size) praise or complain about different things?
3. Pricing Page Comparison
Feed multiple competitors' pricing pages. Get a structured comparison.
Compare these pricing pages from [COMPETITOR A], [COMPETITOR B],
and [COMPETITOR C]:
[PASTE PRICING PAGE CONTENT FOR EACH]
Produce:
1. Tier structure comparison (number of tiers, naming, positioning)
2. Price point comparison at each tier
3. Feature gating: which features are restricted to higher tiers?
4. Free tier or trial availability
5. Pricing model (per user, per seat, usage-based, flat rate)
6. Notable differences in packaging strategy
7. What each pricing structure signals about target buyer
4. Job Posting Signal Extraction
Feed 10 to 20 recent job postings from a competitor. Identify strategic direction.
Analyze these [NUMBER] job postings from [COMPETITOR NAME]:
[PASTE JOB POSTINGS]
Extract:
1. Roles by department: where are they investing headcount?
2. Technical signals: what technologies, frameworks, or
capabilities appear repeatedly?
3. Seniority distribution: are they building or scaling?
4. Strategic signals: what do these hires suggest about
product direction or market focus?
5. Any patterns that suggest an upcoming product launch
or market expansion?
5. Earnings Call Analysis
Feed a quarterly earnings call transcript. Extract competitive intelligence.
Analyze this earnings call transcript from [COMPANY NAME]:
[PASTE TRANSCRIPT]
Extract:
1. Strategic priorities mentioned by leadership (top 3)
2. Product direction signals (new features, investments, bets)
3. Competitive mentions (any competitors named or referenced?)
4. Market positioning language (how does leadership describe
the company's position?)
5. Customer or revenue signals relevant to competitive analysis
6. Risks or challenges acknowledged
6. Messaging Drift Detection
Feed a competitor's homepage copy from two different time periods. Identify what shifted and what it signals.
Compare these two versions of [COMPETITOR NAME]'s homepage:
VERSION 1 (from [DATE]):
[PASTE EARLIER VERSION]
VERSION 2 (from [DATE]):
[PASTE CURRENT VERSION]
Analyze:
1. What specific language changed?
2. Did the target audience shift?
3. Did the value proposition change?
4. Did the market category framing change?
5. What does this shift suggest about their strategic direction?
6. How should we adjust our own positioning or messaging
in response?
From Prompts to System
Running prompts manually works for ad hoc analysis. It does not scale. If you're running these prompts weekly across five competitors, you've built a manual job, not a system.
The progression:
- Manual prompts. Copy data, paste into an LLM, review output. Good for getting started and one-off analyses.
- Automated pipelines. Use workflow tools like n8n or Make to collect data automatically and pipe it through an LLM API on a schedule. Requires some technical setup.
- Purpose-built platform. A tool that handles collection, LLM analysis, and distribution as a unified system.
The 5-Layer CI Stack in the AI Competitive Intelligence guide covers how to build the full system. The prompts in this guide are the analysis layer.
Where Oden Fits
The six techniques above work manually. Oden operationalizes them. Instead of pasting competitor data into prompts, Oden monitors competitors continuously, runs analysis automatically, and delivers structured outputs: messaging teardowns, pricing comparisons, review analysis, and competitive briefs.
The difference between running these prompts yourself and using Oden is the difference between checking the weather by looking outside and having a forecast system. Both give you information. One runs without you having to remember to look.
Free to start with 500 credits. Paid plans from $89/month.
Frequently Asked Questions
What is LLM competitive analysis?
LLM competitive analysis is the practice of using large language models to process, analyze, and synthesize competitive data: competitor websites, pricing pages, reviews, earnings calls, job postings, and ad copy. It produces structured competitive intelligence that PMMs use for positioning, battlecards, and strategic decisions.
Can I do competitive analysis with ChatGPT or Claude?
Yes. The prompts in this guide work with any major LLM. The limitation is that you need to manually collect and paste the competitive data. LLMs analyze what you provide; they don't monitor competitors autonomously unless connected to collection tools.
How is this different from AI competitive intelligence?
AI competitive intelligence is the full system: collection, processing, analysis, and distribution. LLM competitive analysis is specifically the analysis layer, where an LLM processes competitive data and produces structured output. LLM analysis is one component of a broader AI CI system.
What competitive data works best with LLMs?
Text-heavy sources: pricing pages, homepage copy, blog posts, earnings call transcripts, job postings, and product reviews. LLMs are less effective with visual data like screenshots or product UI unless the visual content is described in text.
How accurate is LLM competitive analysis?
LLMs produce reliable structured analysis when given specific, well-scoped prompts and real data. They can miss nuance, misinterpret ambiguous language, and occasionally hallucinate details. Every output needs PMM review. Treat LLM analysis as a strong first draft, not a finished product.
This guide is part of Oden's product marketing resource library. Related: AI Competitive Intelligence · AI Buyer Intelligence · AI for Product Marketing · Klue vs. Crayon vs. Oden