Competitor Intelligence skill

Competitor Intelligence

Workflow

1. Competitor Identification

Three tiers:

Discovery methods (classic):

• Search your top 5 keywords — who ranks?
• Ask churned customers who they switched to
• Check G2 / Capterra / TrustRadius / PeerSpot category pages
• Monitor "alternatives to [your product]" and "[product] vs" searches
• Track who bids on your brand keywords (use Google Ads' public Ad Transparency Center and Meta's Ad Library — both legal, no scraping required; do NOT click rivals' paid ads to inflate their costs, that is click fraud)

Discovery methods (2026 surfaces — do not skip these):

• AI answer engines / LLM visibility. Ask ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews: "best [category] tools", "alternatives to X", "X vs Y". Record which competitors get named, what claims the model repeats, and which sources it cites. By 2026 a large and growing share of buyer research starts in an answer engine, so being absent or mischaracterized there is a real competitive gap. Re-check monthly; models and their cited sources drift.
• App / cloud marketplaces. AWS / Azure / GCP marketplaces, Salesforce AppExchange, Atlassian Marketplace, Shopify App Store, HubSpot Marketplace, Slack/Notion/Figma app directories. Listings reveal positioning, pricing tiers, install counts, and review sentiment.
• Browser-extension stores (Chrome Web Store, Firefox Add-ons, Edge Add-ons) when relevant to your category — user counts and reviews are public signal.
• Community & social listening. Reddit, Hacker News, Stack Overflow, niche Slack/Discord/Circle communities, LinkedIn, X/Bluesky, YouTube reviews. Search "[competitor] sucks / migration / switched from / pricing" to surface unfiltered sentiment. Use the platforms' own search/APIs; respect rate limits and ToS.
• Public filings & funding — Crunchbase, PitchBook, SEC EDGAR (for public co's), state business registries.

2. Legal & ethical CI rules (read before collecting anything)

CI is gathering public information ethically. Crossing into deception or theft is a legal and reputational liability — and it poisons your data. Bake these rules into every collection task.

Hard "never" list:

• No pretexting / misrepresentation. Never pose as a customer, journalist, investor, or job applicant to extract info. Don't lie about your employer when signing up for a trial or talking to their staff. (Pretexting violates many laws, e.g. US GLBA, and most ethics codes.)
• No confidential or trade-secret info. Don't solicit it, accept it, or use it — especially from a competitor's current/former employees who are under NDA. Receiving misappropriated trade secrets can create liability under the US Defend Trade Secrets Act and equivalents. If a new hire offers their old employer's confidential docs, decline and document the refusal.
• No credential sharing or unauthorized access. Don't share paid-tool logins to view gated competitor data, don't use someone else's account, don't bypass auth/paywalls. Unauthorized access can implicate the US CFAA / equivalent computer-misuse laws.
• No bid-clicking fraud, no fake reviews, no astroturfing for or against a competitor.

Scraping & platform policy:

• Prefer official APIs and public exports (review-site APIs, Ad Library, marketplace listings) over scraping.
• If scraping public pages, respect robots.txt, the site's Terms of Service, and rate limits; collect only what a normal browser would see; never circumvent login or anti-bot controls. Scraping legality is unsettled and jurisdiction-specific — for any large-scale or commercial scraping, get legal sign-off.
• Privacy. Don't collect personal data on competitor employees beyond public business context (name, title, public posts). GDPR/CCPA-type rules apply to personal data even when scraped from public sites. Never store special-category personal data.

Sourcing discipline:

• Label estimates vs. facts. "~120 employees (LinkedIn headcount, Jun 2026)" not "120 employees". Tag every datum with a confidence level (see source register below).
• Route outbound claims through legal/marketing review. Anything that leaves the building — battlecards used verbatim with prospects, comparison pages, ads — must be fact-checked and (for comparative claims) reviewed. See the battlecard proof standards in §7.
• No coordination with competitors. Never use CI work as a channel to discuss pricing, market allocation, or hiring with a competitor — that is antitrust-sensitive. CI observes the market; it does not coordinate it.

This is operational guidance, not legal advice. Competition, IP, scraping, and privacy law vary by jurisdiction — have counsel review your CI program and any comparative marketing.

3. Source register (the backbone of a real CI program)

Every claim that informs a matrix, battlecard, or strategy decision gets a row here. This is what separates a defensible CI program from rumor, and it powers change-detection (§8) and quarterly refresh.

Rules: primary sources beat secondary; always archive the page (vendors edit silently); downgrade confidence to stale automatically after the review date; one source register per competitor, version-controlled or in your CI tool.

4. Feature Comparison Matrix

Normalize before you compare — "has it" is a trap. Score each cell on a consistent rubric, not a binary:

What to normalize (the dimensions juniors miss):

• Depth, not presence. "Has reporting" is meaningless — compare scheduled reports, custom metrics, export formats, API access to the data.
• Plan packaging. Record the tier and price each feature unlocks at, plus seat minimums, usage caps, and overage pricing — not just the entry price. The real comparison is "feature at the plan a buyer like ours would actually buy."
• Regional availability. Data residency, language/locale support, local payment methods, compliance certs (SOC 2, ISO 27001, HIPAA, FedRAMP) — these win/lose enterprise and EU deals.
• Enterprise-only exceptions. Many "gaps" vanish (or appear) only at Enterprise. Note SSO/SCIM, audit logs, custom contracts, on-prem/VPC, dedicated support, SLAs — these are usually quote-only; tag as estimate if unconfirmed.
• Selection bias. Choose features your ICP actually evaluates, not your longest feature list. A matrix that exists to flatter you misleads your own sales team.

Rules:

• Be accurate. Don't mark competitors as "None" when they have partial support: your reps get burned on the call, and false comparative claims create legal risk (§7).
• Update quarterly minimum — features change fast. Tie each cell to a source-register row (§3) so refresh is mechanical.
• Source every claim (link to their docs/pricing page) and archive it.

5. Positioning & strategy (beyond the 2x2)

2x2 matrix — choose two axes that matter to your buyers:

Common axis pairs:

• Ease of use ↔ Feature depth
• SMB focus ↔ Enterprise focus
• Price ↔ Capability
• Self-serve ↔ High-touch
• Horizontal ↔ Vertical/specialized

How to place competitors:

1. Score each competitor 1-10 on both axes
2. Use customer reviews, demos, and published materials (not assumptions)
3. Identify the white space — where are there no competitors?
4. Position yourself in or near the white space (if it has demand — empty quadrants are often empty for a reason)

The 2x2 is a thinking tool, not the deliverable. Turn the analysis into these strategic outputs:

a) Category narrative. How does each competitor frame the category and the buyer's problem (not just their product)? Track the words they own ("data cloud", "all-in-one", "developer-first"). Decide whether you compete inside their category, reframe it, or create a new one — and what proof you need to make a reframe credible.

b) ICP segmentation. Map which segments each competitor actually wins (by size, vertical, region, technical maturity, buying motion). Competitors are rarely strong everywhere. Output: a per-segment "who wins here and why" table that routes your GTM toward segments where you have a structural edge.

c) Pricing & packaging strategy. From the matrix's plan/price data, reconstruct each rival's packaging logic: what's the wedge (free/low entry vs. land-and-expand), what's gated to force upgrades, what's the value metric (seats, usage, events). Decide your packaging response — match, undercut, bundle, or unbundle — and where to draw tier lines.

d) Threat prioritization. Score competitors on a simple grid: Threat = (overlap with our ICP) x (momentum). Momentum signals: funding, hiring velocity (§8), share-of-voice in answer engines (§1), review volume trend, win-rate movement (§6). Focus deep monitoring on the top-right; demote the rest to quarterly.

e) Response plan & gap triage. For each material threat, pick a response: product (close a real gap — feed into roadmap triage), positioning (reframe so the gap doesn't matter), enablement (arm sales with a battlecard, §7), or ignore (document why). Triage product gaps by reach x deal-impact x effort; don't let CI become a reactive feature-copy machine.

6. Win/Loss Analysis

Interview framework (20-min call with recent wins AND losses):

Aggregate analysis (quarterly, minimum 20 interviews):

• Win rate by competitor: Who do we beat most? Lose to most?
• Top 3 win reasons: What keeps winning deals for us?
• Top 3 loss reasons: What keeps losing them?
• Feature gaps cited: What do prospects wish we had?
• Pricing feedback: Are we perceived as expensive, fair, cheap?

7. Sales Battlecards

Proof standards & legal guardrails (apply before any battlecard ships):

• Every comparative claim needs a source-register row and a defensible proof type: their own published docs/pricing, a dated screenshot, a third-party benchmark, or a verbatim public review. No "everyone knows" claims. False or unsubstantiated comparative advertising is actionable (e.g. US Lanham Act §43(a), EU comparative-advertising rules, and self-regulatory bodies like NAD).
• Truthful, not disparaging. State factual differences ("they gate SSO to Enterprise — sourced") rather than opinion or insult ("their security is a joke"). Avoid claims you can't prove today; competitors fix gaps — re-verify before each enablement cycle.
• Use review quotes correctly. Quote verbatim, attribute (source + date), don't edit to change meaning, and respect the review platform's ToS on reuse. Don't pass selected quotes off as representative if they aren't.
• Approved language. Maintain a list of approved phrasings (legal/marketing-reviewed) and prohibited ones. Reps use the approved wording verbatim in writing.
• Deal-stage usage. Battlecards are internal enablement, not customer handouts. Use Landmines/objection-handling in discovery → evaluation; never send the raw card to a prospect. Comparative content that goes external (web "vs" pages, ads) takes the full legal-review path.
• Examples — acceptable vs. risky:

Enablement review cadence: refresh every battlecard at least quarterly (sooner on a competitor's launch/pricing change detected in §8); each refresh is fact-checked against the source register (§3) and, for comparative claims, re-cleared by legal/marketing before release.

Template (one per competitor):

# Battlecard: [Competitor Name]

## Quick Facts
- Founded: [year] | HQ: [city] | Employees: ~[X] | Funding: $[X]M
- Pricing: [starting price] - [enterprise price]
- Target: [who they sell to]

## They Say (their positioning)
"[Their tagline/main claim]"

## We Say (our counter-positioning)
"[How we differentiate — one sentence]"

## When We Win
- [Scenario 1: specific situation where we're stronger]
- [Scenario 2]
- [Scenario 3]

## When We Lose
- [Scenario 1: specific situation where they're stronger]
- [Scenario 2]

## Landmines (questions to ask prospects to highlight our strengths)
- "How do they handle [area where competitor is weak]?"
- "What happens when you need [feature they lack]?"
- "Have you looked into their [known pain point — pricing, support, etc.]?"

## Objection Handling
| Their claim | Our response |
|-------------|-------------|
| "[Competitor claim 1]" | "[Factual counter with proof]" |
| "[Competitor claim 2]" | "[Factual counter with proof]" |

## Proof Points
- [Customer who switched from them to us + result]
- [Head-to-head benchmark or comparison data]
- [Review quote from G2/Capterra]

8. Monitoring & change detection

What to track, how often, and what to automate — automation turns CI from a quarterly scramble into a stream that feeds the source register (§3):

Job-posting taxonomy (turns hiring into a strategy signal): bucket each req — Eng/Platform (scaling), Eng/<new area> (new product line), Sales/Enterprise (moving upmarket), Sales/<region> (geo expansion), Partnerships/Channel, Compliance/Security (chasing regulated buyers), DevRel (developer motion). A spike in one bucket telegraphs the next move months early.

Change-detection workflow:

1. Detect — automated diff/feed/alert fires.
2. Triage — owner classifies: noise / tactical / strategic. Only tactical+ proceeds.
3. Verify & log — confirm against a primary source, archive the page, write/append a source-register row (§3) with confidence.
4. Assess impact — does it change a matrix cell, a battlecard, pricing, or a roadmap priority?
5. Act — update artifacts; for material moves, trigger response planning (§5e) and battlecard refresh (§7).
6. Broadcast — surface in the monthly digest below.

CRM win/loss tagging (so monitoring connects to revenue): add a required primary_competitor field and a loss_reason / win_reason picklist to opportunities; sync battlecard usage. This makes win-rate-by-competitor (§6) a live dashboard instead of a quarterly survey, and flags competitors whose win-rate against you is moving.

Competitive digest (internal, monthly):

• Top 3 competitive moves this month (with source links)
• Win/loss & win-rate-by-competitor trend (from CRM)
• Feature-matrix and pricing/packaging changes detected
• AI-answer-engine visibility shifts
• Recommended battlecard updates and any response actions opened

Cadence summary: website/changelog/social = weekly; reviews/jobs/marketplaces/AI-engines = monthly; full matrix, positioning, battlecards, and hands-on product review = quarterly; press/funding/M&A = as it happens. Every cadence writes back to the source register so confidence never silently goes stale.