Neobank Positioning Engine Skill

Neobank Positioning Engine

Scrapes a crypto neobank's website and its competitors, extracts positioning claims, maps the competitive messaging landscape, identifies white space, and generates a complete messaging framework with actual copy.

Not a generic "brand strategy" template. Every output is grounded in what competitors actually say on their websites, what users actually complain about in reviews, and what positioning territory is genuinely unclaimed.

Trigger

• position [neobank name]
• positioning audit for [neobank name]
• how should [neobank name] differentiate?
• competitive positioning: [neobank A] vs [neobank B] vs [neobank C]

Workflow

Phase 1: Scrape Positioning Data

python scripts/scrape_positioning.py "[Company Name]" "[website URL]"

Run for the target company and 3-5 competitors. The script:

• Loads the homepage + key pages (about, pricing, features)
• Extracts: headlines, subheadlines, value propositions, CTAs, proof points, feature claims, social proof, brand voice signals
• Saves structured output to output/{slug}-positioning.json

For a full competitive analysis:

python scripts/scrape_positioning.py "KAST" "https://kast.com"
python scripts/scrape_positioning.py "Revolut" "https://revolut.com"
python scripts/scrape_positioning.py "Crypto.com" "https://crypto.com"

If scraping fails or returns thin data, fall back to:

• Web search: "[company]" neobank (mission OR "we help" OR "built for")
• Cached positioning data in references/neobank-messaging-map.md
• App store descriptions (often contain concentrated positioning language)

Phase 2: Extract Positioning Elements

Read scraped JSON and extract these elements for each company:

Use references/positioning-frameworks.md for the analytical lens.

Phase 3: Map Positioning Territories

Plot all companies on these four positioning dimensions:

Dimension 1: Audience Spectrum

Crypto Native ←————————————→ Mainstream Consumer
(speaks DeFi)                (speaks "easy money app")

Dimension 2: Trust Model

Self-Custody ←————————————→ Full Custodial
("your keys")               ("we handle everything")

Dimension 3: Value Proposition Core

Yield/Returns ←————————————→ Utility/Spending
("earn 8% APY")              ("spend crypto anywhere")

Dimension 4: Brand Personality

Technical/Builder ←————————————→ Lifestyle/Consumer
("built on ZK proofs")         ("money, simplified")

For each company, assign a position (1-10) on each dimension. Build a 2x2 matrix using the two most relevant dimensions for this specific competitive set.

Phase 4: White Space Analysis

Identify gaps by asking:

1. Unclaimed territories: Which quadrants of the positioning map have no strong player?
2. Unspoken benefits: What do users want (from review data) that no competitor claims?
3. Credibility gaps: Who claims things without proof points? (Opportunity to own with evidence.)
4. Messaging fatigue: Which claims are so overused they've become invisible? (Avoid these.)
5. Audience orphans: Which user segments are underserved by current positioning?

Consult references/neobank-messaging-map.md for pre-mapped positioning of major players.

Cross-reference with app store review data if available. User language in reviews often reveals positioning opportunities that competitor websites miss.

Phase 5: Generate Messaging Framework

Produce a complete messaging framework:

1. Positioning Statement (Geoffrey Moore format)

For [target customer] who [need/pain],
[Product] is a [category]
that [key benefit].
Unlike [primary alternative],
[product] [key differentiator].

Provide 2-3 variants. Each takes a different strategic angle.

2. One-Liner Options (5-7 words max) Generate 5 options. Rules:

• No buzzwords ("revolutionary," "next-gen," "seamless")
• Must pass the "could a competitor say this?" test. If yes, it's not differentiated.
• Must connect to a real proof point or capability
• Should feel inevitable in hindsight, not clever

3. Value Propositions (3 pillars) For each pillar:

• Headline (under 8 words)
• Supporting sentence (one sentence, specific)
• Proof point (number, case study, or verifiable claim)

4. Messaging by Audience For each identified target segment:

• Their language (how they describe the problem)
• Hook (what gets their attention)
• Proof (what convinces them)
• CTA (what action to push)

5. What NOT to Say List 5-10 specific phrases, claims, or angles to avoid. For each:

• The phrase
• Why it fails (overused / unsubstantiated / wrong audience / competitor-owned)

6. Competitive Response Matrix For each major competitor:

• Their strongest claim
• Their weakest point
• How to position against them without naming them

Phase 6: Render and Deliver

Save the structured brief as JSON to output/{slug}-brief.json, then render as PDF:

python scripts/render_positioning.py "output/{slug}-brief.json"

Rules

1. Evidence over opinion. Every positioning recommendation must trace back to: competitor website copy, user review language, market data, or structural advantage. "I think you should position as X" is worthless. "No competitor in your segment claims X, users ask for it in 23% of reviews, and you have the technical capability" is useful.

2. Specificity over abstraction. "Differentiate on trust" is a waste of time. "Publish your reserve audit monthly, link it from the homepage hero, and use the line 'See exactly where your money is. Updated [date]'" is actionable.

3. The competitor test. For every positioning claim: could a competitor say the exact same thing? If yes, it's not positioning. It's wallpaper. Push harder.

4. Real language only. Pull actual phrases from websites and reviews. Don't invent marketing speak. The best positioning borrows the customer's own words.

5. Omissions matter. What a company doesn't say is often more revealing than what they do. If nobody talks about fees, that's a territory. If nobody shows their team, that's a trust gap to exploit.

6. No AI slop. No "in today's competitive landscape." No "it's worth noting." Direct. Compressed. CMO-ready.

7. Positioning is a bet. Don't try to be everything. The framework should force a choice. "You can own X or Y. Here's the tradeoff. Here's our recommendation and why."

Examples

Input: "Position KAST against Revolut and Crypto.com"

Process:

1. Scrape KAST, Revolut, Crypto.com websites
2. Extract positioning elements for all three
3. Map on dimensions: KAST is crypto-native + spending-focused, Revolut is mainstream + utility, Crypto.com is crypto-native + yield
4. White space: no player owns "transparent crypto spending" (Revolut hides crypto fees, Crypto.com leads with earn/yield, KAST could own spending transparency)
5. Generate framework: "The crypto card that shows you everything." Three pillars: transparent fees, multi-chain spending, no lock-ups.
6. Render PDF, deliver

Input: "How should a new crypto neobank targeting LatAm differentiate?"

Process:

1. Scrape top 5 LatAm-serving neobanks (Nubank, Mercado Pago, Ualá, dLocal, Lemon Cash)
2. Extract positioning. All cluster around "financial inclusion" and "easy access"
3. White space: nobody owns "USD stability" messaging specifically (they all talk features, not the core anxiety: peso depreciation)
4. Framework: Position around the emotional need (protecting purchasing power) not the feature (crypto/USD access)
5. Generate copy that speaks LatAm financial anxiety, not crypto jargon

Edge Cases

• Pre-launch company (no website yet): Use pitch deck, app store description, social media bios. Positioning analysis becomes purely competitive: "here's the landscape, here's the gap, here's where to plant your flag."
• Very crowded segment (10+ competitors): Don't map all of them. Pick the 3-4 that compete for the same user. Map the rest as "background noise" with one-line positioning summaries.
• Company with no clear positioning: Flag it: "Your current messaging is [list of generic claims]. None pass the competitor test. You have no positioning. Here are three options."
• B2B crypto infrastructure (not consumer neobank): Adjust dimensions. Replace Audience Spectrum with "Developer-focused vs Business-focused." Replace Trust Model with "Self-hosted vs Managed."

Files