Brand visibility

15 min

You are an expert in Generative Engine Optimization (GEO) and LLM visibility strategy. Your task is to generate a structured set of high-quality prompts for monitoring in Adobe LLM Optimizer. These prompts must reflect how real users naturally query AI assistants (ChatGPT, Gemini, Perplexity, Copilot) across the full customer journey.

Input Parameters

Brand: <Brand_Name>

Website: <Website_URL>

Product/Service Focus: <Product_or_Service>

Industry: <Industry>

Language: <Language>

Region: (ISO 3166-1 alpha-2 code, e.g. DE, US, BR, FR) <Region>

Optional — if not provided, infer from Brand, Industry, and Region:

Top 3 Competitors: <Competitor_1>, <Competitor_2>, <Competitor_3>

Market Scope: <Market_Scope> (Global / National / Regional / Local)

Business Size: <Business_Size> (Enterprise / Mid-Market / SMB)

Audience Type: <Audience_Type> (B2B / B2C / Both)

Key USP: <Key_USP> (1–2 sentences: what makes <Brand_Name> genuinely different)

Niche Keywords: <Niche_Keywords> (industry-specific terms, local geography, regulatory context, product jargon)

Language Rule

ALL generated prompts must be written in <Language>. No exceptions — not for comparison queries, not for branded queries, not for technical queries.

Specificity Calibration

Adjust prompt specificity based on Market Scope and Business Size:

Global + Enterprise: broader category queries, international comparisons, industry-standard terminology

National + Mid-Market: country-specific regulations, local competitors, national market context

Regional / Local + SMB: city/region names, local language nuances, community-level concerns, highly specific niche terminology

If Niche Keywords are provided, weave them into prompts naturally wherever a real user in this industry would use them. If not provided, infer relevant terminology from Brand, Industry, and Region.

B2B vs. B2C Prompt Style

If Audience Type is B2B, include these query types in Consideration and Purchase stages:

ROI and TCO queries ("What is the total cost of ownership for…")

Multi-stakeholder queries ("How do I convince my CFO to invest in…")

Compliance and integration queries ("Does <Brand_Name> integrate with [relevant system]?")

SLA and support queries ("What SLA does <Brand_Name> offer for enterprise customers?")

If Audience Type is B2C, include:

Value-for-money queries ("Is <Brand_Name> worth the price compared to…")

Experience-driven queries ("What do real users say about <Brand_Name>?")

Simplicity queries ("Is <Brand_Name> easy to use for someone without technical knowledge?")

Emotional fit queries ("Which [product] feels best for everyday use?")

If Audience Type is not provided, infer from Industry and Product/Service Focus.

USP Integration

If <Key_USP> is provided, weave it naturally into relevant Fight and Features prompts where it fits organically. Do not force it, do not cluster it — the goal is broad coverage, not USP repetition.

Framework

Generate prompts using a matrix of two dimensions:

Dimension 1: The Five F's (Content Types)

Facts & Figures — Objective, structured product data (specs, pricing, dimensions, performance metrics). The foundation LLMs need for accurate retrieval.

Features — Differentiating capabilities and innovations. Not what it IS, but what it DOES differently than alternatives.

FAQs — Real user questions around product usage, compatibility, setup, limitations. High-volume hygiene content that maps directly to LLM queries.

Flock — Trend-driven, category-level hub content (top lists, seasonal themes, use case roundups, buying guides). Captures broad interest and recurring engagement.

Fight — Explicit comparison and competitive content. Brand vs brand, category vs category, direct head-to-head evaluations.

Dimension 2: Customer Journey Funnel

Awareness (15% of total prompts): User is exploring a problem or category. No brand preference yet.

Consideration (45% of total prompts): User is actively comparing options, evaluating fit, handling objections. THIS IS THE PRIORITY STAGE — AI assistants consolidate choices here and users increasingly delegate comparison and shortlisting to LLMs.

Purchase (25% of total prompts): User is ready to act — pricing, availability, deals, setup, onboarding.

Retention (15% of total prompts): User already owns the product — usage optimization, troubleshooting, upgrades, loyalty.

Output Rules

Generate exactly 100 prompts.

Distribute across the 5F x Funnel matrix with the weightings above. Not every F applies to every funnel stage — use judgment. Expected heavy cells:

Consideration x Features (highest volume)

Consideration x Fight (second highest)

Consideration x FAQs

Purchase x Facts & Figures

Awareness x Flock

Retention x FAQs

Ensure 20–30% of prompts are branded (explicitly mention <Brand_Name>). The rest are unbranded category/intent queries.

Write prompts as natural conversational questions — exactly how a real person would talk to an AI assistant. No keyword-style queries.

For Consideration prompts specifically:

Include direct comparison queries ("X vs Y for [use case]")

Include shortlisting queries ("Best 3 options for [specific need]")

Include objection-handling queries ("Is X worth it given [concern]?")

Include fit/suitability queries ("Which [product] is best for [specific persona/context]?")

For B2B: include multi-stakeholder, ROI, and integration queries

Use the named competitors (or inferred competitors) in Fight and Consideration prompts.

Use Niche Keywords naturally throughout prompts — especially for Regional/Local and SMB inputs.

Topics should usually contain 5-10 Prompts. This Rule has exceptions based on topic.

Create 2-3 Categories

Quality Check Before Output

Verify ALL of the following before producing output. If any criterion fails, revise and re-check:

~45% of prompts are Consideration stage

~25% are branded

All 5 F's are represented

Competitors appear in Fight and Consideration prompts

No keyword-style queries — all are natural language questions in <Language>

Specificity level matches Market Scope and Business Size (inferred if not provided)

B2B or B2C query types are present where applicable

Exactly 100 prompts total

Amount of topics must be under 20% of prompts generated

Output — Two Files

File 1: Excel (for review)

File 2: CSV (for tool import)

Produce a clean CSV with exactly 4 columns and no additional content — no header explanation, no markdown, no commentary before or after: category,topic,prompt,region

Rules for the CSV:

Region must be the ISO 3166-1 alpha-2 code from (e.g. DE, US, BR) <Region>

No BOM, no trailing spaces, no blank rows

Download as txt

ABV Prompt set generator 3.txt