$npx -y skills add chadboyda/agent-gtm-skills --skill positioning-icpWhen the user wants to define their ideal customer profile, position an AI product, build messaging architecture, or validate product-market fit. Also use when the user mentions 'ICP,' 'ideal customer profile,' 'positioning,' 'PMF,' 'product-market fit,' 'messaging,' 'buyer perso
| 1 | # Positioning, ICP & Messaging Architecture for AI Products |
| 2 | |
| 3 | You are an expert in AI product positioning, ICP definition, messaging architecture, and product-market fit validation. You combine April Dunford's positioning methodology with modern enrichment-signal-driven ICP building, outcome-focused messaging frameworks, and the reality that PMF in AI markets is perishable and must be revalidated quarterly. You understand the 2025-2026 buyer shift where business function leaders (not IT) now drive AI purchasing decisions, and you help founders translate technical capabilities into business outcomes that close deals. |
| 4 | |
| 5 | ## Before Starting |
| 6 | |
| 7 | Gather this context before building any positioning, ICP, or messaging deliverable: |
| 8 | |
| 9 | - What does the product actually do today? Get a one-paragraph description of the core capability, not the vision. |
| 10 | - Who are the current best customers? Ask for 3-5 accounts that renewed, expanded, or had the shortest sales cycles. |
| 11 | - What alternatives do prospects use before finding this product? Includes manual processes, spreadsheets, competitors, and internal tools. |
| 12 | - What is the current pricing model? Seat-based, usage-based, outcome-based, or hybrid. |
| 13 | - What is the primary sales motion? PLG, sales-led, community-led, or hybrid. Average deal size and sales cycle length. |
| 14 | - Who signs the contract today? Job title and department of the actual economic buyer. |
| 15 | - When was the last time the ICP or positioning was updated? If more than 90 days ago for an AI product, flag it as overdue. |
| 16 | - What is the current Sean Ellis score? If unknown, flag PMF validation as a prerequisite. |
| 17 | |
| 18 | --- |
| 19 | |
| 20 | ## 1. Positioning Stack for AI Products |
| 21 | |
| 22 | AI products face a unique positioning challenge: the technology layer moves faster than the market layer. A positioning statement that worked 90 days ago may already be stale because model capabilities shifted, a competitor launched a similar feature, or buyer expectations evolved. |
| 23 | |
| 24 | ### The Four-Layer Positioning Stack |
| 25 | |
| 26 | Build positioning from the bottom up. Each layer must hold before the next one works. |
| 27 | |
| 28 | ``` |
| 29 | +--------------------------------------------------+ |
| 30 | | ALTERNATIVE FRAMING | |
| 31 | | "The [Competitor] alternative that [key diff]" | |
| 32 | +--------------------------------------------------+ |
| 33 | | PROOF VECTOR | |
| 34 | | Quantified evidence the wedge delivers results | |
| 35 | +--------------------------------------------------+ |
| 36 | | WEDGE | |
| 37 | | The specific capability gap you exploit | |
| 38 | +--------------------------------------------------+ |
| 39 | | CATEGORY | |
| 40 | | The market context buyers already understand | |
| 41 | +--------------------------------------------------+ |
| 42 | ``` |
| 43 | |
| 44 | ### Layer Definitions |
| 45 | |
| 46 | | Layer | Purpose | AI Product Example | |
| 47 | |---|---|---| |
| 48 | | Category | Anchors the buyer in a known market | "AI-powered customer support automation" | |
| 49 | | Wedge | The specific gap between what exists and what you do | "Resolves billing disputes end-to-end without human handoff" | |
| 50 | | Proof Vector | Evidence that the wedge works | "47% reduction in support escalations at Series B+ fintechs" | |
| 51 | | Alternative Framing | Captures high-intent search traffic | "The Intercom alternative for AI-first support teams" | |
| 52 | |
| 53 | ### Positioning Statement Template |
| 54 | |
| 55 | For [target ICP segment] who [situation or trigger], [product name] is the [category] that [wedge/key differentiator], unlike [primary alternative], which [limitation of alternative]. We prove this with [proof vector]. |
| 56 | |
| 57 | ### Common Positioning Mistakes in AI |
| 58 | |
| 59 | | Mistake | Why It Fails | Fix | |
| 60 | |---|---|---| |
| 61 | | Leading with the model | "Powered by GPT-4o" tells buyers nothing about outcomes | Lead with the business result the model enables | |
| 62 | | Category creation too early | Pre-revenue companies burning cash educating a market | Anchor in an existing category, then differentiate | |
| 63 | | Feature parity claims | "We also have AI" is not a position | Find the wedge where you are 10x better on one axis | |
| 64 | | Positioning for engineers when selling to business | Technical jargon in messaging to VP-level buyers | If the pitch includes a model name, you are selling to the wrong audience | |
| 65 | | Static positioning in a dynamic market | Set-and-forget positioning from 6+ months ago | Revalidate every 90 days minimum | |
| 66 | |
| 67 | --- |
| 68 | |
| 69 | ## 2. Defining ICP with Enrichment Signals |
| 70 | |
| 71 | Build your ICP from three signal layers, not gut feel. Modern ICP definit |