# Asb Carol Keystones > Facilitates the third step of a proven ideal-customer (ICP) method: refining classified strengths into keystones — the specific characteristics, behaviors, or circumstances that make a customer NEED an extreme version of a strength, badly enough to drive the purchase alone. Takes a strengths chart (S1, S2, … — file or pasted), walks the strengths one at a time asking who requires an extreme version of each, gates every candidate on naming a real market segment that typifies it (no segment = table stakes, cut), and records survivors in KEYSTONES.md (K1, K2, … with [S] references and example segments). Delivers the strategic verdict when few or none survive: the product isn't compelling yet. Load when the user has classified strengths and asks who needs them, who their target market is, or 'turn our strengths into keystones.' Do NOT load to classify strengths and weaknesses (the previous step), to derive deal-breakers or the anti-market (the next step), or to write the final ideal-customer definition (later). - Skill: `asmartbear/asb-carol-keystones` (Agent Skill) - Install (CLI): `npx skillmds@latest add asmartbear/asb-carol-keystones` - Raw SKILL.md: https://api.skillmd.com/api/skills/asmartbear/asb-carol-keystones/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: asmartbear (https://skillmd.com/u/asmartbear) - Updated: 2026-09-10 - Page: https://skillmd.com/skills/asmartbear/asb-carol-keystones --- # Keystones: Who Needs an Extreme Version of What You're Best At Passionate purchase decisions don't come from sort-of liking a few features. They come from an overwhelming pull toward one of your strengths — something specific about the customer's circumstances that makes them think "I've got to make this work for us." That pull is a **keystone**, and mapping your keystones is how a strengths chart becomes a target market. This skill walks the strengths one at a time, generates keystone candidates with the user, and holds each to the gate that separates keystones from wishful thinking: a real market segment must typify it. ## The mental model ### What a keystone is A keystone is the critical characteristic that drives a customer's desire for the product — the thing that grabs their attention even in a noisy world, even while other trade-offs are in flux. It is not the strength itself; it is the *customer circumstance* that makes the strength critical. "Fast website" is a strength. "Has a website where even 0.1 seconds faster directly results in more revenue" is a keystone — and it immediately names its market: e-commerce stores, where studies put conversion of fast sites at multiples of slow ones, and where the giants have measured revenue loss per tenth of a second of load time. Same pattern across a real keystone chart: "scalable" becomes *has traffic spikes at unpredictable times* (news sites whose story can break any moment); "secure" becomes *just got hacked and is still terrified* or *has a legal mandate for vendor certifications* (the regulated industries); "inexpensive to build on" becomes *big plans, limited budget* (every small business, every "do more with less" marketing department). ### The generative question For each strength, ask: **what characteristics, behaviors, or circumstances REQUIRE an extreme version of this strength?** Imagine the customer for whom this strength is absolutely critical to their strategy — or so delightful it pulls them in through pure joy. Require is the operative word: a circumstance that merely *prefers* the strength generates a trade-off shopper — a buyer who weighs your strength against a competitor's and picks whoever wins the spreadsheet — not a keystone that drives a sale by itself. (Trade-off shoppers are a fine, large audience; the method's point is that targeting the keystone customer is what wins them too.) ### The gate: no segment, no keystone After naming a keystone candidate, describe real-world examples of companies or people who embody it. If you cannot name a significant market segment that typifies the keystone, it isn't a keystone — the strength is probably table stakes: true, useful, not compelling enough to drive a purchase alone. That's fine — that's why it's called brainstorming — but the candidate is cut, on the record, not recorded hopefully. The segments named should be described by circumstance and behavior ("e-commerce stores where speed measurably moves revenue"), not raw demographics; the sharper the description, the more useful every downstream step becomes. ### Keystones concentrate in the head, not everywhere You won't find keystones by asking what your best customers have in common — most of that is common to *all* your customers, including the worst. A trait that describes everyone predicts nothing about buying, paying, or staying. The right keystones concentrate in the best customers and show up weakly, or not at all, in the bad ones. If the strengths chart carries a best-vs-worst differential, mine it: the differences are where keystones live. ### The one-keystone verdict If the walk ends with one keystone — or none — that is not a failed exercise; it is the most valuable strategic lesson available: **the product isn't compelling yet.** Say it plainly, because learning it now is a multiplier — fixing it increases sales, and it was hidden. The twofold response, from the source: (a) increase marketing spend in the markets where keystones DO exist (with zero keystones, this half is empty and the verdict says so), and (b) use market and customer research to determine what to build next that would address a real keystone. Don't pad the chart to avoid the verdict — and don't pick what to build yourself: a feature chosen from inside the strengths chart is an invented keystone, exactly the wishful thinking the gate kills. With one survivor, phrase it precisely: compelling to that one segment only. ### Why this is worth doing at all The customers keystones find are the head of the profit distribution — in the canonical case, roughly 5% of customers producing 40% of profit, which justifies spending many times more to acquire each one. The keystone chart is the map to exactly those customers. ### Vocabulary - **Keystone (K1, K2, …)** — a customer characteristic, behavior, or circumstance that requires an extreme version of a strength; cites its [S-number]. - **Example segments** — the real-world companies or people who typify the keystone; the gate every keystone must pass. - **Table stakes** — a true, useful strength that compels no one by itself; recorded as a cut, not a keystone. - **Requires vs. prefers** — the bar: a keystone circumstance makes the strength critical, not merely nice. ## The refiner's posture ### Be clear, not clever Write to be understood, not admired. The work here wrestles with hard concepts, and clever metaphors, wordplay, or cute turns of phrase make them harder to grasp, not easier. Say plainly what you mean. If a sentence reads more clearly without a flourish, cut the flourish. State the actual point rather than gesturing wittily at it. ### Restate references; never cite a bare token When you mention a numbered or lettered item to the user — K4, W2, O17, H3, and the like — add a few plain words on what it actually is ("K4 — the owner whose career rides on the site"). A bare token is unreadable to a human who saw it defined hours or days ago: the tag is for traceability, the gloss is for comprehension. Keep the tag for accuracy; always add the gloss. ### Generate freely; gate ruthlessly Keystone-finding is brainstorming — propose candidates yourself, liberally, alongside the user's: the imaginative leap from "strength" to "whose circumstance requires this" is exactly where an outside perspective helps. But every candidate, yours or theirs, faces the same two tests before recording: does the circumstance *require* the extreme version (not merely benefit from it), and can a significant real segment be named that typifies it? Run both visibly. A predefined way to press harder: if a devil's-advocate interrogation skill is installed in the environment (for example *Rude Q&A* / `asb-rude-qa`, from the same author as this method), invoke it against the draft keystone chart with this brief: *attack these keystones — find every one whose segment is vague, tiny, or invented; every one where the circumstance merely prefers the strength rather than requiring it; every segment described so broadly it describes everyone. Don't accept wishful or hand-wavy defenses.* If no such skill is available, run that interrogation yourself, visibly — the per-candidate tests inline, the batch attack at the closing sweep. ### Press segment descriptions into specifics "Small businesses" is not a segment description; "e-commerce stores processing enough revenue that a measurable conversion lift pays for premium hosting" is. When the user names a vague segment, ask for the example companies — actual names, or a description sharp enough that a salesperson could qualify against it. Generic words (many, serious, growing, quality-focused) get pressed into numbers, behaviors, and circumstances. A keystone whose segment can't be made specific after pressing fails the gate. ### The user owns the market knowledge Whether a named segment is significant, whether those customers really exist, whether the circumstance actually occurs — the user knows their market and their call stands, after one honest press ("you said the head of your customer base is 3+-location groups; this keystone describes single-location shops — who's it for?"). What their call cannot do is skip the gate: a keystone with no segment attached doesn't enter the file no matter whose idea it was. ### One strength per exchange Walk the strengths in order, one per exchange; a strength may yield zero, one, or several keystones, and candidate variants within one strength are one exchange's work. The opening move is small: what was read, which strengths look most keystone-rich, then the first strength. Compress ceremony on request, never structure. ## How to use this skill ### Phase A — Ingest Read the strengths chart (default: `STRENGTHS-WEAKNESSES.md` in the current directory; or pasted). Both-classified attributes participate on their strength reading — the weakness half belongs to the next step. Note the best-vs-worst differential if the chart (or its observations file, if present) carries one; it seeds keystone candidates. If no strengths chart exists, don't refine from nothing: offer the on-ramp — run the classification step first (if a classification skill from this method's author is installed, for example *Strengths & Weaknesses* / `asb-carol-strengths`, name it), or capture a defensible strengths list in chat now with the caveat that unvetted strengths yield unvetted keystones. For the quick capture: hold each claimed strength to a mini genericness test (would any competitor claim the same words? then press for the mechanism or the measured gap behind it), assign S-numbers as they settle, grab the worst-customer picture too (the closing concentration check needs it), and create the KEYSTONES.md immediately — captured strengths recorded in its preamble, caveat included — so a died session loses nothing and a resume finds the strengths on disk. Note in Next steps that no weaknesses were captured, so the deal-breakers step must capture or classify them first. If a KEYSTONES.md already exists at the target location, read it: in-progress header means resume — pick up at the strength the header names, don't re-litigate settled keystones. Complete means ask whether to revise. Output: `KEYSTONES.md` in the same directory as the input file. If the input was pasted and no path is known, ask where the method's files should live before creating anything (default: the current directory) — never scatter files silently. ### Phase B — Walk the strengths Open small: what you read, which strengths look keystone-rich, any best-vs-worst differential worth mining — then the first strength. Per strength: 1. **Ask the generative question** — who requires an extreme version of this? Offer your own candidates alongside the user's; the best-vs-worst differences are prime material. 2. **Test each candidate**: requires-not-prefers, then the segment gate — name the real-world examples. Press vague segments into specifics. 3. **Record survivors** with [S-number] references and their example segments. A *strength* that yields no keystone is recorded in the file's table-stakes section with one line of reasoning; a failed *candidate* under a strength that did yield keystones dies in chat with its reasoning — no file entry needed. 4. Move to the next strength. **Record to the file as you go.** Create `KEYSTONES.md` at the first settled keystone; append after each; keep the header pointer current (strengths walked, current strength). The file is the memory, not the chat. If files aren't accessible, re-emit the full draft in a fenced block every strength or two. ### Phase C — Sweep and close - **Concentration check** — for each keystone: would this show up weakly or not at all in your *worst* customers? A keystone that describes everyone is re-tested and usually dies; one that's partially present in the worst survives with a "prioritize for honing" flag — the deal-breakers step tightens it. - **Segment reality check** — the batch press (delegated or self-run, per the posture): segments real, significant, specific. - **Count the survivors.** Two or more solid keystones: proceed. One or zero: deliver the verdict plainly — the product isn't compelling yet — with the twofold response recorded in the file (spend where keystones exist; research what to build that addresses a real keystone). This is a finding to act on, not a failure to soften. Finalize: remove the in-progress header and close with the handoff. The next step mirrors this one on the weaknesses — deal-breakers and the anti-market — and then *hones these keystone segments* by appending the qualifiers the deal-breakers force; if a deal-breakers skill from this method's author is installed (for example *Deal-Breakers* / `asb-carol-dealbreakers`), name it: "when you're ready, run `asb-carol-dealbreakers` — it works on the weaknesses and then sharpens this file's segment descriptions." ### The file structure ```markdown # Keystones — > ⚠️ IN PROGRESS — the walk is not complete. Strengths walked: S-numbers>; currently on: . If you are resuming, continue > there. (This note is removed at finalization.) ## Keystones **K1.** [S1] Example segments: **K2.** <…> [S3] Example segments: <…> ## Table stakes (true, but compels no one — cut) - — no significant segment requires an extreme version: . [S2] ## Verdict (only when one or zero keystones survive) , or to no one yet — plus the twofold response: (a) spend where the keystone exists, (b) research what to build that would address a real keystone.> ## Next steps ``` K-numbers are stable once written — later steps cite them and the deal-breakers step edits this file's segment lines under them. Never renumber. ## Refusal conditions - **No strengths chart.** Keystones derived from unvetted strengths inherit their vagueness. Offer the classification step or a quick defensible capture first, caveated. - **A keystone with no segment, however insisted.** The gate is the method: no significant real-world segment typifies it = table stakes, recorded as a cut. The refusal is of recording it as a keystone, never of the strength — which stays true and useful, just not compelling alone. - **Vague segments.** "SMBs," "growing companies," "people who care about quality" — pressed into circumstance and behavior until a salesperson could qualify against the description, or the keystone fails the gate. - **Padding to avoid the verdict.** When one or zero keystones survive, that verdict is the deliverable. Inventing weak keystones to make the chart look healthy defeats the exercise and poisons every downstream step. - **Picking the target segment.** Which keystone market to pursue is strategy, made later with deal-breakers, inciting events, and the full definition in hand. This file maps the options; it doesn't choose. - **Deal-breakers or the anti-market.** The weakness side is the next step's job — including the honing of this file's segments.