Scribe known-for queries and free/paid ping packaging

Owner words drive a small frozen query set; free capped ping + heuristic.

docs/strategy/BUSINESS_PLAN.md

@@ -88,7 +88,14 @@ POR work: an intuitive scorecard a business owner understands. Top-level outcome
 
 Keep to **three top-line outcomes**, each with plain “why not yet” underneath — not a 40-row SEO audit. Dual scoreboard still applies: hygiene work feeds rungs 1–2; recommendability is the differentiation line.
 
-**Open — measuring “would recommend”:** early v1 may need to be simulated / heuristic (“based on what’s public, an assistant still couldn’t answer X”) rather than live pings of ChatGPT. Honesty bar TBD with Ty.
+**Open — measuring “would recommend” (working packaging idea, not locked):**
+
+- **Owner input:** business owner describes the business in their own words — what they want to be “known for.” GeoLocal derives the query set from that + place + category; owner does not invent keywords.
+- **Routine queries:** generate from that input, then freeze a small static set (roughly 5–10), re-run routinely — not a giant library.
+- **Free tier (idea):** one real live ping (e.g. ChatGPT), capped (e.g. one run per business), plus a free heuristic so the scorecard isn’t blank if the ping fails.
+- **Paid (idea):** same questions across a versioned panel of popular AI assistants (names will drift — treat panel as versioned, not sacred).
+
+Honesty / cost / abuse controls still TBD.
 
 ### Free path after checkup (working direction — not locked)
 
@@ -140,13 +147,14 @@ If something here conflicts with `CANONICAL_STRATEGY.md`, resolve on purpose in
 | 2026-09-15 | Process: nothing locked yet — distill POR as we build; porbot scribes so key insights are not lost |
 | 2026-09-15 | Working nuance: 2017 hygiene usually first *fix* work; dual scoreboard from day one (hygiene + AI recommendability); skip forced hygiene phase if foundation already solid |
 | 2026-09-15 | Working direction: owner scorecard ladder — knows you exist / knows what you do / would recommend you (wording TBD); open whether v1 recommend = heuristic vs live assistant pings |
+| 2026-09-15 | Working idea: owner “known for” → small frozen query set; free = capped single live ping + heuristic; paid = versioned multi-assistant panel |
 
 ---
 
 ## Next conversation topics
 
 1. Wordsmith the three scorecard rungs; define plain “why not yet” evidence under each
-2. Decide v1 “would recommend” measurement: heuristic / simulated vs live assistant pings
+2. Decide free/paid assistant-ping packaging (caps, panel list, heuristic fallback) and cost/abuse controls
 3. Primary reader for v1 of this plan (internal / investors / partners)
 4. Offer / packaging detail (what free record includes vs what paid unlocks beyond fulfillment) — no invented numbers
 5. 90-day proof: what “reputable enough that AI checks us” requires in coverage and quality