---
name: cosmos-luminosity-market-validation
description: Use when Margarita asks to extract an audience, offer, content, sales, or launch algorithm for Cosmos Luminosity from supplied courses/transcripts, research an English-speaking audience, validate a product before building a site, or grow a target audience through source-grounded content experiments.
---

# Cosmos Luminosity Market Validation

## Funnel-stage separation: awareness result vs paid trial

Do not repeat the same interaction mechanic at adjacent funnel stages. If awareness already contains a short interpretive survey/check-in, the paid trial must not be another questionnaire, assessment, or prettier report.

For the current CL professional-transition route:

- **Awareness/free entry:** short survey → transparent possible interpretation.
- **Paid trial:** a result-matched AudioBoost → one bounded action in real life → optional Evidence Review.
- **Deeper paid product:** broader transition architecture and multiple experiments.

A result-matched AudioBoost is selected from answer patterns, but it is **not personalized audio, diagnosis, prediction, or proof that the system fully knows the woman**. Personalize only the bridge text, selection of the relevant authored journey, and bounded next action.

Whenever `AudioBoost` first appears publicly, define it plainly before using brand language: an original Cosmos Luminosity audio journey using voice, imagery, guided questions, pauses, attention, body and state to help the listener meet an insight experientially and carry it into a real-life action. Do not assume a new visitor knows whether it is a podcast, meditation, affirmation track, or therapy.

Before audio exists, a temporary result card may honestly preview the matched journey and proposed price, but must show a non-interactive availability status—not a disabled CTA that resembles a broken purchase button. Do not promise playback, checkout, duration, or access terms until they exist and have been verified.

## Trial-product depth and artifact validation

> For the approved separation between the free Transition Insight and a result-routed paid AudioBoost, including routing boundaries, commercial status, human-language rules and multipart voice-approval workflow, use [`references/trial-to-routed-audioboost-architecture.md`](references/trial-to-routed-audioboost-architecture.md).

A completed interactive flow is not evidence of product value. For Cosmos Luminosity trial products, validate both the **decision work** and the **save-worthiness of the result** before treating the prototype as viable.

- Do not lead with a generic quiz, score, mini worksheet, or small “experiment card.” For accomplished women in professional transition, these can feel infantilizing even when visually polished.
- Depth should come from consequential distinctions rather than more questions: Preserve / Release / Redesign; hidden functions of current work; tensions among mind, body, state and material reality; credible continuations; key assumptions; reversibility; boundaries; evidence; continue/stop criteria; and Evidence Review.
- The result should be a usable adult artifact—such as a **Transition Decision & Experiment Brief**—not a formatted transcript of inputs. It should help the woman think, act, have a conversation, and later review evidence.
- The free experience should still complete in one session. Do not mistake “deeper” for a mandatory multi-day commitment, email gate, long course, or compulsory return.
- Preserve authorship: no assigned type, readiness score, prediction, or external verdict.
- Validate with behavioral questions: Would she save, print, reopen, take into a conversation, or use the artifact during an experiment? “Looks beautiful” is insufficient evidence.
- Keep the paid boundary explicit: the free trial may create one substantial brief and one experiment; the paid Gateway owns the fuller transition map, deeper scenarios, reversibility assessment, supported testing and Evidence Review.

See `references/trial-depth-and-save-worthy-artifacts.md` for the detailed product and validation checklist.

## Guided-investigation standard

When a trial still feels like a polished worksheet, do not solve the problem by adding more direct reflection prompts. The product must help the woman **discover a distinction she could not easily formulate before entering**.

- Do not ask her to self-author every conclusion (`What should you preserve?`, `What should you release?`, `What is your hypothesis?`) and then return a formatted transcript. That is input capture, not product intelligence.
- Begin with **concrete episodes**, then use contrasts and counterfactuals: what happened; what was hardest; what would change under better conditions; what would remain unacceptable; when the work still felt alive; what people seek from her without title or industry.
- Investigate the **hidden contract** with current work: identity, belonging, status, structure, visibility, intellectual intensity, proof of worth, sunk investment, and protection from uncertainty. Ask which functions must be preserved but not necessarily through the current role.
- Compare both costs: the risk of movement and the accumulating cost of no change. Include reversibility and present capacity so an experiment remains proportionate.
- Synthesis must be transparent and falsifiable. Present: **possible reading → evidence from her answers → contradictory evidence → key unknown → bounded test**. Never present a diagnosis, type, score, prophecy, or authoritative verdict.
- Support materially different branches rather than one predetermined answer. At minimum distinguish: `redesign before exit`, `direction may be complete`, and `mixed transition / compare two conditions`.
- Generate the experiment from the **unproven assumption**, not merely from a problem category. Include boundary, time window, observable evidence, continue rule, stop/redesign rule, and Evidence Review question.
- The final artifact should synthesize rather than mirror inputs. A useful structure is: possible reading, grounds, contradiction, hidden contract, cost of no change, Transition Square (`Carry Forward / Release / Redesign / Discover`), two credible continuations, key assumption, and experiment protocol.
- Preserve editability: generated hypotheses and experiments are proposals the woman can revise or reject.

For the detailed question architecture, rule-based synthesis pattern, artifact structure, and QA matrix, use `references/guided-transition-investigation.md`.

## Cold-entry trial: interpretive depth without reflection burden

For a free first-touch experience, do not confuse depth with the amount of writing required from the woman. A 20–30 minute guided investigation may belong inside a paid Gateway, but it is usually too much commitment for cold awareness traffic.

- Target roughly **4–6 minutes**, six short screens, mostly constrained choices, and at most one optional short text response.
- Ask for raw signals, not conclusions: presenting question, moments of resistance, counterfactual response under better conditions, living professional actions, feared loss, protective strategy, and divergence among mind/body/actual behavior.
- The product—not the user—must do the interpretive work. Never ask her to name the hidden contract, formulate the reframe, or author the final hypothesis before receiving value.
- Interpret from **relationships among answers**. The result should reveal one primary pattern, an alternative reading, three specific grounds, what cannot yet be concluded, and one discriminating experiment.
- Maintain several materially different patterns so the experience does not force everyone toward redesign, exit, or experimentation. Useful classes include: outgrowing the price of participation; mastery alive while identity/context ends; ambition changing form; analysis as protection from beginner status; exit fantasy compensating for missing boundaries; depletion obscuring discernment; and mixed transition.
- Phrase interpretation as `possibly / may / your answers point toward`, never as diagnosis. Keep a counter-reading and epistemic boundary visible.
- The final sheet should be one concise, save-worthy insight document—not a multi-page compilation of every input.
- Test the matrix with deliberately opposing answer combinations and verify that they produce meaningfully different interpretations.

See `references/short-interpretive-transition-trial.md` for the compact question architecture, interpretation contract, pattern matrix, and acceptance tests.

> For public-social Sales Safari evidence standards, incremental collection, source verification grades, bias controls, and translation into the CL platform journey, use [`references/social-sales-safari-evidence-and-platform-translation.md`](references/social-sales-safari-evidence-and-platform-translation.md).
>
> For the 2026 synthesis of affluent English-speaking women’s therapy, coaching and spiritual/esoteric demand, competitor ladders, and the founder-free pre-LumoSide route, use [`references/affluent-women-demand-and-product-ladder-2026.md`](references/affluent-women-demand-and-product-ladder-2026.md).
>
> For the canonical audience definition, `Grounded Transition Architecture` category, three-information-layer model, positioning copy, product value sequence, messaging pillars, and fit boundaries, use [`references/canonical-audience-and-grounded-transition-positioning.md`](references/canonical-audience-and-grounded-transition-positioning.md).

## Core rule: keep evidence layers separate

Before synthesizing, divide claims into:

1. **Named source** — explicitly stated in the supplied audio, video, transcript, workbook, or expert material.
2. **Margarita / Cosmos Luminosity** — her stated desires, homework, existing products, and approved philosophy.
3. **Agent adaptation** — proposed segmentation, offer, experiment, metric, funnel, customer route, automation, or platform decision.

If Margarita asks for “только то, что было в материалах,” return only layer 1 plus the explicitly requested parts of layer 2. Never silently turn extraction into a strategic blueprint. Put adaptations in a separate labelled section only when requested.

Do not introduce Magic Cloud, LumoSide, subscriptions, MVPs, automation, metrics, or a technical platform merely because they fit the project. Include them only when the named source set contains them or Margarita asks for adaptation.

For full procedures, Katya Ro Reels source notes, and the public Instagram topic-page workflow for finding and verifying direct Reel links and visible view counts, read `references/source-grounded-audience-and-content-validation.md`.

For the October 2026 English audience evidence bank, segment decision, verified VOC, Reels viewing territories, CL meaning architecture, brand-integration ladder, and long-run Sales Safari persistence pattern, read `references/english-audience-and-reels-findings-2026.md`.

> For the women-only multi-source Social Sales Safari schema, evidence ledger fields, geography reporting, compressed platform-as-research route, and timeout-safe collection protocol, read [`references/women-transition-social-safari-protocol.md`](references/women-transition-social-safari-protocol.md).
>
> For translating women-between-identities research into one coherent CL product ladder—assigning Magic Cloud, AudioBoosts, LumoSide, Portal, Gateways, embodied work, cacao and premium distinct non-competing roles—read [`references/women-between-identities-product-line.md`](references/women-between-identities-product-line.md).
>
> For the 2026 company-image comparison, Katya Ro loyalty-funnel reconstruction, 549-review mining synthesis, product UX implications, and the approved professional-transition positioning/bio, read [`references/company-models-product-funnels-and-review-mining-2026.md`](references/company-models-product-funnels-and-review-mining-2026.md).

## Source-grounded course homework adaptation

When Margarita sends a course chapter or workbook and asks to preserve it and complete its assignment for Cosmos Luminosity, treat the source as both a library item and a staged business exercise—not as permission to synthesize every downstream deliverable at once.

1. Preserve the original and extracted text before analysis; register the source in the CL library.
2. Extract the exact homework spine and create one cumulative working document with a visible checklist.
3. Work one stage at a time when later outputs depend on earlier choices. Present the stage artifact and obtain correction/approval before writing dependent product copy, sales posts or site changes.
4. Keep three layers explicit in every stage: **named source → confirmed CL context → agent adaptation**.
5. Never fill evidence-dependent fields with plausible invention. Real mini-interviews, testimonials, client outcomes, current revenue, prices, paid counts and acquisition sources remain open until supplied or verified.
6. Treat source product formats as functional roles, not mandatory implementation. Reconcile them with CL’s approved delivery boundaries—for example, replace a suggested consultation with a self-guided equivalent when founder participation is intentionally absent.
7. Separate creator fit from market evidence: what Margarita loves and does naturally can generate hypotheses, but does not prove audience demand or willingness to pay.
8. Do not publish or silently alter the live site during a strategic homework pass unless Margarita explicitly asks for implementation.

For the chapter-specific source spine, anti-fabrication rules and stage-one acceptance criteria, use `references/katya-ro-product-money-people-homework.md`.

## Guided source-homework and product provenance

When Margarita asks to complete a supplied course/workbook exercise, conduct it conversationally rather than auto-filling the entire framework:

- ask one source prompt at a time;
- accept multipart voice/text and wait for her explicit `«конец»` before synthesis;
- preserve her answer separately from the named source and agent adaptation;
- after synthesis, show/save the cleaned answer, then move to the next prompt;
- do not invent interview feedback, revenue, conversion, pricing or demand to complete blank fields.

Before asking her to describe an existing offer, verify product provenance: whether it originated with Margarita, customers, or an agent; its original meaning and delivery; behavioural evidence; and current approval status. A strategic draft does not turn an agent-proposed offer into Margarita’s product. In particular, `Gateway` originally names Margarita’s live head-touch/head-massage experience; never silently repurpose it as a digital transition offer.

When the product is unknown, research current competitors and adjacent mechanics first, then present 3–4 materially different concepts with experience, output, reason to pay, differentiation, build complexity, and the cheapest transaction test. Keep offer existence separate from evidence of demand.

For the full Katya Ro homework sequence, one-question facilitation protocol, provenance table and competitor-to-concept workflow, use [`references/katya-ro-product-homework-workflow.md`](references/katya-ro-product-homework-workflow.md).

## Trial-layer progression: do not repeat the awareness interaction

When the awareness layer already uses a short interpretive check-in, the trial layer must not become another questionnaire, diagnostic, map, or polished worksheet. Reuse the existing answers to route the woman into a materially different experience.

A preferred Cosmos Luminosity route is:

> interpretive check-in → possible reading → AudioBoost selected from the answer pattern → one bounded real-life experiment → optional Evidence Review → paid continuation

- The interpretation serves thinking; the AudioBoost supports state; the experiment creates evidence.
- Use a small library of shared, editorially authored AudioBoosts. Do not fabricate detailed personal futures or call generic/generated audio “personalized.”
- Personal relevance comes from transparent routing, a short explanation of why this journey opened, and the experiment selected afterward.
- Do not show a catalogue and ask the woman to prescribe her own AudioBoost.
- Place the paid continuation after the lived experience, not immediately after the text result.
- Do not promise this route publicly until the relevant recordings and delivery path are actually ready.

For the five-journey library, routing signals, site copy boundary, result-page transition block, and validation events, use `references/transition-insight-to-audioboost-routing.md`.

## Product-before-platform sequence

For an unvalidated English-first offer, use one of two modes.

### Standard sequence

1. Identify people Margarita genuinely wants to work with.
2. Mine current public voice-of-customer language.
3. Interview representatives without pitching or leading.
4. Find a repeated situation, trigger, desired result, attempted alternatives, and purchase behaviour.
5. Create one minimum paid test.
6. Sell and deliver manually.
7. Collect completion, outcome, objection, and continuation evidence.
8. Revise or repeat.
9. Build the durable site journey and automate only the repeated, validated process.

### Accelerated build-and-validate mode

When Margarita explicitly wants to move into platform creation quickly and a working site/ecosystem already exists, do not make a long standalone research phase a prerequisite. Build one narrow, complete and instrumented route while validation runs inside it:

1. use the existing research to choose one transition situation;
2. conduct three short qualified conversations in parallel;
3. place a brief check-in inside the platform;
4. return a useful personalized result, not merely a thank-you page;
5. offer three paid alpha places for one bounded continuation;
6. use completion, action, payment and continuation behaviour to shape the next release;
7. defer broad automation and product-line expansion until the route repeats.

A site should not be used to discover an entire market through undirected building. It **can** be used as a research instrument when the route is narrow, useful, measurable and connected to a real paid test.

For the complete accelerated protocol, current-stage model, build-now/defer split and evidence ledger, read `references/accelerated-platform-validation.md`.

## Audience research standard

- Segment by life situation and job-to-be-done, not demographics alone.
- Save exact English quotes before translating them.
- Every public quote presented as evidence needs a direct source URL and date where available.
- Do not infer wealth, gender, profession, geography, or readiness to pay from names, profile photos, location aesthetics, job titles, or community membership.
- For women-only research, include a record only when the author self-identifies as a woman (`woman`, `F`, explicit first-person bio/profile evidence). Record the exact evidence; otherwise mark gender unknown and keep the item outside the women-only count.
- Report geography only from explicit self-report. Separate **geography of the corpus** from **geography of the market**, and always name source/platform bias before saying where the audience appears concentrated.
- Prefer behavioural payment signals: prior adjacent purchases, independent purchase authority, active search, dissatisfaction with alternatives, and ability to apply the result. Keep income/capital, category spend, willingness to act, and willingness to buy this offer as four distinct signals.
- Search for disconfirming evidence and differences between apparently similar subgroups.
- Do not build a persona from fewer than five independent data points in a coherent segment.

### When Margarita asks to skip interviews

Do not continue prescribing interviews as a mandatory gate. Replace primary interviews with a documented public Social Sales Safari and clearly label the resulting ICP as provisional. Mine several source contexts where public access permits; if the accessible corpus is platform-heavy, preserve the useful language while explicitly limiting market-level conclusions.

Collect per record: exact quote, direct URL, date/platform, self-identification evidence, explicit geography, role/stage, resource signals, trigger, push, pull, what must be preserved, fears, doubts, attempted solutions, actual purchases/prices, what helped or failed, tool openness, action readiness, and CL fit/exclusion.

Public research can validate repeated language, situations, objections, prior category spend, and observable action. It cannot by itself prove demand for a new Cosmos Luminosity offer. If direct conversations are intentionally skipped, use the platform itself as the research instrument: recognition page/check-in → useful first result → concrete paid pilot/deposit → completion/outcome/continuation evidence.

Confidence labels:

- **High:** repeated unprompted theme across 5+ people or 3 source types, supported by behaviour.
- **Medium:** 2–4 people or one source type, partial behavioural support.
- **Low:** a single source, unclear profile, prompted answer, or stated intention only.

## Content growth and validation

When using viral-format research:

- distinguish the source method from current platform advice;
- borrow abstract structure, never wording, footage, story, or identity;
- judge outlier performance relative to the creator’s normal reach, not raw views alone;
- verify that comments and new followers resemble the desired audience;
- combine reach content with expert/trial content;
- keep reach, relevant follower growth, trial uptake, conversations, purchases, completion, and outcomes as separate signals;
- never report reach as product validation or followers as willingness to pay.

## Deliverables

Depending on the ask, produce:

- source-only algorithm;
- VOC quote bank with URLs;
- JTBD comparison by segment;
- purchase/alternative/gap map;
- interview guide;
- minimum paid-test brief;
- viral-form library with original CL adaptations;
- validation report separating facts, hypotheses, and decisions.

## Positioning discipline for the integrated model

When translating therapy/coaching/spiritual research into Cosmos Luminosity positioning:

- do **not** describe CL as a blended therapy + coaching + esoteric service;
- lead with the recognizable customer situation—an accomplished woman outgrowing a professional chapter while the next is not yet clear—not with an invented category or integration claim;
- use **Professional Transition Support for Accomplished Women** as the market-facing shelf unless Margarita approves a later change;
- describe the method as a private whole-person transition system connecting mind, body, state, inner knowing and real-world evidence; `Grounded Transition Architecture` may remain internal rather than the headline category;
- preserve specialist boundaries and the woman’s authorship;
- keep body/state and symbolic layers opt-in while practical constraints and real-world evidence remain mandatory;
- define the audience through transition situation, demonstrated accomplishment, maneuverability, agency, and an active consequential choice—not age, title, luxury aesthetics, or HNW imagery alone;
- lead with the beachhead rather than presenting every adjacent transition as equally primary;
- distinguish a canonical strategic document from live site implementation: when Margarita asks what the audience/positioning is, save and present the strategy first; do not silently rewrite the site unless she asked for implementation;
- after Margarita approves with “save,” update the canonical positioning source and product-marketing context together, bump versions/changelogs, create a short approved-copy artifact, verify the saved text, and return the complete approved wording;
- treat approval of a microcopy change (bio line, CTA, tagline) as a canonical-copy update, not a chat-only preference: update the canonical positioning document, the short approved-copy artifact, and `.agents/product-marketing.md`; update the Russian sense translation too; add a concise changelog note explaining the semantic reason; then search the working corpus for the superseded exact phrase and confirm that no active occurrence remains;
- for Cosmos Luminosity CTA language, distinguish an experiment from an exam: when the intended meaning is a small reversible first step, prefer language such as `Begin with an experiment` over `Test it in life`, which can imply proving that a choice is correct.

## Pitfalls

- Do not overbuild strategy when the user asked for extraction.
- Do not assume Russian demand transfers to English-speaking audiences.
- Do not merge relocation, divorce, career transition, multipassionate identity, and practitioner-business problems without evidence that they share one job.
- Do not ask “Would you buy?”; ask what people already bought, why, and what happened.
- Do not automate an untested offer.