§ The cold look
Now the hard part. Most of the raw sources are public — anyone can scrape Netflix Tudum. So the honest question is not "what data do we have" but "what do we have that someone would pay for, and can't easily rebuild?" This page audits the asset cold — what's commodity, what's a moat — then ranks the ways to monetise it, names the risks, and lands on a recommendation.
The instinct is to value the data. Wrong frame: the data is mostly public. Value lives in five things that are not easily copied — and being clear-eyed about which is which is the whole game.
| Asset | What it is | Defensibility | The cold truth |
|---|---|---|---|
| Accumulated history | 8+ yrs of daily worldwide charts (Spotify) + 7 yrs of France behavioural signal, growing daily | Strong moat | Charts are ephemeral — you cannot backfill a day you didn't collect. This is the one asset a competitor cannot buy or rebuild, only out-wait. Every day the cron runs it deepens; every day it's off, it doesn't. |
| Validation IP | Which features proxy mood and which are noise — the Phase-1 work (tempo × savings r = −0.52, mode-major r = +0.43) | Strong moat | Cost real, pre-registered work to learn — and we published the failures, which is itself the trust asset. Knowing what not to claim is worth as much as the signal. |
| Triangulation method | The cross-source read — seven families agreeing, one diverging = the insight | Medium moat | The method is replicable in principle, but the calibration (how the layers weight against each other) is tacit and earned. Defensible while we're ahead, not forever. |
| Wealth-tier & cross-asset read | The trade family — mass→mid→elite mood (Spotify → Amazon → auctions) + brand equity set against cultural attention | Medium moat (once built) | Unique and premium: nobody else sets a share price beside cultural attention, and no one reads mood at three income altitudes at once. Today partly roadmap (markets build-out + ART/Salle integration) — but the highest-margin differentiator for the luxury and finance buyer when it lands. |
| Voice + receipts | Editorial register + every-number-traceable discipline | Medium moat | Brand and trust compound but can be matched by a disciplined rival. The receipts angle is a genuine first-mover position in the AI era — see the LLM thread below. |
| The raw public sources | Netflix, Trends, Wikipedia, GDELT, etc. | Commodity | Individually worth ~nothing — anyone can fetch them. Their value is only in the combination, the history, and the read. Do not price the data; price the decision. |
We are not a data vendor — the data is public. We own unbackfillable history, validated knowledge of what the signal means, and a method + voice that turn ten public feeds into a decision. Price the decision and the trust, never the rows.
Different buyers want different cuts of the same pipeline. The product line already maps to them; the question is which to lean into first.
The flagship buyer. Wants the country read as a decision — "speak heritage, hold premium." Pays for confidence + freshness + receipts. Report £700 · Pro £2,400/yr.
Labels, A&R, sync, catalogue. Wants artist momentum, breakout candidates, cross-border flow, sync targets. This is where our existing logos (UMG, Sony, PRS) live — fastest first revenue.
Wants purchase-side demand — the Amazon "music demand" exclusive. A narrower, premium edition.
New, and the most important. The Cadence Skill ships the country data into Claude/GPT. The buyer in 2027 may be an agent, not a person. Per-seat → per-query → API.
The trade family is built for them: maisons want the wealth-tier read (HNW confidence via auctions); banks want brand-equity-vs-attention. The highest-margin, enterprise-priced cut.
Not a buyer — a channel. A free monthly index, cited, is the cheapest authority-building engine we have.
The terminal buyer. Prices the pipe on rebuild cost — years of unbackfillable history + LLM distribution + a cited public index. The exit, not the revenue.
Seven routes, scored on revenue potential, defensibility and how soon they can ship. The ranking is opinionated — it leans toward what compounds (history, citations, distribution) over what's merely sellable today.
| Route | What it is | Ship | Revenue | Defensibility | Verdict |
|---|---|---|---|---|---|
| ① Reports + Pro | Monthly country read; one-off & subscription | Now | £100–200k Y1 | Medium | The base. Card-swipe priced, proves demand, funds everything. Start here. |
| ② Music Industry edition | A&R / sync / catalogue intelligence | Now | Premium per-report | High | Fastest real money — existing logos, unique 8-yr corpus. Do in parallel with ①. |
| ③ Cadence Skill / API | Data inside the buyer's own LLM; later a query API | Q3 | Per-seat → per-query | High | The strategic bet. AI answers need provenance; ours are machine-citable. This is where the category goes. |
| ④ The Cadence Index | One free, validated mood number per country per month | Q3 | £0 direct | High (indirect) | Not revenue — the citation/authority engine (Big Mac / Fear-&-Greed play). Pipeline already produces it. Highest ROI line item. |
| ⑤ Enterprise / white-label | Private instance, bespoke markets, the pipe under someone's brand | On demand | £30–100k/yr each | High | One land changes the year. Needs the multi-market gap closed first. |
| ⑥ Scored predictions | Named, dated, publicly-scored calls per edition | Q3 | £0 direct | High (indirect) | Track-record-as-marketing. Honest either way; turns the pre-registration discipline into a recurring asset. |
| ⑦ Raw data licensing | Selling the feeds / aggregates directly | — | Low | Weak | Avoid as the core play. The data is mostly public; licensing commoditises us and risks the NPILABS/Spotify terms. Only as a bespoke enterprise add-on. |
Spotify deprecated audio features Nov 2024. The mood anchor's forward freshness depends on in-house extraction (Essentia/Musicnn, 3–4 wks eng). Until that ships, the anchor slowly ages. This is the single biggest technical risk — fund it.
Seven of ten live sources are France-only. The "25+ markets" story is schema-ready, not data-real. Enterprise and exit value both hinge on closing this — it's a config-and-cron job, but it has to actually run.
The deep asset is licensed via NPILABS, not owned outright. Spotify Charts CSV + Deezer + YouTube give redundancy; the relationship is a strength but also a concentration risk to manage.
The country-level forecast claim failed Phase 1 and was withdrawn. The product is descriptive today. That's honest and still valuable — but it caps pricing until the brand-level backtest (Phase 2) passes. Don't re-inflate the claim without the receipt.
Now: ship Reports + Pro (①) and the Music Industry edition (②) — the base revenue and the fastest real money, both fundable today off the existing corpus and logos.
Next quarter: the three compounding plays — the Cadence Skill/API (③, the category's direction), the free Cadence Index (④, the authority engine), and scored predictions (⑥, track-record marketing) — and stand up the trade family (stocks-as-signal + the ART/Salle auction tier), the premium differentiator the luxury and finance buyers pay most for. None is big revenue alone; together they build the moat that makes ⑤ and the exit possible.
Underneath all of it: keep the cron running (history is the only thing nobody can rebuild), fund in-house audio features (protect the anchor), and close the France-only gap (unlock enterprise + exit). Do not chase raw-data licensing — it commoditises the one position the AI era will actually pay for: a behavioural-culture pipe whose every number carries a receipt.
The exit framing follows from the asset audit: not "early-revenue report shop" (Stylus, ~£19M) but "the validated behavioural-culture pipe with LLM distribution and a cited public index" — priced on what it would cost an acquirer to rebuild, and the one line on that invoice they can't pay down at any price is the years of history they didn't collect.