V3
DETERMINANT MARKET STUDY · PLAN UPDATE · CONFIDENTIAL

From map to decision

What changed in the study on 17 July, the logic behind each change, and the run to the August business plan workshop.
PREPARED 17 JULY 2026 · FOR THE WORKING SESSION WITH DEE, TUESDAY 21 JULY
DECISION HISTORY · superseded as the entry point by the August read-first
01

Why this update: the study was building a strong instrument, aimed at the wrong terminal output

On 17 July we audited the whole study against the outcome fixed in the 29 June kickoff: DETERMINANT's next plan gets built market-in, which requires two upstream decisions, what type of brand DETERMINANT is and what price tier it sits at, plus the investment required to hold that position. The audit found the instrument in good shape and the aim drifted. Six findings, each with its receipt:

  1. The study stopped converging.The original plan ended with "August: positioned recommendation delivered into the 3-year plan." Later scoping ("no recommendation at this stage", "rolling library, no fixed end date") deleted that step. The study terminated in a map plus an open-ended library, with no point where a call gets made.
  2. Only half the question was being served.Price is the X axis of every view. No instrument existed for the brand-type decision at all.
  3. The hypotheses were aimed at a market no longer on the roster.H2 tests a band defined by Everlane, Bonobos and LV. All three were dropped when Dee confirmed the shortlist on 16 July. The roster changed; the hypotheses did not.
  4. H1 was promoted from instinct to constraint without a test."No viable position below the current tier" was hardcoded into the map's price bands as "closed to DETERMINANT" while the market size of that band reads null. It is the only hypothesis nothing could kill.
  5. The decisive number ran on the softest instrument in the study.The spend required to hold each tier was a 1-to-5 judgment score calibrated against listed-company S&D ratios, which mostly measure rent and store payroll, not brand investment.
  6. The feasibility screen had no input."What is DETERMINANT's core competency, concretely" has been an open question since kickoff. Without it, no option can be ruled out and every door looks available.

None of the evidence is wasted. The staff survey (n=158), the confirmed roster, the price benchmark discipline and the map engine all stand. v3 re-aims them at the decision.

02

The core question, reframed: marketing is one way to hold a price, and it is the one DETERMINANT can least afford

THE QUESTION UNTIL 17 JULY
At what price tier should DETERMINANT sit, and how much marketing and brand spend does that tier require?
THE QUESTION FROM V3
What holds price at each tier, and which of those mechanisms can DETERMINANT actually run?

The old question assumes price is held by buying marketing. Our own shortlist falsifies that assumption four times over. Price is held by different mechanisms, paid in different currencies:

比音勒芬
77% gross margin (FY2024 filing) held by self-developed fabric, high-end malls, membership and a no-discount rule. Product and retention, not media: only 4.9% of revenue goes to advertising against 29.2% to stores, staff and rent.
龙牙
RMB 1.8B revenue, 52% repurchase held by precise sizing and function translated to commuter value. Product proof.
Vuori
Premium price held by fabric-first product and word-of-mouth community seeding, with no KOL spend.
衬衫老罗
Margin held by factory-direct trust and stored-value retention. Near-zero brand spend.
打版 factories
Found 20 July. KKRS ERIC, DUTTMOKN, GFOR, Gremade hold 388-679 at 85-100% of ticket on rented luxury patterns and premium fabric, sold direct online. No stores, no ambassadors, no media.

DETERMINANT China runs a 66.9% gross profit against a 70% healthy bar (up 4 points on FY2025, internal report YTD Jun-26), with a 2025 EBITDA loss of RMB 15M. The media mechanism is the one mechanism the P&L cannot fund. The study had spent its full weight on exactly that mechanism. The internal financials since confirmed it quantitatively: selling cost runs 55.1% of sales, leaving a spread of 11.8 points, and the split is roughly 24% media against 19.5% physical. 比音勒芬 runs the mirror image at 4.9 media / 29.2 physical for a spread near 37. DETERMINANT pays for the expensive mechanism and is cutting the one the premium door actually requires.

What this changes in the map. The invented "brand + creative investment" score (1 to 5) is deleted as an axis. Gross margin, a measured number, replaces it. The default view becomes shirt take-home price x gross margin, colored by mechanism: at every price point, who holds margin, and by what means. DETERMINANT's dot sits at 66.9% and the viewer can see at a glance which mechanisms clear it.

This also dissolves a problem the study could not solve: a single "spend to hold the tier" number never existed, because each mechanism is paid in its own currency. Fabric R&D, product engineering, channel capital, media. The doors below price each in its own terms.

03

Two decisions, two surfaces: brand type is chosen once, price tier is chosen per market

Dee's kickoff named two upstream decisions. v3 gives each its own surface and its own altitude:

04

The six doors: every viable answer, now carrying real evidence

The mechanisms that hold price are the type options. The doors were written down on 17 July, before collection, so the data could confirm or kill rather than decorate. Updated 21 July: the full price dataset (226 of 234 cells) and the internal financials have since landed, so the status column below now reflects measured evidence rather than expectation. Every door is still open to the screen; none has been closed on the strength of this table alone. One door was added after registration. D6 did not exist in the 17 July set: the 打版 segment was found during collection, and it holds price by a mechanism none of the original five describes. It is flagged as post-registration rather than folded in quietly, because the point of writing the doors down early is to make exactly this kind of surprise visible.

DoorPrice held byProof brandsWhat it demandsStatus
D1 · Mid-high national Channel scale and breadth Youngor 雅戈尔 (top cross-shop brand, 78 mentions), HLA, JoeOne, LILANZ National channel scale and franchise mechanics DETERMINANT does not have; a volume fight against an Esquel cost base. OPEN
The default door by gravity, 58% of purchases. The price data now cuts against it: their tickets are fiction (Youngor holds 44%, SeptWolves 37%) and their real prices land on ours (JoeOne 282, SeptWolves 224, Youngor 344 against DET 318). DETERMINANT already runs this door's economics, spread 11.8, without its scale. Choose it deliberately or kill it.
D2 · Focused premium Product authority, no-discount discipline, private-domain retention 比音勒芬 (77% GM per FY2024 filing, 13 straight growth years, real shirt 769 held at 87-95%), ICICLE 之禾 (2846) Margin headroom DETERMINANT still lacks (66.9% vs 77%), fabric R&D, discount discipline, and above all physical retail: the proof brand spends 29.2% of revenue on stores, staff and rent, the exact line DETERMINANT is cutting from 9 stores to 4. 比音勒芬 took 13 years; the plan has 3. OPEN
Dee's instinct (H2). The gate moved: margin is no longer the only bar (66.9 now, +4pp), the binding question is whether DETERMINANT will fund physical presence. The proof brand's machine is retail, not media, six to one. One tell worth knowing: its signature shirt is 100% polyester, so the hold is brand and retail, not fabric. Fabric provenance is a card Esquel holds and it does not.
D3 · Function hero Function proof translated to commuter value 龙牙 (RMB 1.8B, 52% repurchase), 白小T, 布先生. Separately, the premium outdoor wallet leak (Descente, Arc'teryx, Kolon, China only) is pressure evidence, not a model DETERMINANT could run: it proves function holds price, at a scale we cannot reach. Product engineering and hero-SKU discipline. Plausibly closest to Esquel's real competency; the least media-dependent door. OPEN
The sleeper, and the price data strengthened it. The wallet trio holds at zero discount: Descente 790 polo and 590 tee, Arc'teryx tee 1000, Kolon at 90% hold. Function buys the strongest price integrity measured in the study, at 2-3x DETERMINANT's price, with almost no media. 龙牙 runs one-price honesty (169-319, no ticket theatre) at RMB 1.8B.
D4 · Digital-native Design voice, content-native community Nice Rice, Pane, Bananain 蕉内 Not taken further unless the confirm dive surprises. PRE-KILLED
0 of 121 mainland respondents cross-shop Nice Rice. One confirm dive only. Correction, 20 July: the earlier "no woven shirt" finding was an AI search failure, the shirt exists at 598 held at 100%. The kill therefore rests on the demand signal alone, and that signal stands.
D5 · Factory-direct The factory is the credibility; near-zero brand spend 衬衫老罗 229, 霞湖世家 238, 欧定衬衫 790, all at zero discount; modern variant Quince Cost discipline and a transparency story. Esquel manufactures for Ralph Lauren and Hugo Boss, both sitting on our own map at 41 and 18 mentions. Prices honestly at or below where DETERMINANT sits today, so it is a margin-and-volume bet, not an elevation bet. REOPENED
H1 closed this door on instinct; Dee's own Quince and Kirkland references keep pointing at it. 欧定 holding 790 at zero discount proves factory-direct does not mean cheap. Test it, then keep or kill on data.
D6 · Factory premium Design authority rented onto a factory cost base, sold direct into the white space The 打版 segment, found 20 July: KKRS ERIC 388, DUTTMOKN 466, GFOR 594, Gremade 顾大师 679, all holding 85 to 100% of ticket. Second ring already at 600-1200 (RIZZUP, 老乱, 东崛, SageSalt). Design authority DETERMINANT would have to acquire rather than rent, plus fabric buying and content discipline. No stores, no ambassadors, no media. The economics are the ones Esquel already has; the missing input is the design credibility these players borrow from luxury houses. NEW · POST-REGISTRATION
Added 21 July, after the segment appeared in collection. The only door whose brands occupy the empty 350-620 band, one step above DETERMINANT's 318. They rent precisely what Esquel owns outright: the factory, the fabric provenance, and the Ralph Lauren and Hugo Boss client list. Distinct from D5 because it points up-market, not down.
05

The screen: five numbers per door, and a designed way out if all doors fail

Each door must show, in its own currency:

If every door fails the screen, that is a finding, not a failure. The 2025 loss driver was a RMB 5M celebrity livestream with too little inventory: an execution problem no positioning choice fixes. The per-brand price-vs-perceived-value diagnostic already in the study decides the question: if DETERMINANT sits below its perceived value, the problem is underpricing and operations, not position. The memo is allowed to reach that verdict.

The memo the doors produce

One section per live door, one snapshot per claim, map attached as the appendix:

06

Changelog: what v3 changes in the build

AreaWasNow
Core questionWhat tier, and how much marketing to hold itWhat holds price at each tier, and which mechanism DETERMINANT can run
Terminal outputMap plus rolling deep-dive library; recommendation deferredFive priced doors, a screen, and a decision memo before the workshop
HypothesesH1 a fixed constraint; H2 aimed at a band no longer on the rosterRewritten as the doors; H1 becomes a testable claim (D5). D6 added post-registration on 21 July when collection surfaced the 打版 segment
Map Y axisBrand + creative investment, scored 1 to 5Gross margin, measured; mechanism as color
Schemabucket, three elevation scores, MGFI archetype, identity alignment, archetype, AI discoverabilityCut. Added: grossMargin, mechanism, yearsToProof, apacFlex, coreCompetency, door, heroProduct. lifestyleBreadth promoted to map tier. Polo prices kept in schema, not collected.
MethodsTwelve methods, all activeSix. Keep benchmark tables, triangulation, price-band sizing. Promote cross-market price checks (APAC flex), the prior poll, scenario modeling (it was always the memo). Cut composite scoring, archetype alignment, correlation matrix, EP lens. Defer the Tmall SKU tracking and persona panels.
Deep divesCompetitor stories with a verdict attachedModel proofs. Each outputs margin, cost in its own currency, years to proof, competency, transferability. Tier-2 model studies become Map B rows. Quince readmitted as a model exemplar. D4 shrinks to one confirm dive.
APACFull peer map and studyField map ships; door analysis China-first with an APAC-survival column; full peer study deferred
Data collectionGM teams collect prices and channel dataAlfred + AI sweep for ticket prices and listed-company filings, source-flagged, sample hand-checked. Take-home prices hand-collected for door exemplars only. GMs contribute judgment through a short prior poll, not labor.
SiteBriefing + mapAdds this page (/plan/) and next a /doors/ surface for Map B and the screen. Survey pages publish alongside.

What stays

The staff survey (n=158) and everything it settled. The Dee-confirmed roster, since extended by collection to 39 brands (32 China + 11 APAC, own brands included). The price benchmark discipline: three anchor items, ticket vs take-home, pre-registered expectations. The schema-driven map engine, which absorbs all of the above as data changes only. The eight-brand dive shortlist, re-tasked rather than replaced.

07

Three points to confirm with Dee on Tuesday

Each changes something previously aligned, so each is put forward for confirmation rather than absorbed silently.

1 · The study now ends in a decision structure, not a picture

"Full picture first, recommendation later" was right for the framework phase. For the workshop it leaves the hardest step unowned. v3 delivers named doors with costs and a screen, so the room debates a small set of priced options instead of reasoning from a 38-brand terrain. The map remains attached as the evidence layer.

2 · APAC narrows from a second map to a survival column

The 8 July alignment said China + APAC both. The data since shows two disjoint markets and an APAC set that is mostly well-documented global brands. Proposal: ship the APAC field map as built, run the door analysis China-first, and make "does this door survive in APAC, at what price gap" an explicit column. A full APAC peer study follows later if wanted.

3 · Factory economics go back on the table, as two doors

H1 ("no viable position below the current tier") closed this door before the study began. Dee's own references, Quince as the long-term economic model and Kirkland as the taxonomy anchor, keep pointing at it, and Esquel's manufacturing story is the strongest single asset on the map. Collection then showed there are two different moves here, not one, and they point opposite ways on price. D5, factory-direct, prices honestly at or below where DETERMINANT sits today and sells the factory story; 欧定 holds 790 at zero discount, so it does not mean cheap. D6, factory premium, is what the 打版 players are doing right now: rent the design authority, keep the factory cost base, and sell into the empty 350-620 band one step above DETERMINANT's 318. A second ring is already trading at 600-1200, so that space is filling now rather than later. D6 was not in the pre-registered set; it is flagged as a post-registration addition because collection surprised us. The ask is not to choose either one; it is to let the data price both, and to say which of the two you actually mean when you say factory.

08

Timeline: two weeks, one gate

Fri 17 Jul · done
Plan v3 written (this page) and published. Map rebuilt on the reframe: gross-margin axis, mechanism and door fields, the model-map view. Deliverable template live at /report/ with automatic gap tracking.
Sat 18 Jul · done
Wilkie and the GMs confirmed no bandwidth for collection; the study runs self-serve (Alfred + AI). Listed-company wave 1 pulled and loaded: 13 real gross margins on the map, sources logged. First hard finding: 比音勒芬 holds 77% gross margin by spending 40.3% of revenue on selling; the volume floor runs 44 to 52; Youngor's main brand holds 72.9% on declining revenue.
Sun 19 to Mon 20 Jul · done
Forty real take-home prices hand-collected; ticket fiction measured (old-school tier holds 11-44% of ticket) and the tier collapse found (their real prices land on DETERMINANT's 318). 打版 luxury-block segment discovered and mapped. Internal financials read: CN spread 11.8 with the media/physical mix inverted vs the D2 proof brand. Price bands locked on the real-price gaps. Walkthrough rehearsal tonight.
Tue 21 Jul · THE GATE
Working session with Dee: the full evidence walkthrough (order below), then the three confirmations and the core-competency question. Dee picks a door to lead with or narrows to two.
Wed 22 to Fri 25 Jul
Prior poll to the brain trust on the confirmed doors. Collection continues: exemplar take-home prices hand-collected, AI ticket sweep across the roster, source flag on every AI cell with a 10-cell hand check.
Mon 27 to Wed 29 Jul
Doors priced and screened. Six snapshots drawn. Memo v2.
Thu 30 to Fri 31 Jul
Map ships with real data, honoring the end-of-July commitment. Memo goes to Dee.
Early Aug
Business plan workshop builds the 3-year plan on the chosen door. Deep dives continue as a rolling library of model proofs.

The Tuesday walkthrough: show everything, in argument order

Everything built so far gets shown; the order keeps it a decision conversation rather than a tool tour. Each stop makes exactly one claim.

Update, 20 Jul: the internal financial reports landed, so door costs no longer stop at benchmark ranges; DETERMINANT's own selling structure (55.1% CN, split media vs physical) prices each door against the real cost base. If the direction changes on Tuesday, it changes with two weeks in hand rather than in the workshop.