Precise × Polymarket

Every market forces a price to prove itself against a real outcome.
We do that for spend.

Precise prices media decisions the way your markets price events, and grades itself against what actually happened.

You · board 01Spencer opens and hands to Peter; his vouch carries the first two minutes. Then this line, once, slowly. Do not explain it; the boards do that. If Anthony pushes here, jump to board 06 (key 6), give him the holdout sentence (you pick the slice we never see; you write the acceptance rule), then return to 02. If the call compresses to fifteen minutes, the spine is 01 → 03 → 06 → 07 on the number keys: board 04 becomes one beat on the HELD row, 05 appears only if MMM comes up, 08 is the goodbye.

Your machine is already good. It is just blended.

Install‑to‑deposit funnel on one side. Every major DSP on the other. The decision that connects them is made at blended grain.

We join the two sides and price the decision at log grain: what each incremental dollar returned, where the next one should go.

The next four boards are the screens we stand up. Their feeds go underneath in week one; everything in teal is Precise.

You · board 02Their words, not ours: the funnel and the DSP list came from the Jul 9 call. Do not name a share number or a competitor unless they do first. MMM gets its full answer on board 05, not here.

The read: their feeds, joined and priced

Numbers on these four boards are illustrative; the screens are the week‑one build.

Precise addsmarginal CPA and headroom per partner, beside the blended numbers they already readWhy it winsblended CPA says who looked good; marginal CPA says who earns the next dollar. Same logs, one join, different question

Today · the blended readThe report says The Trade Desk looked cheapest: $226 per deposit, averaged over the whole flight.

Joined · the priced readThe join says TTD still earns the next dollar: $268 marginal ± $31, with $400k+ of headroom — while DSP D's $419 blended hides a $540 marginal: saturated.

what this board is doing: joining AppsFlyer to DSP logs and pricing the NEXT dollar, instead of averaging the last one

polymarket × preciseWorld Cup flight · retrospectiveshadow mode · no spend authority

Flight: Jun 11 – Jul 19Spend joined: $4.82M · 91%First deposits: 18,940Blended CPA: $254Outcome: first deposit · 7‑day window

Partner	Channel	Spend	First deposits	Blended CPA	Marginal CPA1	Band	Headroom
The Trade Desk · CTV	CTV	$1,610,000	7,120	$226	$268	± $31	$400k+
StackAdapt · CTV	CTV	$940,000	3,410	$276	$342	± $44	limited
DSP C · video	OLV	$780,000	3,890	$201	$219	± $26	$300k+
DSP D · display	Display	$620,000	1,480	$419	$540	± $88	saturated
Creator lane	Social	$470,000	—	—	unjoined	—	not yet2
Flight		$4,820,000 (joined $4.42M)	18,940	$254	$291 blended

Precise · the coverage read, said first 91% of spend joins cleanly at placement grain; the creator lane does not join at all today, which is itself a finding. Marginal prices run above blended everywhere, as they should late in a flight: the question is never who was cheap on average, it is where the next dollar still buys deposits inside the band.

Board 03The joined read. Their columns on the left, exactly as AppsFlyer and the DSP report them; the teal columns exist only because the join exists.

You · board 03Walk it left to right: "your columns, then what joining buys." The coverage read is said before any claim, and the unjoined creator lane is a finding, not a failure. One beat on headroom, then move.

The book: cut and fund, with a refusal

Precise addsranked moves with expected results and reasons, and a stated refusal when the math cannot support oneWhy it winsa recommendation you can act on has a source, a destination, an amount, a prediction, and a band; anything less is commentary

Today · the recommendationAn email that says "shift budget toward CTV."

The decision rowCut $180k from DSP D display, fund TTD CTV prime, +710 first deposits predicted at $254–269, confidence high — and when the math cannot support a move, the row says HELD.

what this board is doing: making every recommendation a complete, gradeable decision — source, destination, amount, prediction, band, or a stated refusal

polymarket × precisethe growth opportunity bookshadow mode · what day 12 of 28 looks like

Move	Cut	Fund	Amount	Expected result	Confidence	Reason	State
01	DSP D display	TTD CTV prime	$180,000	+710 FDs · CPA $342 → $254–269	high	display saturated; CTV holds headroom inside band	PROPOSED
02	StackAdapt late‑night	DSP C video	$120,000	+540 FDs · within band	high	marginal gap $123 at equal reach	PROPOSED
03	—	TTD CTV sports	$90,000	capacity review	—	projected volume exceeds observed capacity; funding above it would be a guess	HELD

Precise · what happens to a row A person approves or declines; either way the decision is written down completely (what was known, what was chosen, what we predicted, what happened, what it taught the system) and graded when deposits mature. Row 03 stays held until capacity is observed, and that refusal is the reason rows 01 and 02 deserve trust.

Board 04The book. The row is the deliverable; the held row is the character.

You · board 04Row three is the credibility beat: HELD. If Jenner asks why we would show a refusal, that is the answer, out loud: the held row is what makes the funded rows safe to sponsor. If Jenner engages on the rows, hand him the pen: what would a row have to show after one closed flight for him to take it into his own allocation meeting? His answer is the success definition. If Anthony presses on identification: key 6, the holdout sentence, come back.

Paths: the best route per ad type

Precise addsa continuously recalibrated route per ad type, where MMM gives one quarterly mix for everythingWhy it winsthe brand film and the referral code should not buy the same surfaces; pricing the path per ad type is the question their taxonomy can finally answer

Today · MMMOne mix, refit quarterly, for the whole account.

PathsA route per ad type, re‑priced weekly: the in‑match :30 runs live CTV in eligible states then retargets OLV at $238 ± 29 — and the creator lane says "cannot price until joined."

what this board is doing: composing priced decisions into the best route per creative — above MMM and below it at once

polymarket × precisepath pricing · by ad typeshadow mode · recalibrated weekly

Ad type	Route today	Priced route	Next dollar	Identified?
In‑match :30	Live sports CTV, national	live CTV in eligible states → retarget OLV	$238 ± 29	yes
Brand film :60	Premium CTV + OOH, everywhere	premium CTV → social video echo, 7‑day	$310 ± 61	partial: OOH unjoined
Performance :15	Programmatic video, broad	OLV in‑app → display recapture	$205 ± 22	yes
Creator post	Flat fees, untracked	cannot price until joined	—	no, not yet
Referral code	Sports media syndication	code‑level, cleanest join in the stack	$121 ± 9	yes

Precise · the honest boundary A route is claimed only where the data identifies it; partial and unjoined rows say so instead of guessing. That is the difference between path pricing and multi‑touch storytelling.

Board 05Paths per ad type. MMM answers what mix, quarterly, for the whole account; this answers which route, weekly, for each creative.

You · board 05This is the full MMM answer (we priced it coming up at 70): above and below at once. Below at decision grain, above because priced decisions compose into routes. Say the boundary line; do not go deeper live. If Anthony presses on identification: key 6, the holdout sentence, come back.

The grading: predictions against deposits, holdout revealed

Precise addsa graded track: predicted against realized, in front of both teams, with a blind holdout scored before revealWhy it winsthis is the same discipline their book runs on: the price proves itself against the resolution, or it corrects

Today · the QBRThe deck that grades itself: everything worked, trust us.

The graded trackSealed predictions against realized deposits, revealed in front of both teams: four within band, one missed — and the miss stays on the board.

what this board is doing: making the instrument prove itself the way a market does — the price against the resolution

polymarket × precisecalibration · what month 1 looks likeholdout: a blind slice, acceptance rule theirs

Partner	Predicted CPA	Realized	Band	Slice
TTD CTV	$268	$259	within	open
DSP C video	$219	$231	within	open
DSP D display	$540	$497	within	open
Holdout partner 1	sealed → $284	$291	within	REVEALED
Holdout partner 2	sealed → $198	$243	missed	REVEALED

Precise · the miss stays on the board This is month one's board as it will look, the miss included: a holdout that misses stays up, gets a named cause, and the next meeting starts from it. Grading in public is what earns trust with the next dollar.

Board 06The grading board. The sealed predictions will come out of escrow at reveal: the shape of the board, not a record. The miss is presented with the wins, which is the whole point.

You · board 06 · Matt's boardSay first: invented numbers, real mechanism; the only real thing on this board is the offer. Then the holdout sentence to Anthony, verbatim and early: "You pick the slice we never see; you write the acceptance rule." His identification question (priced at 75) is answered by this board, not by an argument. The miss row is deliberate; point at it.

The ask: five things, one closed flight

1AppsFlyer rows for one closed flight (your pick; World Cup is our default)

2One DSP's delivery and cost, at the finest grain that moves

3Field dictionary and join keys

4First‑deposit definition, window, maturation

5The decision report you use today

Sample rows first. Coverage and joinability read before any claim. Shadow mode; nothing re‑tagged, nothing touched.

You · board 07The ask is deliberately small: five items, one flight, sample rows not a pipe. We priced acceptance‑in‑principle at 60 and no‑date‑on‑the‑call at 65, so do not push for a date live; name the owner and let the date land in the thread. Two of the five are choices handed to the room out loud: which closed flight, and which DSP goes first. Ask Jenner to pick, and write the pick down as his.

What happens next

1Sample landsFive items, one flight

2Coverage readWhat joins, what does not, said first

3Ranked rowsCut‑and‑fund, holdout sealed

4The revealHoldout opened, misses shown with hits

Before NFL money moves, you know whether this instrument earns a seat.

You · board 08Close with the season clock, not urgency theater: NFL is the reason the retrospective is worth doing now. Leave with a named owner for the sample and Peter holding the thread. Transcript feeds the run book tonight; the Polymarket board grades itself before we sleep.

The three switches · flip during the 2:10 deck review; re‑confirm at 4:30, do not reopen

Switch 1Which line opens board 02: the blended‑grain line as written, or Peter's own framing from the Jul 9 call. Owner: Adam.

Switch 2The price answer if asked directly: scope‑first ("the retrospective sizes it") or a number range. Owner: Spencer. Decide once, say it identically.

Switch 3Who says the five‑item ask: Spencer as organizer, or Matt as the one who reads the sample. Owner: the room.

The lab: we already point Precise at your markets

To stress‑test our own substrate, we built an observatory on Polymarket's public data: sealed market censuses, recipe courts, a paper‑trading loop that grades itself.

So far it refuses to claim an edge. The paper book is down $31.72, the confidence intervals cross zero, and every recipe has failed its own court. The refusals are the instrument working.

It cannot trade — no wallet, no signer, by design. But it means our machinery already speaks your market's language: sealed frontiers, payoff courts, markouts, refusals.

If there is appetite, this is an open door to take Precise deeper inside Polymarket and Publicis than media.

Spencer's pocket board · only if the room invites itNinety seconds, "for fun" register — Spencer loves the betting frame; let him carry it. The nulls ARE the pitch: we graded ourselves on their own markets and refused every unproven edge. Do not oversell: it is a research lab, it cannot trade, artifacts last refreshed mid‑July, and it never touches ad data. The deeper‑inside line is an invitation, not a proposal.