A read of what 140 companies across the advertising and marketing ecosystem claim on their own pages, what evidence sits beside each claim, and where the open ground is. 2,213 claims quoted with their source links; 740 of them about outcomes. Every claim, every argument and every conclusion in this deck was adjudicated by MAVEN for what the evidence establishes and how far it travels.
The outcomes market · September 2026 · Calidescope LLC · The Throughline × MAVEN
home pages carry a claim about an outcome. 41 print the word itself.
of 2,213 captured claims are about an outcome or carry a number that stands for one.
parts of the ecosystem use it, from creative production to clean rooms.
different things the word points at, depending on who is saying it.
Reach. Attention. An asset produced. A click. An attributed conversion. An incremental sale. All of them are called outcomes, on pages that sit next to each other in the same buyer's evaluation.
| Sector | The dominant object the word points at | Outcome claims | |
|---|---|---|---|
| S01 | Walled gardens | A conversion attributed inside the platform’s own reporting | 55 |
| S02 | DSPs · buy side | A campaign result attributed by the platform, plus hours saved | 50 |
| S03 | SSPs · sell side | Buyer performance and publisher yield, sold from one page | 57 |
| S04 | Retail & commerce media | The sale, read from the retailer’s purchase data | 44 |
| S05 | CTV & streaming | Reach on the TV set and the attention it holds | 38 |
| S06 | Measurement | The advertiser’s ROI, and the lift from adopting the measurement | 49 |
| S07 | Verification & attention | Performance attributed to screened or scored placements | 46 |
| S08 | Data, identity, clean rooms | A client result attributed to a better-resolved record | 46 |
| S09 | Marketing clouds · CDP | Revenue on channels the buyer already owns | 51 |
| S10 | Creative tech & AI | Production: assets, hours and cost per asset | 66 |
| S11 | Workflow & B2B GTM | Pipeline dollars and process time | 49 |
| S12 | Agencies | Growth, sparsely quantified | 36 |
| O20 | Outcome sellers (deep read) | A business event credited to exposure; 7 of 20 against a control | 153 |
MAVEN: each definition is the dominant construct visible in that sector's captured pages, not an exhaustive account of every outcome claim inside it. Retail media and measurement in particular carry secondary constructs the single line does not hold.
A buyer comparing two decks has no common denominator. Production hours and incremental sales arrive under the same heading.
That it happened, that it was credited to the ad, and that the ad caused it are distinct. Pages rarely say which one is on offer.
depend on evidence found elsewhere on the same site. The claim often appears before the evidence that supports it.
Most figures arrive without the window they were measured over or the comparison they were measured against, so nothing can be restaged.
and every guaranteed object is upstream of a sale: fraud, attention, incremental reach, delivery.
No page we captured states what the seller owes if the number is missed.
companies publish any pricing page at all. In 103 cases there is nothing to read.
MAVEN: the absence is an absence in public material. 103 missing pricing pages and private contracts mean this cannot establish that no seller's fee moves — only that the public promise stops before the remedy.
Two instruments, two jobs, run in two rounds. The Throughline composes the account of the market. MAVEN adjudicates it — first the claims and the arguments, then every sector conclusion this deck draws from them. Neither ranks sellers.
Most sellers can answer one or two. The map is which ones they can answer at the same time.
The five questions are MAVEN's, drawn from its round-one findings package.
Most claims reach an observed or attributed business result. Forty-one are measured against a control. None establishes what a buyer should expect next time.
Transfer strength across five levels. Nothing reaches level 4. Of 290 named-client claims, 277 sit at level 1.
Independent provenance strengthens who established a result. Twenty-six of the 28 still sit at transfer level 2.
Of 536 arguments. Nothing collapses. The work is in the 101 where the argument says more than the evidence carries.
Adjudication: MAVEN, 17 September 2026. Counts as returned.
Every statement this deck makes about a sector was split into its six parts — the object, the claim pattern, the evidence pattern, the transfer read, the commercial risk, the white space — and adjudicated against the claims underneath it rather than against the deck.
across 13 sectors, each carrying the counts it rests on.
all of them claim-pattern counts, read straight off the record.
the finding stands; the sentence needs its limit printed beside it.
the finding stands; the verb claimed more than the evidence carries.
Nothing failed. Twelve of the nineteen repairs are the same repair, in the same place: the transfer read.
MAVEN's summary of what round two found, and the rule this version is written to. The facts underneath the market account survived the audit intact. What did not survive was a handful of words — nobody, nothing, every, least, does not travel, unclaimed, expect — where authority leaked as the evidence was compressed from claims to sectors to market patterns.
Compression is not the error. Compression is the job. The error is letting the boundary fall off on the way up.
Between them they cover the nineteen propositions that partially held and four of the five market patterns. In every case the fact underneath was left exactly where it was.
Each sector sells a different object under the same word, supports it with a different kind of evidence, and leaves a different thing unstated. Every row below carries MAVEN's verdict on the conclusion drawn from it.
| Sector · the dominant object | Outcome claims | With a figure | 3rd party | Guar. | Not observed on any captured page · MAVEN verdict | |
|---|---|---|---|---|---|---|
| S01 | Walled gardens A conversion attributed inside the platform’s own reporting |
55 | 37 | 2 | 0 | No page in this segment puts an outside party's name next to an outcome number, so an outcome figure that a… Holds · bounded |
| S02 | DSPs · buy side A campaign result attributed by the platform, plus hours saved |
50 | 30 | 1 | 2 | Nothing in this segment pairs a stated price with an outcome someone outside the seller measured; a published… Holds · bounded |
| S03 | SSPs · sell side Buyer performance and publisher yield, sold from one page |
57 | 33 | 7 | 0 | A sell-side fee that moves with an independently measured buyer result is not observed on any page we… Partially holds |
| S04 | Retail & commerce media The sale, read from the retailer’s purchase data |
44 | 23 | 0 | 0 | A retail media network that publishes the incrementality method and the result it produced on the same page… Holds · bounded |
| S05 | CTV & streaming Reach on the TV set and the attention it holds |
38 | 17 | 1 | 0 | No seller in this segment joins a CTV outcome to a named verifier and a published method on the page where… Holds · bounded |
Closer to the sale on the retail side, closer to the screen on the TV side. Where the seller owns the purchase data the figures are richer, and the conditions a second buyer would need to read them across are not published.
MAVEN on retail media: the closed-loop, sale-level pattern is well supported. It does not establish that every outcome in the sector is retailer-owned purchase data, and limited transfer between retailers is not zero transfer.
| Sector · the dominant object | Outcome claims | With a figure | 3rd party | Guar. | Not observed on any captured page · MAVEN verdict | |
|---|---|---|---|---|---|---|
| S06 | Measurement The advertiser’s ROI, and the lift from adopting the measurement |
49 | 27 | 1 | 1 | The segment that adjudicates other sellers' outcomes publishes almost no outside adjudication of its own… Holds · bounded |
| S07 | Verification & attention Performance attributed to screened or scored placements |
46 | 27 | 2 | 4 | No front door in the segment states how a placement bought by an agent or carried in an AI answer is checked… Partially holds |
| S08 | Data, identity, clean rooms A client result attributed to a better-resolved record |
46 | 32 | 2 | 0 | No seller here attaches a commitment, or a figure validated by a named outside party, to the match rate the… Partially holds |
The sectors whose product is adjudication publish little adjudication of their own. Measurement carries one third-party figure across 49 outcome claims. Verification holds four of the market's eight guarantees, all of them over cost per view, completion, savings or attention.
MAVEN on measurement — the strongest capture boundary in the read: Kantar Media is not represented by Kantar-branded pages, Nielsen's case studies are uncaptured, and Circana came back condensed. These statements describe the captured measurement material. The missing pages are exactly where outside validation and methods would live.
| Sector · the dominant object | Outcome claims | With a figure | 3rd party | Guar. | Not observed on any captured page · MAVEN verdict | |
|---|---|---|---|---|---|---|
| S09 | Marketing clouds · CDP Revenue on channels the buyer already owns |
51 | 41 | 1 | 0 | An incremental revenue figure from a named client's holdout, with the period, the baseline and the holdout… Holds · bounded |
| S10 | Creative tech & AI Production: assets, hours and cost per asset |
66 | 53 | 3 | 0 | A published creative test that reports the incremental media result — cells, spend, period and the… Holds · bounded |
| S11 | Workflow & B2B GTM Pipeline dollars and process time |
49 | 38 | 1 | 0 | A published attribution rule behind a pipeline number — what counts as influence, over what window, against… Holds · bounded |
| S12 | Agencies Growth, sparsely quantified |
36 | 6 | 1 | 0 | An agency that publishes its remuneration terms next to one client result read against a stated… Holds · bounded |
| O20 | Outcome sellers (deep read) A business event credited to exposure; 7 of 20 against a control |
153 | 77 | 6 | 1 | No seller in this segment publishes a rate per incremental outcome with a stated remedy, even though the… Holds · bounded |
Creative and workflow sell the operation rather than the market result: assets made, hours removed, pipeline influenced. Agencies sell growth with 6 stated figures across 132 claims. The outcome sellers are the exception, and the only group where a controlled design regularly appears on the page.
MAVEN on agencies: agency capture is unusually incomplete — Horizon, WPP and Accenture Song work pages among the failures. The sparse-quantification pattern survives; a market-wide absence does not. On the outcome sellers: this sector exposes more of the conditions needed to assess transfer than any other in the read, which is not the same as establishing that its results transfer.
An interpretive map. Horizontal: where the claimed outcome sits relative to commercial consequence · Vertical: public evidence authority, read as a synthesis of the sector · Bubble: outcome claims, measured
Nothing sits in the top right. The sectors whose claimed outcome is closest to the sale read it from their own data; the sectors with the most independent evidence claim outcomes furthest from it.
MAVEN: the relative topology is supportable as analytical judgment; the positions are not measured scores. Agencies are the most interpretive placement — weak evidence must not pull the claimed outcome leftward. The outcome sellers' upper-right position is the best supported on the map.
230 of 536 arguments depend on evidence found elsewhere on the same site.
Holds · bounded · 'Elsewhere on the same site' is not synonymous with 'case study,' and off-home-page evidence does not always fully establish the proposition.
Retail media, clouds and outcome sellers publish the richest figures and read them from systems they own. A pattern read across the sectors, not a measured law.
Partially holds · The directional pattern is interpretive: 'distance from sale' and seller-controlled data were not coded as a market-wide measured relationship.
Measurement and verification carry three third-party figures between them across 95 outcome claims.
Holds · bounded · The 3/95 count holds. 'Least' is not literally established because other sectors, including Retail Media, have zero third-party outcome claims.
In creative, assets and hours are counted where media results would be — 53 of 66 outcome claims carry a figure, almost none a media result. In workflow it is pipeline and process time.
Partially holds · Creative strongly supports production outcomes; Workflow is primarily pipeline/process time, not production.
Across 2,213 claims and 627 pages, no captured page states a remedy tied to a result. Eight guarantees exist and all cover something upstream of the sale.
Partially holds · No captured page states a remedy tied to a result. The record cannot establish that no seller's actual fee moves because private contracts and many pricing surfaces are unseen.
| Sector | Not observed on any page we captured | Closest to it today | MAVEN | |
|---|---|---|---|---|
| S01 | Walled gardens | No page in this segment puts an outside party's name next to an outcome number, so an outcome figure that a named third party measured, on a stated method and period, is… | LinkedIn Marketing Solutions, Snap for Business, Meta for Business | Holds · bounded |
| S02 | DSPs · buy side | Nothing in this segment pairs a stated price with an outcome someone outside the seller measured; a published fee next to an independently measured result, with a make-good… | Viant, Basis Technologies, Madhive | Holds · bounded |
| S03 | SSPs · sell side | A sell-side fee that moves with an independently measured buyer result is not observed on any page we captured. | TripleLift, OpenX, Media.net | Partially holds |
| S04 | Retail & commerce media | A retail media network that publishes the incrementality method and the result it produced on the same page, with the measuring party named. | Albertsons Media Collective, Instacart Ads, Koddi | Holds · bounded |
| S05 | CTV & streaming | No seller in this segment joins a CTV outcome to a named verifier and a published method on the page where the outcome is claimed. | Innovid, Samsung Ads, Vizio / Walmart | Holds · bounded |
| S06 | Measurement | The segment that adjudicates other sellers' outcomes publishes almost no outside adjudication of its own results. | Analytic Partners, Mutinex, iSpot.tv | Holds · bounded |
| S07 | Verification & attention | No front door in the segment states how a placement bought by an agent or carried in an AI answer is checked. [Throughline interpretation, not entailed by the outcomes… | HUMAN Security, Scope3, DoubleVerify | Partially holds |
| S08 | Data, identity, clean rooms | No seller here attaches a commitment, or a figure validated by a named outside party, to the match rate the category rests on; the accuracy number that anchors the offer is… | LiveRamp, Experian Marketing Services, TransUnion | Partially holds |
| S09 | Marketing clouds · CDP | An incremental revenue figure from a named client's holdout, with the period, the baseline and the holdout share printed on the page. | Braze, Iterable, HubSpot | Holds · bounded |
| S10 | Creative tech & AI | A published creative test that reports the incremental media result — cells, spend, period and the business-as-usual comparison — rather than the production saving. | Marpipe, Movable Ink, VidMob | Holds · bounded |
| S11 | Workflow & B2B GTM | A published attribution rule behind a pipeline number — what counts as influence, over what window, against the account's prior pipeline. | 6sense, ZoomInfo, Madison Logic | Holds · bounded |
| S12 | Agencies | An agency that publishes its remuneration terms next to one client result read against a stated counterfactual would be alone in this segment. | WPP, Monks, Stagwell | Holds · bounded |
| O20 | Outcome sellers (deep read) | No seller in this segment publishes a rate per incremental outcome with a stated remedy, even though the measurement design that would settle one is already on seven of the… | Haus, Moloco, Recast | Holds · bounded |
MAVEN: each row is an absence in captured public material. None of them establishes feasibility, buyer demand or an unclaimed market opportunity. Two rows are Throughline interpretation rather than a finding the outcomes evidence entails — verification's agent-and-AI-answer gap, and the independence criterion under data and identity — and are marked as such.
Retail media, outcome sellers and clouds all name a business event and count it.
Geo holdouts, ghost bidding and matched markets appear on seven of the twenty outcome sellers' pages.
No captured page says what the seller owes when the named outcome, measured against that control, is missed.
Stated at the size the evidence allows: the three parts exist in the market and never appear together on a page we captured. Whether buyers would pay for the combination, and whether a seller can carry the risk, is the next thing to test with buyers rather than a claim this read can make.
MAVEN: this slide holds as written. It stays at exactly the size the evidence permits and correctly leaves buyer demand and seller risk capacity as things to test rather than findings.
The market has no proof.It holds substantial evidence. The question is what each piece establishes.
Only the claims with third-party evidence are proven.Independent provenance is not a synonym for proof, and it is not transfer.
Only two companies guarantee a business outcome.A market-wide count needs a denominator a public crawl does not have.
No seller puts a fee at risk.Private contracts, make-goods and negotiated terms are not public. Nothing here reads them.
The missing configuration is the opportunity.An absence in public material is not evidence of demand.
The five refusals are MAVEN's, held through both rounds.
| Not captured | Count | Why it matters | MAVEN's ruling on it |
|---|---|---|---|
| Pricing pages | 103 | of 140 companies publish none. Any fee-at-risk language would live there. | 103 missing pricing pages plus private contracts make 'nobody's fee moves' unsupportable. Stronger finding: no captured page states a remedy tied to… |
| How-it-works or method pages | 83 | missing. Method statements that would lift a figure from attributed to causal. | 83 missing method pages mean the record can say causal machinery was not exposed in capture, not that sellers lack it. |
| Customer or case-study pages | 46 | missing, or reachable only as a PDF or a gated form. | 46 missing/gated case-study surfaces can contain named-client, third-party, method and boundary information. Positive observed patterns remain more… |
| Sites that need a full browser | 5 | Netflix Ads and Reddit yielded nothing; Warner Bros. Discovery, Epsilon and Adform are partial. | Especially material where a company contributes zero or partial claims. Do not generalize from fetch failure to market absence. |
| Animated figures | 6 | counters that render as zero to a text fetcher and to an agent, including Disney, Demandbase and Integrate. | Usually unlikely to overturn the market architecture, but it can change sector-level metric density and specific examples; preserve as known capture… |
Named for the adjudication rather than assumed away. Four questions are answerable with targeted work rather than a wider crawl: the contract terms behind public guarantee language, whether sellers carry business-result risk privately, the current mechanics of Publicis' The Pact, and how Pinterest's acquisition changed the tvScientific proposition.
What do you mean by outcome? How do you know it happened? How do you know advertising caused it? Why should I expect it for me? What do you put at risk if it does not happen? Nobody is accused of anything, and the differences between two decks that use the same word become visible in about a minute.
The instrument that makes this publishable is the second one. A market account is a compression, and compression is where authority leaks. Keep the finding; bound the verb.
Method · 140 companies across 12 sectors plus 20 outcome sellers, read 16–17 September 2026 · 627 pages captured verbatim, home page first · 2,213 claims quoted with source links · Adjudication by MAVEN (Michael Risucci): round one, 17 September 2026 — 740 outcome claims scored for source, claim and transfer strength, and 536 arguments put through an entailment audit; round two, 18 September 2026 — 78 sector propositions, 5 market patterns, 13 matrix placements and 97 capture-boundary items. Verdicts as returned, in adjudication/*.csv · figures describe the captured public material, not the whole market · company data from the advertising database (maddb.ai).