Calidescope LLC · The Throughline × MAVEN

The Outcomes MarketThe record underneath

Everyone sells “Outcomes.” How far each seller stands behind it is printed on the page. This is the evidence behind that read: 140 advertising and media companies, read page by page, with every claim kept verbatim and every count reproducible from the files below.

Read 16–17 September 2026 · adjudicated 17–18 September 2026 · published 28 September 2026

Start here

Three ways in, shortest first.

14 stages

The argument

What 140 advertising and media companies promise on their own websites, and what the gap is worth to whoever closes it first.

22 stages · 25 min

The deep dive

Sector by sector, with the constructs, the evidence pattern and the white space in each one.

How it was built

The method

The sample, the coding, the definitions, and the eight places the record contradicts itself.

The record in figures

Each one is recomputed from the files on this page.

138companies read

627pages captured

2,213claims kept verbatim

740about an “Outcome”

433of those state a figure

28carry a figure from outside the seller

8put something at risk

The thirteen categories

Twelve categories of the advertising and media supply chain, ten companies each, plus twenty companies that sell on “Outcomes” directly. Each row opens its own memo and its own claims file.

Category, and what the word means there	Cos	Claims	Outcome	Figure	Outside	Files
Walled gardens & platformsA conversion attributed inside the platform’s own reporting	10	144	55	37	2	read·claims
DSPs & buy-side platformsA campaign result attributed by the platform, plus hours saved	10	153	50	28	1	read·claims
SSPs & sell-sideBuyer performance and publisher yield, sold from one page	10	166	57	33	7	read·claims
Retail & commerce mediaThe sale itself, read from the retailer’s purchase data	10	158	44	23	0	read·claims
CTV & streamingReach on the TV set, and the attention it holds	10	130	38	17	1	read·claims
Measurement & attributionThe advertiser’s ROI, and the lift from adopting the measurement	10	159	49	26	1	read·claims
Verification & brand safetyPerformance attributed to screened or scored placements	10	159	46	23	2	read·claims
Data identity & clean roomsA client result attributed to a better-resolved record	10	154	46	32	2	read·claims
Marketing clouds CDP & engagementRevenue on channels the buyer already owns	10	156	51	41	1	read·claims
Creative tech & AI contentProduction: assets, hours, and cost per asset	10	172	66	53	3	read·claims
Workflow & media operationsPipeline dollars, and process time	10	159	49	38	1	read·claims
Agencies & holding companiesGrowth, unquantified	10	132	36	6	1	read·claims
Outcome sellers (deep read)A business event credited to exposure	20	371	153	76	6	read·claims
All thirteen	140	2213	740	433	28	claims.csv

The adjudication

MAVEN’s second pass ran at the sector-statement level. These are the files it ran against.

File	Size	What it holds
sector_propositions.csv	78 rows	Each sector statement split into its six parts, with the counts it rests on.
five_patterns.csv	5 rows	The five market-wide patterns, and what each rests on.
matrix_placements.csv	13 rows	Where each sector sits on claim and proof. Judgment, with the counts beside it.
uncaptured_by_sector.csv	92 rows	What each sector’s read could not see.
uncaptured_evidence.csv	5 rows	The five gaps that apply across the whole read.

The full data

Open in a spreadsheet. Every figure published anywhere in this study can be rebuilt from these seven files.

File	Size	What it holds
claims.csv	2,213 rows	Every claim, verbatim, with the page and URL it came from.
pillars.csv	536 rows	Every argument a company makes, and whether evidence sits with it.
companies.csv	140 rows	One row per company: the read, the hero line, the guarantee check.
pages.csv	627 rows	Every page captured, with its word count.
sector_stats.json	13 objects	The counts every figure on this site is generated from.
sample_frame.csv	140 rows	How the sample was chosen.
outcomes20_frame.csv	20 rows	The twenty outcome sellers, and why each one is in.

Two research notes

The reading behind the precedent and the agentic comparison. Each one names its sources and what it could not establish.

Research note

The audience guarantee and the make-good

Dated evidence for the precedent: who guaranteed what, when, and what the remedy was. Every entry carries its source.

Research note

Agentic claims capture: same 64 companies, same instrument

Sixty-four companies re-read on 30 September for claims about AI agents. Eighty claims coded on the same five rungs.

Five things the record will not support

Each one is available from the figures above. None survives what was captured.

01
The market has no proof.
It has plenty. The question is what each number shows, and how far it travels.
02
Vague language means weak evidence.
Walled gardens use the plainest language of anyone and show a method on none of their claims.
03
Outcome sellers are more honest.
They show their method twice as often. Honesty was not measured.
04
No seller puts a fee at risk.
Sellers have guaranteed a business result with a stated remedy at least five times since 2011.
05
The gap is the opportunity.
An absence on a web page is not proof of demand.

With thanks to Marketecture. The company records, segment definitions and news history behind this sample came from the Marketecture Advertising Database. Their API is what made a 140-company frame buildable in two days. The frame is better for it, and the errors in it are ours. maddb.ai · Marketecture Media

Use it. The claim text, the coding and the source URLs are yours to check, argue with, or extend. Where a figure here disagrees with a figure elsewhere, the CSV is the record and the page is the summary. Corrections are welcome and will be published with credit. — Bret Leece, bret.leece@gmail.com