The Revenue AI Report

Buyers push back on AI slop, and the receipts are in

Label identical human writing as AI and engagement falls. Human-written articles drew 5.44 times the traffic. The backlash is measurable across writing, music, and images.

The short answer

What does the research show about Buyers push back on AI slop, and the receipts are in?

Label identical human writing as AI and engagement falls. Human-written articles drew 5.44 times the traffic. The backlash is measurable across writing, music, and images. AI slop is the term the market settled on for machine output shipped at volume with no one accountable for whether it is any good. The pushback against it is not a vibe. It shows up in engagement tests, traffic comparisons, platform policy, and court filings, and it shows up in the revenue motion itself: the same buyers who ask where you use AI will quietly discount everything you send them that reads like nobody wrote it.

Evidence

  • The penalty is about perceived machine authorship, not quality. Identical human writing lost engagement the moment it was labeled AI. Disclosure strategy is a revenue decision, not an editorial one.
  • Buyers are bad detectors and harsh judges at the same time. Most cannot tell AI content from human content, and they punish whatever they suspect is AI. That combination means sloppiness is the real exposure: generic, unattributed, unedited output gets pattern-matched as slop whether or not a model wrote it.
  • The defensible position is receipts. Human-written content outperformed AI content 5.44x on traffic, and human-assisted outperformed pure AI. The market is not rejecting AI. It is rejecting output nobody stood behind.

Supporting pages

AI slop is the term the market settled on for machine output shipped at volume with no one accountable for whether it is any good. The pushback against it is not a vibe. It shows up in engagement tests, traffic comparisons, platform policy, and court filings, and it shows up in the revenue motion itself: the same buyers who ask where you use AI will quietly discount everything you send them that reads like nobody wrote it.

The cleanest demonstration is a labeling experiment, not a quality test. Neil Patel's team took human-written articles, told 1.8 million visitors that a third of them were written by AI, and watched engagement fall on the labeled articles. Nothing about the content changed. The label did the damage. That is the distinction revenue teams keep missing: the penalty attaches to perceived machine authorship, and it attaches before the reader has judged a single sentence on merit.

The production-side numbers point the same direction. NP Digital published 744 articles across 68 websites and tracked them for five months: the human-written articles drew 5.44 times the traffic of the AI-written ones. A follow-up run of 1,000 articles found human-assisted work outperforming pure AI output. And in a detection test run in January 2026 with 100 people shown AI and human content side by side, most could not reliably tell which was which, which means the penalty and the detection gap coexist: buyers punish what they think is AI, and they are bad at knowing what actually is.

The pattern repeats outside writing. Deezer disclosed that AI-generated tracks reached roughly 10,000 uploads a day in early 2025, passed 20,000 a day by spring, and reached roughly 30,000 a day, around a third of all uploads, later that year, and built a detection and labeling system because listeners objected. Getty Images banned AI-generated submissions outright. Brand24 tracked 228,200 mentions of AI-generated content in 2026 and found 48 percent of them negative. And in the field experiment that matters most to this audience, Luo and colleagues showed that disclosing chatbot identity before an outbound sales call reduced purchase rates by more than 79.7 percent.

Editorial disclosure: the editor runs marketing at an AI-native GTM company. That is a conflict on any question about whether AI works in revenue teams, and it is disclosed inside every chart footer that touches vendor claims.

What this page is

The measurable buyer reaction to machine-generated content, across a labeling experiment, a traffic comparison, platform policy, and sentiment tracking.

The argument

The penalty attaches to perceived machine authorship rather than to quality, and it fires before the reader judges a sentence. Buyers punish what they think is AI and they are bad at telling what actually is.

How to read it

  • The labeling experiment is the cleanest evidence because the content was held constant. Only the label changed.
  • The traffic comparison is production data from one publisher with a disclosed incentive, and it is charted with that noted.
  • Detection results and penalty results coexist. People cannot reliably identify AI writing and still punish what they believe is AI writing.

Human-written articles drew 5.44 times the traffic of AI-written ones.

744 articles across 68 websites, tracked for five months. Traffic indexed to human-written articles at 100.

100%

Human-written articles

18.4%

AI-written articles

5.44x more traffic for human-written content

What this does not say

This does not show AI writing cannot rank. It shows what happened to one publisher's set of 744 articles.

Publisher
NP Digital
Sample and method
744 articles across 68 websites, tracked for approximately five months, published 2023 to 2024. Vendor-published marketing research
Field dates
Not published by the source.
Source
No primary URL reachable at research time.

Run by a content marketing agency whose services include human-written content. The incentive is disclosed, the sample is real, and no independent replication exists.

Medium confidence

The pushback is measurable in four different ways.

Each row is a separate study or disclosure. Confidence tags are per figure, not per theme.

  • The label penalty

    Visitors shown identical human-written articles; engagement fell on the third labeled written by AI. Neil Patel, published October 2024. Exact effect size was not published

    1.8M

    Medium confidence

  • Sentiment on AI content

    Of 228,200 tracked mentions of AI-generated content were negative. Brand24, 2026

    48%

    High confidence

  • The detection gap

    People shown AI and human content side by side; most could not reliably tell which was which. Neil Patel, January 2026. Small n, directional only

    100

    Low confidence

  • Disclosure penalty in sales

    Purchase rates when chatbot identity was disclosed before an outbound sales call. Luo et al., Marketing Science, peer-reviewed field experiment

    -79.7%

    High confidence

What this does not say

Each row stands on its own publisher. The ledger is a collection, not a pooled estimate.

Publisher
Multiple sources, see row labels
Sample and method
Compiled from primary sources. Every row links to its publisher in the sources list below
Field dates
August 2026
Source
No primary URL reachable at research time.
High confidence

AI slop flooded music faster than any other medium.

AI-generated tracks uploaded to Deezer per day, as disclosed by Deezer. Thousands of tracks per day.

0918263530Early 2025Spring 2025Late 2025AI tracks uploaded per day, thousands
  • AI-generated uploads per day

What this does not say

Upload volume is not listening. Nothing here measures whether anyone played the tracks.

Publisher
Deezer newsroom disclosures
Sample and method
Deezer reported roughly 10,000 AI tracks a day in early 2025, over 20,000 by spring 2025, and roughly 30,000, around a third of all uploads, by late 2025
Field dates
Not published by the source.
Source
No primary URL reachable at research time.

Company self-report. Deezer sells a detection product built on these numbers, which is an incentive worth naming.

High confidence

The penalty attaches to the label, not to the capability.

Concept view. Where measured pushback shows up, and where AI output passes without comment.

Where the pushback shows up

  • Writing

    Engagement falls on content labeled AI, even when humans wrote it

    Measured

  • Music

    Platforms label, demonetize, or exclude AI uploads as volumes tripled in a year

    Platform policy

  • Images

    Getty Images banned AI-generated submissions outright

    Policy

  • Sales outreach

    Disclosing the chatbot cut purchase rates by more than 79.7 percent

    Peer-reviewed

Where it does not

  • Undetected AI

    Most people could not tell AI from human content in a side-by-side test

    n = 100

  • Human-assisted work

    Outperformed pure AI output in a 1,000-article run

    Measured

  • Answer engines

    AI engines cite structured receipts regardless of who wrote the prose

    Observed

Perceived machine authorship, not actual quality

What this does not say

Documented pushback is not market-wide sentiment. It records where objections were recorded.

Publisher
Assessment by The Revenue AI Report
Sample and method
August 2026. Sources for every measured row are linked below
Field dates
August 2026
Source
No primary URL reachable at research time.

The right column is observation, not a controlled test. The detection gap was measured with n = 100 and may not generalize.

Medium confidence

Also in the record

Figures that sit alongside these charts.

  • The penalty is about perceived machine authorship, not quality. Identical human writing lost engagement the moment it was labeled AI. Disclosure strategy is a revenue decision, not an editorial one.
  • Buyers are bad detectors and harsh judges at the same time. Most cannot tell AI content from human content, and they punish whatever they suspect is AI. That combination means sloppiness is the real exposure: generic, unattributed, unedited output gets pattern-matched as slop whether or not a model wrote it.
  • The defensible position is receipts. Human-written content outperformed AI content 5.44x on traffic, and human-assisted outperformed pure AI. The market is not rejecting AI. It is rejecting output nobody stood behind.

The brief

What is going on here, and why it matters.

The charts above are the evidence. This is the read: what the data describes, the mechanism behind it, where the argument could be wrong, and what a revenue team does about it.

01

The label does the damage

Neil Patel's team took human-written articles, told 1.8 million visitors that a third of them were written by AI, and watched engagement fall on the labeled ones. Nothing about the content changed. That isolates the penalty to perceived machine authorship, which is a reputational variable rather than a quality variable.

In a detection test run in January 2026 with 100 people shown AI and human content side by side, most could not reliably tell which was which. Put the two results together and the operating rule is uncomfortable: your buyers cannot detect AI writing, and they will punish anything that reads like nobody was accountable for it.

02

The production side agrees

NP Digital published 744 articles across 68 websites and tracked them for five months. Human-written articles drew 5.44 times the traffic of AI-written ones. A follow-up run of 1,000 articles found human-assisted work outperforming pure AI output, which is the finding with the most practical value in the set: the winning configuration is assisted, not automated and not manual.

This is the same shape as the agent reliability condition. A cheap human check before publication is the verification step, and publication is effectively irreversible once it is indexed and attached to your brand.

03

The pattern is not confined to writing

Deezer disclosed that AI-generated tracks reached roughly 10,000 uploads a day in early 2025, passed 20,000 a day by spring, and reached roughly 30,000 a day, around a third of all uploads, later that year, and built a detection and labeling system because listeners objected. Getty Images banned AI-generated submissions outright. Brand24 tracked 228,200 mentions of AI-generated content in 2026 and found 48 percent negative.

Three different markets, three different responses, one direction. Where supply of machine output goes vertical, platforms build labeling and audiences build discount rules.

04

What this means for the revenue motion

The same buyers who ask where you use AI will quietly discount everything you send them that reads like nobody wrote it. That discount does not show up in a reply-rate dashboard as a content problem. It shows up as declining response across an entire channel, which usually gets diagnosed as list fatigue and treated with more volume.

The defensible configuration is human-assisted production with a named author and a visible standard, which is also what the peer-reviewed disclosure finding argues for. Volume is the thing being punished, and volume is the thing most AI content programs are optimized to produce.

What to do with it

The move, by seat.

CMO
Move to human-assisted production with named authors. The follow-up run of 1,000 articles found that configuration outperforming pure AI output.
Demand gen
Watch channel-level response, not per-asset engagement. The perceived-AI penalty shows up as channel decay.
Sales leaders
Cap outbound volume before the whole channel gets discounted. The penalty attaches to the sender, not the message.

Questions this page answers

What the data says, in plain language.

What does the research show about Buyers push back on AI slop, and the receipts are in?
Label identical human writing as AI and engagement falls. Human-written articles drew 5.44 times the traffic. The backlash is measurable across writing, music, and images. AI slop is the term the market settled on for machine output shipped at volume with no one accountable for whether it is any good. The pushback against it is not a vibe. It shows up in engagement tests, traffic comparisons, platform policy, and court filings, and it shows up in the revenue motion itself: the same buyers who ask where you use AI will quietly discount everything you send them that reads like nobody wrote it.
What does the figure "Human-written articles drew 5.44 times the traffic of AI-written ones" show?
744 articles across 68 websites, tracked for five months. Traffic indexed to human-written articles at 100. Source: NP Digital. 744 articles across 68 websites, tracked for approximately five months, published 2023 to 2024. Vendor-published marketing research Confidence: Medium. Caveat: Run by a content marketing agency whose services include human-written content. The incentive is disclosed, the sample is real, and no independent replication exists.
What does the figure "The pushback is measurable in four different ways" show?
Each row is a separate study or disclosure. Confidence tags are per figure, not per theme. Source: Multiple sources, see row labels. Compiled from primary sources. Every row links to its publisher in the sources list below Fielded August 2026. Confidence: High.
What does the figure "AI slop flooded music faster than any other medium" show?
AI-generated tracks uploaded to Deezer per day, as disclosed by Deezer. Thousands of tracks per day. Source: Deezer newsroom disclosures. Deezer reported roughly 10,000 AI tracks a day in early 2025, over 20,000 by spring 2025, and roughly 30,000, around a third of all uploads, by late 2025 Confidence: High. Caveat: Company self-report. Deezer sells a detection product built on these numbers, which is an incentive worth naming.
What does the figure "The penalty attaches to the label, not to the capability" show?
Concept view. Where measured pushback shows up, and where AI output passes without comment. Source: Assessment by The Revenue AI Report. August 2026. Sources for every measured row are linked below Fielded August 2026. Confidence: Medium. Caveat: The right column is observation, not a controlled test. The detection gap was measured with n = 100 and may not generalize.
What else sits alongside these figures?
The penalty is about perceived machine authorship, not quality. Identical human writing lost engagement the moment it was labeled AI. Disclosure strategy is a revenue decision, not an editorial one. Buyers are bad detectors and harsh judges at the same time. Most cannot tell AI content from human content, and they punish whatever they suspect is AI. That combination means sloppiness is the real exposure: generic, unattributed, unedited output gets pattern-matched as slop whether or not a model wrote it. The defensible position is receipts. Human-written content outperformed AI content 5.44x on traffic, and human-assisted outperformed pure AI. The market is not rejecting AI. It is rejecting output nobody stood behind.
Where does this data come from?
Every figure is reproduced from a named publisher: NP Digital, AI vs human content study, Neil Patel, labeling experiment, Neil Patel, detection test, Brand24, AI content sentiment report, Deezer newsroom, AI upload disclosures, Luo et al., Marketing Science. Sample, field date, and confidence are shown on each chart. Sources marked as vendor research are labelled on the page.
What could not be confirmed?
The exact effect size of the Neil Patel labeling experiment. The post reports direction on 1.8 million visitors but does not publish the percentage. The positive and neutral shares behind Brand24's 48 percent negative figure. The report leads with the negative share and does not break out the remainder in its summary. Whether the label penalty fades with repeated exposure. No published longitudinal version of the labeling test exists.

Cite this page

Permanent URL and suggested citation.

https://www.therevenueaireport.com/research/ai-slop-backlash

Kvarfordt, Jonathan. "Buyers push back on AI slop, and the receipts are in." The Revenue AI Report, Research Library. https://www.therevenueaireport.com/research/ai-slop-backlash

Figures on this page are reproduced from the publishers listed below. Cite the original publisher for the underlying data, and this page for the compilation and framing.

Sources

Every publisher used on this page.

If a metric, model term, or method on this page is unfamiliar, every one of them is defined in The AI and Revenue Dictionary. Sample size, field date, and confidence tags are explained there too.

Could not confirm

What we looked for and did not find.

Claims found during research and not charted

  • The exact effect size of the Neil Patel labeling experiment. The post reports direction on 1.8 million visitors but does not publish the percentage.
  • The positive and neutral shares behind Brand24's 48 percent negative figure. The report leads with the negative share and does not break out the remainder in its summary.
  • Whether the label penalty fades with repeated exposure. No published longitudinal version of the labeling test exists.

How to cite this research

Written by Jonathan Kvarfordt, Founder and Principal Analyst, The Revenue AI Report. Published under CC BY 4.0.

APA

Kvarfordt, J. (2026). Buyers push back on AI slop, and the receipts are in. The Revenue AI Report. Retrieved from https://www.therevenueaireport.com/research/ai-slop-backlash

MLA

Kvarfordt, Jonathan. "Buyers push back on AI slop, and the receipts are in." The Revenue AI Report, 31 Aug. 2026, www.therevenueaireport.com/research/ai-slop-backlash.

BibTeX

@misc{kvarfordt2026aislopbacklash,
  author = {Kvarfordt, Jonathan},
  title = {Buyers push back on AI slop, and the receipts are in},
  year = {2026},
  publisher = {The Revenue AI Report},
  url = {https://www.therevenueaireport.com/research/ai-slop-backlash}
}

Share this research

Posting to Instagram or TikTok? Copy the link, it carries the title, summary and share image.

Next theme

Job Loss, Fact or Fiction?

Seven in ten Americans now expect AI to shrink the job market. The payroll data, the residency match, the postings index and the Fed's own surveys describe a different economy. This piece measures the gap, lists the losses honestly, and gives each GTM seat one job.

Subscribe

Get the weekly issue built on this data.

Arrives weekly by email. Free. Unsubscribe anytime. By subscribing you agree to our Privacy policy and Terms. We never sell or share the list.