ACQI Content Quality Audit: Key Findings
- AI citation is now measurable across five major platforms, with four of the five moving from zero tracked citations to measurable activity within twelve months.
- Optimization was the only quality signal in this sample that showed a measurable correlation with traffic. Articles scoring 70% or higher on content optimization pulled 5.4 times more monthly traffic than those scoring between 40 and 49 percent.
- Optimization, readability, and fact accuracy operated independently of one another. Improving one metric did not predict improvement in the others.
DesignRush ran a benchmark of 100 published articles to test whether common content quality signals (fact accuracy, readability and content optimization), relate to performance across organic traffic and AI citation.
Scored collectively as the AI Content Quality Index (ACQI), a framework developed by DesignRush, the results show that they do not.
AI citation refers to instances where AI assistants directly reference and link to publisher content within generated responses.
The benchmark was conducted by DesignRush in collaboration with Originality.ai.
Using the company's Lite 1.0.2 model, we evaluated fact accuracy, readability, and content optimization for each article, then cross-referenced those scores with Ahrefs traffic and citation data.
Originality.ai provided analysis credits for the evaluation.
How Content Optimization is Scored
Originality.ai’s Content Optimization Score is a 0 to 100 predictive SEO metric that evaluates content against pages currently ranking for a given keyword, weighing factors like keyword coverage, heading hierarchy, topical authority, and GEO signals such as citations and statistics rather than rewarding keyword stuffing.
Because this is a predictive SEO metric, some correlation with observed traffic is expected by design, so the finding confirms the model aligns with real-world outcomes in our sample, rather than proving the metric caused the traffic.
Optimization Still Drives Performance
The ACQI tested whether on-page quality signals still correlate with traffic in a multi-surface search environment.
They do, but the relationship is steeper than most editorial teams assume.
Articles scoring 70%+ on content optimization pulled 5.4× more monthly traffic than articles in the 40 - 49% band.
The middle bands (50 - 59% and 60 - 69%) clustered near the baseline.
The traffic payoff isn't linear and concentrates sharply in the highest optimization tier.
The relationship wasn’t gradual. Traffic didn’t climb with every small improvement.
The real jump showed up after the 70% mark.
At the full-sample level, the relationship between optimization scores and organic traffic was modest overall, but disproportionately driven by the top-scoring articles.
The gains are concentrated where structural alignment is strongest.
That alignment reflects how search engines and AI systems parse content.
High optimization scores reflect disciplined execution such as structured headings, clear keyword placement, and scannable formatting, which are the signals search engines and AI systems are designed to parse.
Traffic rewards that clarity.
Optimization was the only quality signal in this sample that showed a measurable correlation with traffic.
Readability and fact accuracy behaved as statistically independent signals, and the three metrics moved independently.
In practice, that means content quality is not a single continuum.
It is three separate systems operating in parallel:
- Acquisition (optimization)
- Comprehension (readability)
- Credibility (fact accuracy)
Improving one does not automatically improve the others.
A piece can rank and still be unreadable, or read clearly and still rely on unverifiable claims.
It can be factually rigorous and structurally invisible to search.
No composite score captures all three; each requires its own workflow and its own review discipline.
Readability is the Most Controllable Variable
Readability did not correlate with traffic. That’s not its job.
Instead it reflects cognitive load, meaning how much effort it takes to process a sentence. It doesn't determine whether content is found. It determines whether it's easy to process once it is.
Across the 100-article sample the average score sat in the ideal range, with more technically dense pieces clustering on the lower edge.
There was also no meaningful relationship between readability and fact accuracy. Dense writing wasn’t any more rigorous, and accessible writing wasn’t any less factual. Comprehension and credibility operate independently.
For publishers, readability is controllable and clarity can be improved in a single editorial pass.
Credibility Is the Least Predictive and the Hardest to Fake
Fact accuracy behaves differently. Across the 100 article sample, fact accuracy scores averaged 70.7%, with a wide range.
The lowest score (18%) appeared in a sponsored CEO interview, where most claims were proprietary and unverifiable.
Several brand campaign articles that cited press-release data scored near 100%.
Fact accuracy did not predict traffic performance, and it isn't always fixable in an edit pass.
Accuracy reflects the quality and verifiability of the underlying source material.
One of the more counterintuitive findings surfaced in the agency ranking category.
Mid-tier articles outscored top-tier articles on fact checks by nearly 30 percentage points (85.2% vs. 56.2%).
Top-tier rankings incorporate interviews, commentary, and proprietary claims, all of which introduce unverifiable assertions by design.
Quality tier reflects investment while credibility reflects source verifiability. They are different signals.
Fact accuracy also operated independently of readability. Dense content wasn't more factual, and accessible content wasn't less rigorous. Credibility stands on its own axis.
For publishers, the choice isn’t how to improve the score, it’s whether the content format supports verifiability in the first place.
Citation is the New Traffic
One of the more structural findings emerged outside the quality metrics themselves.
AI citation activity is now measurable across ChatGPT, Google AI Overviews, Gemini, Perplexity, and Microsoft Copilot.
Four of those five platforms produced zero tracked citation activity in our dataset twelve months earlier.
The question for publishers isn't whether AI platforms are citing their content.
It's whether anyone on the team can tell you how many times, across which platforms, and for which articles.
This matters because the citation surface reflects something different than traffic.
A citation indicates that an AI system selected that source to support its response.
That's a quality signal worth tracking on its own terms, not just as a downstream proxy for traffic.
The signal is real, though the measurement infrastructure is still catching up.
Platforms use different citation conventions, referral tracking varies across tools, and first-year data is directional rather than trend-verified.
“Optimization still drives acquisition, but traffic no longer tells the whole story.
“Citation visibility across AI platforms is becoming an additional layer publishers need to track,” says Jonathan Gillham, CEO of Originality.ai.
What Publishers Should Take From This
The ACQI tested how different quality systems relate to performance in an AI-mediated search environment.
The findings suggest they do, but not in the way many teams assume.
AI platforms now represent a measurable citation layer across major assistants.
That layer reflects selection rather than clicks, and can be tracked alongside organic traffic.
Within the quality stack the three systems behave independently.
Optimization drives acquisition and still correlates with traffic, with gains concentrated at the top.
Readability governs comprehension and remains the most controllable lever for most publishers.
Fact accuracy governs credibility and reflects the strength of underlying source material rather than editorial effort alone.
No single score captures all three systems. A composite number can mask which one is underperforming.
Three implications follow:
- Instrument AI citation exposure alongside organic traffic.
- Treat readability as an editing discipline, not a stylistic preference.
- Benchmark quality by system (acquisition, comprehension, credibility), not by section or tier.
This looks less like a cycle and more like a structural change.
Traffic remains critical, but it no longer reflects the full picture.
Publishers who measure only traffic are measuring only part of the system.
Content quality is not a single metric. It operates as parallel systems that influence different performance outcomes.
Measuring them as one score obscures how content really performs.
Want to improve your visibility in AI search and stay competitive in 2026? Check out our selection of the Best SEO Agencies to help you build authority, structure your content, and optimize for GEO and AEO.
