We Tested 8 Press Release URL Paths. Here’s What Got Cited.

Brands pay to distribute press releases across high-authority news publications. The likes of Forbes, Reuters, and Fortune. And they’re still not showing up in AI search.

We ran the same content across eight different URL destinations to understand why.

What we found led us to build our newest service, AI Press Release.

Why press releases don’t get cited in AI search

AI retrieval systems use the structure of web addresses as a filter before evaluating content.

Every wire distribution service delivers content to the “press releases” section of the outlets it syndicates to, pages at addresses like /press-releases/ or /news-releases/. AI platforms have learned to treat content at those addresses as promotional and self-reported, screening them before the content itself is even evaluated.

Content in a site’s actual news section carries a different signal. Editorial review implied, research format expected. Citation rates follow accordingly.

Take Yahoo Finance. Its news section (finance.yahoo.com/news/) consistently appeared as a cited source across ChatGPT, Gemini, Perplexity, and Google AIO when we ran relevant queries.

The same content published in the press release sections of comparable outlets almost never appeared as a cited source across the same queries and the same four platforms. The only variable was which section of the website the content lived in.

It’s a gap that better writing doesn’t close. The path signal is set by where the content lives. Getting cited requires getting onto the right path in the first place, and that’s the structural problem AI Press Release is built to solve.

What the test showed across eight paths

We tested eight distinct URL destinations: the same content, published across eight different sections of eight different outlets. The citation behavior on each was then checked across four AI platforms — ChatGPT, Gemini, Perplexity, and Google AIO — against target query types aligned to the release content.

The results fell into three tiers:

Confirmed citation: Yahoo Finance /news/ and Business Insider /markets/. Cited by ChatGPT, Gemini, and Perplexity on brand and informational queries. These paths carry the editorial signal that AI retrieval systems reward.

Rare to never: /press-releases/ paths at Fortune, Reuters, and equivalent wire outlets. Comparable or higher domain authority than Yahoo Finance using the same content. When we ran the same queries through all four platforms, content on these paths almost never appeared as a cited source.

Mixed/conditional: Paths that showed platform-specific behavior, cited on one platform and filtered on others. Not reliable enough to engineer a product on.

The Business Insider /markets/ placement adds a second confirmed citation path for clients where the second citation source matters for their query type. It’s an add-on tier for a reason.

What a research announcement is and why the format matters

The path gets you in the door. The content format determines what the AI does with it once you’re there.

A press release, even a well-written one, is a promotional artifact with the brand as subject. Executive quotes, product launches, and maybe even company news. The brand is announcing something about itself.

A research announcement positions the brand as a publisher. The brand releases original data, analysis, or research findings. The brand is the author of the research, not the subject of the story.

That distinction matters for AI citation in a specific way.

AI platforms are answering informational queries. “What are switching costs for enterprise CRM software?” “What does the data show about mid-market freight brokerage margins?” They’re looking for content that addresses the informational query, not content that promotes a company.

The research-announcement format matches the query type, but the promotional format doesn’t.

Every release we produce is built to eight structural rules, each one backed by research on what AI retrieval systems reward. Methodology section with disclosed data sources. FAQ block with entities echoed in opening words. One stat per section minimum, attributed with publication date. The analyst voice throughout.

The structural rules exist because the format signal matters as much as the path signal. Getting onto Yahoo Finance /news/ with a press release dressed up as a research announcement is not the same as getting there with content engineered to the format that earns citation.

The two layers this product addresses

Most of the conversation about AI visibility treats it as a single system. But in reality, it’s two.

The retrieval layer is what AI platforms fetch in real time. When ChatGPT runs a web search on a query, it’s hitting Google, Bing, Brave API, etc., evaluating what it finds, and surfacing content from that retrieval.

Our service AI Press Release (AIPR) targets this directly. Yahoo Finance /news/ is a domain retrieval system surface. Fresh, dated, research-format content gets surfaced and cited.

The base model layer is different, though.

LLMs are trained on large text corpora, and that training determines what the model “knows” before any retrieval happens. Brand mentions on authoritative sites feed into future training cycles. Every placement contributes to the model’s understanding of what the brand is, what it does, and what it knows about.

Both layers matter for different timescales. Retrieval-layer impact is visible at, say, Day 7. Training-layer contribution compounds over time and becomes visible when the next model is trained and released.

AIPR addresses Layer 2 (Entity Depth) and Layer 4 (Informational Citation) of the Loganix AI Visibility Framework. In plain english: how well AI systems know your brand, and whether they cite your content when answering relevant questions.

Each release builds both simultaneously. The placement earns a retrieval citation now. The distribution across the cascade — PRNewswire, AP News, Benzinga, and downstream syndication — builds the network of authoritative-domain mentions that feed the training layer.

What we deliver and what tends to follow

Three buckets worth being clear about.

What we deliver. A research announcement release is placed on Yahoo Finance /news/ within 7 business days of approval. Distribution through PRNewswire and the syndication cascade. DA 90+ backlinks from the distribution network. Citation tracking at Day 7 post-publication for one-time releases, Day 30 per cycle for monthly clients. A Day 30 report with live URLs, citation results, and decay or growth data.

Placements come with a 12-month guarantee on the premium distribution cascade. If we can’t place, you don’t pay. If a placement gets pulled within 12 months, we remediate.

What tends to follow. Pilot testing showed 100% of placements achieved AI citation within 7 days for target query types. (Disclaimer: That’s what the data shows; it’s not what we warrant.) Citation behavior varies by query, platform, and indexing cycle. The mechanism is engineered. The outcome is observed.

What we don’t deliver. Category-level recommendations. “Best [product] 2026” queries are Layer 3 territory (our service AI Brand Links handles that). Guaranteed rankings. Guaranteed traffic volumes. Specific citation events at specific times.

The product is built for brands that want to appear when someone asks an informational question aligned to their expertise. That’s the layer we’re in.

Who this is for and how to think about cadence

The product is intent-driven. If a brand has a research angle to publish and an AI visibility problem at L2 or L4, they’re a fit.

A research angle means proprietary data, novel analysis, or a genuine finding that the brand can publish under its name. Operational data that most brands have. Survey data. Analysis of trends in their vertical. If there’s no research artifact, there’s no release to write. That’s the qualifying question.

Cadence matters more than most clients expect coming in. A single release establishes a presence. Monthly releases compound it. Each additional release adds authoritative-domain mentions, builds entity depth, and provides retrieval systems with more content to surface for a broader set of queries.

For brands in the Strong tier (DR 40+ and an established footprint in AI search), citation lift tends to appear by Day 7–30 on the first release. For Foundation tier brands (DR under 20, minimal entity presence), the first release typically shows indexing confirmation. Citations follow with cadence, usually by releases three to six. Setting that expectation at intake is part of our process.

One product, one job

The legacy Loganix press release service isn’t going anywhere. It exists for wire distribution, journalist reach, regulatory announcements, and traditional brand PR. That’s a real job with real buyers.

AI Press Release is a different product for a different job. Same distribution network, different content format, different URL path, different purpose. Wire PR earns backlinks and journalist pickup. AIPR earns an AI citation and entity depth.

If a client wants traditional media coverage and DA backlinks from wire distribution, our press release and syndication services handle that. If they want to show up when ChatGPT answers questions about their category, this product does that. They coexist.

The data says to build here

The brands getting cited in AI search right now aren’t getting there by luck. They’re getting there because they published something on a path AI systems trust, in a format AI systems recognize as informational.

Most brands are still publishing promotional content on wire paths and wondering why they’re invisible in AI answers. The format mismatch is a placement mismatch. It doesn’t get fixed by writing better press releases.

This product exists because the test data said to build it.

AI Press Release is available now. Order here.

Use code AIPRLAUNCH for 10% off your first order. Valid through June 20, 2026.