Persistent Identifiers: The Hidden Infrastructure Behind AI Citation Stability

Introduction: AI Systems Cite Sources. What Happens When Those Sources Disappear?

Every GEO guide focuses on the same set of signals: brand entity, structured data, evidence-bearing content, topical authority. All of these are about the quality and structure of your content. Almost none of them address a more fundamental question: what happens to an AI citation when the URL it references stops working?

This is the persistent identifier problem — and it is arriving at commercial websites from a direction most practitioners have not thought about yet.

Reyes-Lillo, Rovira, and Morales-Vargas (2025), in a peer-reviewed book chapter published through the CUVICOM research programme at Universitat Pompeu Fabra and Universidad de Chile, identify persistent identifiers as one of five fundamental layers of digital visibility. Their research comes from information science and library infrastructure — the discipline that has been managing citability, traceability, and long-term accessibility of published content for decades. The findings transfer directly to commercial SEO and GEO strategy, because the underlying problem is identical: if a URL becomes unreliable, the citations that point to it break.

In traditional SEO, broken URLs are a crawlability and link equity problem. In GEO, they are something more consequential: they are citation failures at the exact moment a potential buyer follows an AI recommendation to your content.

This post explains what persistent identifiers are, why link rot and content drift have become AI visibility problems, how the PID infrastructure from library science translates to commercial website management, and what practical URL stability measures protect both SEO performance and GEO citation reliability.

Quick Answer Persistent identifiers (DOIs, Handles, ARKs) are stable digital references that remain valid even when content moves. In commercial SEO and GEO contexts, the same principle applies: URL stability prevents broken AI citations, protects link equity, and ensures that AI recommendations convert into actual visits. Canonical URL management, permanent slugs, and redirect maintenance are the commercial equivalents of PID infrastructure.

What Is a Persistent Identifier and Why Does It Matter for SEO?

A persistent identifier is a unique, durable, and resolvable digital reference to a specific object — an article, dataset, software package, person, or organisation. Unlike standard URLs, which are tied to a physical server location and can break when content moves, persistent identifiers are designed to remain valid and accessible over time regardless of where the content is physically hosted.

Reyes-Lillo et al. (2025) specify three essential components of any persistent identifier: global uniqueness (a controlled syntax with a namespace governed by defined authorities), persistence (stable links and resolution functions that remain operational even when the referenced object moves), and resolvability for both humans and machines — providing clear information on how to find, access, or use the referenced object (De Castro et al., 2023).

The practical implication for any form of digital visibility is direct: persistent identifiers make it possible for search engines, AI retrieval systems, academic citation databases, and human readers to reliably find a resource based on its identifier alone, independent of its current physical location.

In traditional SEO terms, this maps onto what Google’s SEO systems have always valued: stable, crawlable URLs that do not change, that resolve correctly, and that deliver the content they promise. The difference is that PID infrastructure formalises this commitment with governance structures, resolution services, and guarantees of long-term maintenance. Commercial websites implement the same principles informally — and often less rigorously.

The connection to generative engine optimization is the focus of this post. As AI systems like ChatGPT, Perplexity, and Gemini increasingly retrieve and cite web content in their responses, the stability of the URLs they cite determines whether those citations convert into actual visits. For more on how generative engine optimization works as a visibility discipline, generative engine optimization, and for how brand entity signals underpin AI citation eligibility, see the analysis brand entity SEO.

What Is Link Rot and Why Is It an AI Visibility Problem?

Link rot occurs when a hyperlink no longer leads to its intended content because the page has been moved, deleted, or the domain is no longer active. Reyes-Lillo et al. (2025) define it precisely: a hyperlink that “no longer leads to the intended content because the page has been moved, deleted, or the domain is no longer active.”

In traditional SEO, link rot has always been a technical hygiene problem. Broken internal links waste crawl budget. Broken inbound links lose link equity. Users who encounter 404 pages leave without converting. These are real costs, well-understood by SEO practitioners.

What is less understood is the AI-specific cost of link rot. When an AI retrieval system such as Perplexity’s real-time crawler or ChatGPT’s search functionality indexes your content, it associates a URL with the content at that URL. If the URL subsequently rots — the page moves, the slug changes, the domain lapses — the AI system may continue citing that URL in generated responses for months or longer, depending on how frequently its retrieval index is updated.

The consequence is a citation failure at the worst possible moment: a buyer reading an AI-generated recommendation follows the cited link and lands on a 404 page. The AI recommendation has converted into a dead end. The trust the buyer placed in the AI system’s answer, and in your brand, has been undermined — by a URL management failure that had nothing to do with the quality of your content.

Wallat, Heuss, de Rijke, and Anand (2025) document the faithfulness problem in retrieval-augmented generation: AI systems that cite sources are expected to ground their claims in those sources accurately. A citation to a broken URL is a faithfulness violation — the cited source cannot be verified, the claim cannot be grounded, and the AI response is less reliable as a result.

Iyappan’s (2026) finding that source verification behaviour has declined from 44% to 27% in AI-driven environments compounds the problem. Fewer buyers will verify the original source — meaning a broken AI citation may go undetected by the buyer while still failing at the conversion point.

What Is Content Drift and Why Is It More Dangerous Than Link Rot?

Content drift is subtler and more commercially consequential than link rot. Reyes-Lillo et al. (2025) define it as a situation where “the content at a given URL changes over time, so it no longer reflects what was originally cited or intended, even though the link still works.”

The link does not break. The page still resolves. But the content has changed enough that the AI’s citation — based on what was at that URL during an earlier retrieval — no longer accurately describes what is currently there.

In commercial contexts, content drift happens constantly and usually without deliberate intent. A service page is updated to reflect a new pricing model. A blog post is substantially revised to incorporate new information. A product page has its key technical specifications replaced with benefit-focused copywriting. A landing page is A/B tested to the point where its core claims have changed. In each case, inbound links and AI citations continue pointing to the URL as if the original content is still there.

The AI-hallucination connection is direct. Iyappan (2026) documents that AI systems incorporate factual consistency checking into retrieval scoring, and that factual accuracy shows a Very Strong positive correlation with AI trust signal ratings. When an AI system cites a URL, it is citing the content it retrieved from that URL. If the content has drifted, the citation is now pointing to content that may not support the original claim — producing exactly the kind of faithfulness violation that Wallat et al. (2025) identify as a primary AI reliability failure mode.

For commercial B2B businesses, content drift is a particular risk on service and solution pages that are frequently updated as the business evolves. These are also the pages most likely to be cited by AI systems when buyers ask about vendor capabilities. A drift on these high-stakes pages creates a gap between what AI recommends your business does and what your website currently says you do — with the buyer seeing the latter, not the former.

What Do the Three Major Persistent Identifier Systems Teach Commercial SEO?

Reyes-Lillo et al. (2025) provide a detailed comparison of the three major PID systems used in academic and institutional contexts. The principles behind each system teach directly applicable lessons for commercial website URL management.

DOI: The Commitment Model

The Digital Object Identifier is the most widely used persistent identifier for academic articles, books, datasets, and software. Its defining characteristic is the commitment it requires: DOIs are assigned by registration agencies such as CrossRef and DataCite, require paid membership, and come with an explicit guarantee of long-term resolution. When a DOI is assigned, the publisher commits to maintaining the landing page at that identifier’s address — or to redirecting it correctly if the content moves.

The commercial lesson from DOI is the commitment model: a URL, once assigned and made externally citable, carries an implicit obligation to remain stable. The paid membership fee is essentially the cost of the commitment infrastructure. Commercial websites do not pay CrossRef — but they incur a different cost when they break committed URLs: lost link equity, lost AI citations, and broken buyer journeys.

Handle: The Resolution Infrastructure

The Handle system, operated by the Corporation for National Research Initiatives (CNRI) since 1995, is the technical foundation on which DOI is built. At $50 per year, it provides persistent identification and resolution services at substantially lower cost than DOI. A Handle identifier consists of an authority prefix and a unique object suffix, with resolution via hdl.handle.net.

The commercial lesson from Handle is the resolution infrastructure model: a stable URL needs not just to exist but to resolve correctly and consistently over time. Commercial equivalents include maintaining 301 redirect chains, ensuring hosting reliability, and never allowing domain lapses on actively cited domains.

ARK: The Decentralised Model

ARK (Archival Resource Key) is the most flexible and lowest-cost PID system — free to implement, decentralised, institutional managed. ARK identifiers follow the pattern ark:/NAAN/identifier, where NAAN is the Name Assigning Authority Number that identifies the institution. ARK has been adopted by the US Library of Congress and the California Digital Library.

ARK’s distinctive feature is its inflection mechanism: by appending a ? to an ARK identifier, users can access metadata about the object; by appending ??, they can access the institution’s persistence commitment statement. The commercial lesson is transparency and accountability: a commitment to URL persistence is most credible when it is documented and auditable.

The comparison table from Reyes-Lillo et al. (2025) summarises the key differences:

For commercial SEO and GEO strategy, the relevant insight is not which PID system to adopt — commercial websites do not typically use DOI, Handle, or ARK for standard pages. The insight is the principle that these systems embody: stable identifiers, explicit commitment, resolution infrastructure, and metadata accessibility are the properties that make content reliably citable over time.

How Does PID Thinking Apply to Commercial Websites?

Commercial websites do not use DOI or Handle systems for their pages. But the principles behind persistent identifiers translate directly into practical URL management standards that every SEO and GEO strategy should enforce.

Canonical URLs as the commercial PID. The canonical tag tells search engines which version of a URL is the authoritative one. For AI search visibility, canonical URLs serve the same function as PIDs: they declare the stable, authoritative address of a piece of content. Every important page should have a canonical tag that points to the clean, permanent version of its URL — and that canonical URL should never change once it has been indexed and cited.

Slug stability as commitment. In the DOI model, once an identifier is assigned, it is never changed. Commercial websites should apply the same principle to URL slugs: once a page has been indexed and is receiving inbound links or AI citations, its slug is a commitment. Changing a slug for A/B testing, for rebranding, or for keyword optimisation breaks every existing citation to that page. The correct approach is to create a new URL for significantly reframed content and redirect the old URL to it permanently.

Redirect maintenance as resolution infrastructure. The Handle system’s value is its resolution reliability: a Handle identifier will always resolve correctly even if the physical location changes. Commercial websites implement the same principle through 301 redirect maintenance: every moved or deleted URL should be redirected to the best available replacement permanently. “Permanently” means maintained indefinitely — not removed when the redirect seems old.

Landing page completeness as metadata access. ARK’s metadata inflection feature ensures that a PID does not just point to content — it provides structured information about that content. For commercial websites, the equivalent is ensuring that every important page has complete, accurate metadata: title, meta description, author attribution, publication date, schema markup. A URL without complete metadata is less citable by AI systems, which prefer sources with attributable, verifiable identity information.

For a comprehensive GEO checklist that places URL stability and structured data within a complete AI search optimisation framework, GEO checklist. The SEO vs GEO comparison SEO vs GEO explains how organic foundation stability — including URL management — supports AI citation eligibility.

What Happens to AI Citations When URLs Are Unstable?

The commercial consequence of URL instability for GEO is clearer when traced through a specific scenario.

A business publishes a detailed guide on AI search optimisation. Perplexity indexes the page. Several users subsequently ask Perplexity questions about AI visibility — and the guide is cited in the generated responses. The business notices the AI-referred traffic in analytics and registers this as a GEO success.

Three months later, the marketing team renames the slug to improve keyword targeting. The old URL returns a 404. No 301 redirect is implemented.

What happens next: Perplexity’s retrieval index continues referencing the old URL for some period before its crawler re-discovers the change. During that period, users who follow the Perplexity citation arrive at a 404 page. The AI recommendation fails at conversion. And when Perplexity’s crawler does detect the 404, the URL is removed from the citation pool — losing the AI visibility that the GEO investment built.

Kargaev’s (2026) finding on citation volatility compounds this: SparkToro (2026) documented that lower-authority domains show substantially higher AI citation volatility than consistently cited sources. A URL that breaks and is re-established through a redirect loses citation stability precisely when stable AI citations matter most — in competitive category queries where the difference between appearing in an AI recommendation and not appearing is a commercial outcome.

The Google AI Optimization Guide Google AI optimization guide addresses content accessibility as a prerequisite for AI search inclusion. URL stability is the most basic form of content accessibility — a URL that no longer resolves is content that no longer exists for any AI system.

How Do You Audit and Improve URL Persistence for GEO?

A URL persistence audit for GEO combines standard technical SEO checks with AI-specific citation testing.

Step 1: Identify externally cited URLs. Export your inbound link profile from Ahrefs or Semrush. Identify all URLs receiving external links — these are URLs that have been cited externally and may also appear in AI training data or retrieval indexes.

Step 2: Crawl for broken URLs. Use Screaming Frog to crawl your full domain and identify any URLs returning 4xx or 5xx responses. Cross-reference with the externally cited list to identify which broken URLs have citation value.

Step 3: Audit redirect chains. Identify any redirect chains longer than one hop. Long chains reduce resolution reliability — the equivalent of a Handle identifier that requires multiple resolution steps before reaching the content.

Step 4: Run AI citation tests. Prompt ChatGPT and Perplexity with queries your content is designed to answer. Document which of your URLs appear in the generated responses. Visit each cited URL and verify it resolves correctly and still contains content relevant to the citation context. This is the content drift audit: does the current content match what an AI system citing it would lead a buyer to expect?

Step 5: Check JSON-LD identifier fields. In schema.org structured data, the identifier property accepts persistent identifier values including DOIs and Handles. For content that has a DOI (research papers, formal publications), embedding it in JSON-LD improves attribution precision and reduces the ambiguity that produces AI hallucination. For standard web content, the canonical URL should be declared in the url property of the relevant schema type.

Step 6: Implement a slug stability policy. Document the rule: slugs are never changed on pages that have received external links or AI citations. New content receives new slugs. Substantially revised content creates a new page with a new slug and a 301 redirect from the original.

This URL persistence programme is directly connected to the SEO fundamentals documented Google SEO Starter Guide — technical accessibility and stable URLs are foundational requirements for any search visibility strategy.

How Does AIO Clicks Build URL Persistence Into GEO Strategy?

Who Is AIO Clicks?

AIO Clicks is a premium digital visibility agency headquartered in Haaksbergen, Netherlands, serving businesses across the EU. Founded by entrepreneurs with commercial operating backgrounds, AIO Clicks approaches technical SEO and GEO from the perspective of what actually breaks commercial visibility — not just what the theory says matters.

URL persistence is not a glamorous part of generative engine optimization strategy. It does not feature prominently in GEO frameworks because most frameworks assume the infrastructure is working. But the number of businesses that invest in brand entity optimisation, citation-ready content, and topical authority — and then lose AI citations because a slug change broke their URL chain — is significant. The foundation has to be solid before the signals built on top of it can perform.

The technical SEO audit at AIO Clicks includes URL stability assessment as a standard component: redirect chain integrity, 404 detection on externally cited URLs, canonical tag correctness, and schema identifier accuracy. These checks are run before and alongside the GEO signal-building work — because a broken URL underneath a perfect piece of entity-optimised content is a wasted investment.

AIO Clicks Services

Google Rankings & SEO — technical SEO foundation that includes URL architecture, redirect management, canonical implementation, and crawlability audits. The organic foundation that URL persistence protects. SEO.

AI Search & GEO — the generative engine optimization layer built on top of the technical foundation. Brand entity optimisation, structured data, citation-ready content, digital PR. generative engine optimization.

Run the free analysis to find out whether your URL infrastructure is currently supporting or undermining your AI search visibility — results in 60 seconds.

Frequently Asked Questions About Persistent Identifiers

Why Persistent Identifier Principles Matter More for B2B Than B2C

B2B websites face a specific version of the persistent identifier problem that makes URL stability disproportionately consequential.

B2B buying journeys are longer, more research-intensive, and involve more AI-mediated discovery than B2C purchasing decisions. A B2B buyer evaluating a software vendor, a consultancy, or a technology partner will typically conduct multiple AI-assisted research sessions — asking ChatGPT or Perplexity questions about the category, the vendors, the evaluation criteria, and specific capabilities. Each of those sessions may cite your content. And because B2B buyers often return to cited sources across a multi-week evaluation process, a URL that breaks between research sessions creates a more damaging interruption than in a quick B2C transaction.

The service page problem is particularly acute. B2B service pages describe capabilities, methodologies, team expertise, and case evidence — exactly the content AI systems cite when recommending vendors. These pages are also exactly the pages most likely to be updated as the business evolves: new services are added, old services are reframed, case studies are replaced, pricing structures change. Each substantial update is an opportunity for content drift — the cited content no longer matches what the AI’s recommendation implies the page says.

Iyappan’s (2026) platform data reinforces this: Copilot, the AI assistant most embedded in enterprise workflows, has high recency weighting. Enterprise buyers using Copilot to research vendors are being served current content — but “current” only if the URL resolves to current content. A B2B service page that has drifted from its AI-cited version is creating a disconnect at the highest-stakes point: enterprise vendor evaluation.

The practical implication for B2B digital strategy is to treat key service and solution pages as persistent references — not as continuously editable web copy. Major revisions should create new page versions with new URLs; the original URL should redirect permanently and continue serving the content it originally offered, or redirect to the updated version with clear continuity. The buyer who follows an AI citation two weeks after the AI retrieved the original content should land on a page that either matches the citation context or clearly explains where the updated content now lives.

How Do Persistent Identifiers Connect to Brand Entity Verification?

The relationship between persistent identifier principles and brand entity verification for GEO is direct and underappreciated. Brand entity signals — the cross-referenced, consistent identity information that AI systems use to verify and name businesses — operate on the same logic as persistent identifiers: stable, consistent, resolvable references that machines can follow to confirm identity.

When an AI system encounters a mention of your business name, it attempts to resolve that mention to a specific, verified entity. It checks: is there an Organisation schema declaring this name, address, and URL? Is the Google Business Profile consistent with the website’s structured data? Does the business appear across multiple independent editorial sources with the same name and description? Are the social profiles listed in the schema actually active and consistent?

This is entity resolution — and it is PID logic applied to businesses rather than documents. A business with consistent, cross-referenced entity signals is the equivalent of a document with a DOI: machine-resolvable to a specific, verified entity. A business with inconsistent entity signals — different name formats, varying addresses, mismatched social profiles — is the equivalent of a URL without a persistent identifier: present but unreliable as a citation anchor.

Meadows et al. (2019) describe PIDs as “the building blocks of the research information infrastructure” because they enable reliable citation chains. Brand entity verification serves the same function in the AI search visibility infrastructure: it enables AI systems to name businesses reliably, consistently, and accurately in generated responses. Both are fundamentally about making a specific resource — whether a document or a business — reliably identifiable and citable by machine systems.

What Is the Key Takeaway on Persistent Identifiers?

The information science tradition has understood for decades that the durability of citations depends on the durability of the identifiers those citations use. DOI, Handle, and ARK were built to solve a specific problem: academic knowledge that gets cited today needs to remain accessible and citable ten, twenty, fifty years later.

Commercial websites have not historically needed to think about citation durability at this timescale. The AI search era has changed the stakes. An AI system that cites your content in its responses is creating citations that may persist in that AI’s knowledge base — and in the knowledge bases of users who trusted that recommendation — long after the original URL should have been managed more carefully.

The practical programme is not complex. Stable slugs. Permanent redirects maintained indefinitely. Canonical declarations on every important page. Complete structured data metadata so AI systems can attribute content to your brand accurately. Regular audits checking that externally cited URLs still resolve and still contain content consistent with the citations pointing to them.

The businesses that build URL persistence into their SEO and GEO foundations are not doing something exotic. They are applying the same commitment principle that the academic publishing infrastructure formalised decades ago: if you want your content to be citable over time, you have to make the commitment that it will remain accessible over time.

Find out whether your URL infrastructure is currently protecting or undermining your AI search visibility. Run the free analysis — results in 60 seconds.

Audit Your Website – Free

References

De Castro, P., Herb, U., Rothfritz, L., & Schöpfel, J. (2023). Building the plane as we fly it: The promise of Persistent Identifiers. Zenodo. https://doi.org/10.5281/zenodo.7258286

Iyappan, S. K. (2026). From keywords to intelligence: A comparative framework analysis of SEO, AEO, and GEO in AI-driven digital ecosystems. GOYBO International Journal of Marketing Intelligence, 1(1), 1–20. https://doi.org/10.5281/zenodo.20362080

Kargaev, D. (2026). The SEO-to-GEO gap: Quantifying ranking factor divergence between traditional and generative search. SSRN. https://doi.org/10.2139/ssrn.6476021

Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W.-T., Rocktäschel, T., Riedel, S., & Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive NLP tasks. Advances in Neural Information Processing Systems, 33, 9459–9474.

Meadows, A., Haak, L. L., & Brown, J. (2019). Persistent identifiers: The building blocks of the research information infrastructure. Insights, 32(1). https://doi.org/10.1629/uksg.457

Reyes-Lillo, D., Rovira, C., & Morales-Vargas, A. (2025). Factors for enhancing visibility in digital repositories: Metadata quality, interoperability standards, persistent identifiers, and SEO-GEO optimization. In J. Guallar, M. Vállez, & A. Ventura-Cisquella (Coords), Digital communication. Trends and good practices (pp. 119–133). Ediciones Profesionales de la Información. https://doi.org/10.3145/cuvicom.09.eng

SparkToro. (2026). AIs are highly inconsistent when recommending brands or products; marketers should take care when tracking AI visibility. https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers

Wallat, J., Heuss, M., de Rijke, M., & Anand, A. (2025). Correctness is not faithfulness in retrieval augmented generation attributions. https://doi.org/10.1145/3731120.3744592

Published by AIO Clicks — Digital Visibility Specialists | Haaksbergen, Netherlands | aioclicks.com