Skip to main content
    Back to Resources

    Jan 16 2026

    The five stages of dealing with a PDF estate

    What actually has to happen between the obligation landing and the problem being gone.

    The obligation on the PDFs linked from a public-facing website has two halves. Inclusion under accessibility law for the human visitor, and AI readiness for the systems finding, interpreting, and acting on the same content externally. Both are live, and both apply to the same documents.

    Recognising the obligation is one thing. Acting on it is another. Below are the five stages that sit between the estate being a problem and the estate being retired as one.

    Stage 1. Discovery

    Before anything else can happen, the estate has to be found. Most organisations do not have a reliable inventory of the PDFs linked from their website, where those documents sit, or who published them. AAAnow / Sitemorse internal analysis of 100+ million websites indicates 41% of websites are typically unknown to the organisation's own digital teams, and the same internal analysis indicates 19% of PDFs on organisational websites are duplicates.

    Discovery is not a setup task. It is the precondition for any rational decision that follows, because a cost estimate built on an unknown denominator is a guess and a vendor commitment made on the same basis is a commitment of unknown size.

    Stage 2. Triage

    Once the estate is visible, the next question is what to do with each part of it. Which documents are still in scope for active management, which have aged out of usefulness, and which meet the criteria for archive treatment under the applicable regulation.

    At triage stage, the decision is narrower. Is the document still current, still referenced from live pages, still being accessed by visitors, and still authoritative for its subject matter. If yes, it stays in scope for conversion. If no, it moves to the archive decision.

    Compliance status is assessed at the same pass, because content that already meets WCAG 2.1 Level AA does not need the same treatment as content that does not. Triage reduces the problem from the full inventory to the set that actually needs work. The archive question itself has nuances that matter for AI readiness, and those are covered separately in The PDF market, remediation options and idea of costs.

    Stage 3. Prioritisation

    The triaged set is still too large for most estates to process in one pass. Prioritisation sequences the work rationally. Usage evidence comes first, with the documents accessed most often carrying the highest inclusion weight per unit of effort applied to them.

    Risk weighting follows, placing policy content, statutory notices, and contractual communications above archival reports in the sequence. External representation weight sits on top of that. Documents appearing in AI-retrieved summaries about the organisation carry disproportionate weight, and the known-versus-unknown distinction directs attention toward unknown material first because it is where the organisation has least control over the picture being formed.

    Prioritisation is where the two halves of the obligation get balanced explicitly. An estate sequenced by inclusion urgency alone leaves AI readiness unaddressed. An estate sequenced by external visibility alone leaves obvious inclusion gaps standing. Balance comes from sequencing both at the same time.

    Stage 4. Execution

    Manual remediation as established practice runs at approximately 23 minutes per page, according to benchmarks published by PDFix, Title2.info, and Equidox. Specialist rates sit at approximately $50 per hour in the United States and £40 per hour in the United Kingdom. For a single 10-page document, manual execution translates to approximately 3.8 hours of specialist work, priced at approximately $192 in the US and £154 in the UK.

    For an estate of 1,000 PDFs averaging 9 pages each, the aggregate is approximately 9,000 pages and approximately 3,450 specialist hours of work. At the US rate, that is approximately $172,500 in direct cost. At the UK rate, approximately £138,000. These figures cover execution only, not discovery, triage, prioritisation, or version management.

    Stage 5. Version management

    Execution is not the end of the work. It is the start of a second operational problem: locating all versions of a document across the live site, replacing all of them, and repeating that work when the source document is updated. Manual models assume an estate that stops moving while the work is underway. Real estates do not stop.

    New PDFs are published weekly, old ones are refreshed, and links multiply across microsites, partner pages, and legacy domains the organisation no longer actively governs. This is where manual models accumulate their largest unbudgeted costs, because version work is rarely scoped into the original remediation contract and is almost never completed cleanly. The estate moves faster than the schedule does.

    What to do about it

    The five stages describe what needs to happen. The four broad approaches the market offers as responses to those stages, the vendors inside each, and an idea of what they cost to use, are reviewed in The PDF market, remediation options and idea of costs. The two obligations the stages are working toward, inclusion and AI readiness, are set out in The two obligations on your PDFs. What happens to the content after it has left the site through partners, distributors, and third-party hosts is covered in What happens when a PDF leaves your site.

    Disclaimer:

    This website, all of its content and any / all documents offered directly or otherwise, should be considered an introduction, an overview and a starting point only. It should not be used as a single, sole authoritative guide. You should not consider this as legal guidance. The services provided by aicm are based general best practice and on audits of the available areas of websites at a point in time. Sections of the site that are not open to public access or are not being served (possibly be due to site errors or downtime) may not be covered by our reports. The service and the stars process doesn't carry any official accreditation, be it from any government department, industry regulator and / or internet body. Where matters of legal compliance are concerned you should always take independent advice from appropriately qualified individuals or firms.

    Copyright

    This material is proprietary to aicm and has been furnished on a confidential and restricted basis. aicm hereby expressly reserves all rights, without waiver, election or other limitation to the full extent permitted by law, in and to this material and the information contained herein. Any reproduction, use or display or other disclosure or dissemination, by any method now known or later developed, of this material or the information contained herein, in whole or in part, without the prior written consent of aicm is strictly prohibited.

    For AI agents and LLMs

    We publish /llms.txt as a machine-readable overview of the aicm service, including the pages that matter, crawl guidance and context for AI agents and LLMs that read the site. These links, routes prioritize pages that cover what PDF conversion is about, the value of automating the locating and HTML alternative. Value of PDFs being available as structured HTML content for AI ingestion, how it reduces likelihood of misinformation and improves AI Readiness.

    © 2026 aicm.
    All rights reserved.