The 2027/2028 Math: Why Manual-Only Remediation Cannot Meet the Extended Deadline

For chief financial officers and procurement leadership at universities approaching the IFR-extended ADA Title II compliance dates, the operational question is no longer whether to remediate but at what cost and on what timeline. The April 2026 Interim Final Rule extended the deadlines to April 26, 2027 (large public entities) and April 26, 2028 (smaller entities and special districts). That extension changes the arithmetic, but not enough to rescue any compliance strategy that depends on linear human review of a non-linear archive. The math is worth working through carefully, because the conclusion has direct procurement consequences: not for the question of which AI tool to buy, but for the prior question of whether AI assistance is optional. The defensibility framework these numbers feed into is laid out in the defensibility standard; this article narrows in on the budget and timeline arithmetic alone.

Three numbers determine the outcome. Archive size, per-page remediation cost, and reviewer throughput. Each is reasonably bounded by published industry data, and each is large enough that small adjustments do not change the conclusion.

Archive size at a mid-size US university

A mid-size US university LMS holds, conservatively, between 50,000 and 200,000 distinct documents. The number is not theoretical; institutional data from any large Canvas, Brightspace, Blackboard, or Moodle tenant produces a count in this range. A document is a course reading, a lecture slide deck, an assessment, a recorded lecture transcript, a syllabus, an exam paper, an answer key, a lab manual, a problem set, or a faculty handout. The count grows by 5,000 to 20,000 documents per academic year, depending on enrolment and course inventory size, before any new initiative adds further content.

Two ranges to anchor the discussion. A small liberal-arts institution with 2,000 students and 350 courses across 4 active years of LMS content has an archive in the 30,000-to-60,000 document range. A mid-size public R1 with 25,000 students and 3,500 courses across 6 active years has an archive in the 150,000-to-300,000 document range. The second case is the more common shape of an ADA Title II covered entity in the population-≥50,000 jurisdiction tier.

Average pages per document varies by content type. Course readings cluster around 15 to 30 pages per document; lecture slide decks at 20 to 40; assessments at 5 to 15; lab manuals at 30 to 80. A blended average of 15 pages per document is conservative on the low side. A 100,000-document archive at this average is a 1.5-million-page remediation problem.

Per-page manual remediation cost

The per-page cost of expert manual remediation is the variable most often understated in budget projections. Industry surveys and contracted-vendor pricing put expert reviewer remediation between $35 and $150 per page, with the variance driven by document complexity. A simple Word document with a clean structure remediates at the low end of that range. A scanned PDF requiring OCR, table tagging, alt-text generation, and reading-order correction remediates at the high end. STEM-heavy academic content, with equations, diagrams, and figures requiring content-aware description, sits at or above the high end.

A blended $80 per page is a defensible middle for a representative academic archive — neither punitive nor optimistic. Lower-end cost projections frequently appear in vendor pitches but assume document complexity that does not match real LMS content. For budget conversations, the $80 figure should be treated as the planning baseline.

A 1.5-million-page archive at $80 per page is a $120 million remediation budget. That number is larger than the annual operating budget of most US university accessibility programmes by an order of magnitude. It is larger than the total annual operating budget of many academic libraries. For a CFO, the budget number alone is enough to conclude that linear manual remediation against a serious archive is not a viable strategy.

Even halving each input — a 50,000-document archive, $40 per page, and 10 pages per document — produces a $20 million budget. For most institutions this remains an order of magnitude beyond the operational accessibility line item.

Throughput against the deadline

Cost is only one constraint. Throughput is the other. A trained accessibility specialist working on academic content can fully remediate a complex document in roughly half a day to a day, or three to six documents per standard work-week. For non-academic content the throughput is higher, but academic content is the load-bearing case.

Four reviewers working steadily produce roughly 1,000 fully remediated documents per year. A 100,000-document archive at this throughput is a 100-year backlog. Scaling reviewers helps linearly: forty reviewers produce 10,000 documents per year, a 10-year backlog. Forty trained accessibility specialists at full burdened cost is, at conservative US compensation levels, an annual personnel budget of $4 million to $6 million — reaching $40 million to $60 million across a decade, on top of the per-page remediation cost.

The deadline arithmetic that follows is straightforward. From May 2026 to April 2027 is approximately 11 months. From May 2026 to April 2028 is approximately 23 months. Even in the smaller-entity tier with the longer deadline, no institution staffing four to forty reviewers will reach a 100,000-document archive by the deadline. The institution that hires aggressively, doubles the team to eighty reviewers, and pays the $80-million-plus to compress the timeline still does not reach 100,000 documents in 23 months at three-to-six-per-week throughput.

The architecture problem is non-linear archive growth against linear human capacity. Each new semester adds documents at a rate that, by the time the existing backlog is processed, has produced a new backlog of comparable size. A programme that "finishes" in 2028 by force of expenditure has no answer for the new 2029 content the same programme has not yet remediated.

The procurement consequence for CFOs

The IFR's preamble does not extend deadlines because regulators expected universities would use the year for slower manual remediation. The preamble cites resource constraints and staffing limitations as reasons for the extension — and DOJ's own characterisation of the technology landscape, that "advanced technology, such as generative AI, does not yet reliably automate the remediation of inaccessible content at scale," confirms that the gap is real on both sides of the strategy choice. The extension acknowledges that neither manual-only nor AI-only achieves the standard at scale.

For CFOs, the procurement question is not "which approach do we fund?" It is "what is the configuration of AI assistance and human review that closes the gap before the deadline at a budget the institution can fund and a throughput the calendar permits?" Manual-only is foreclosed by the math. AI-only is foreclosed by the regulator's own statement and by the alt-text-accuracy failure modes covered in hallucinated alt text as a Section 504 risk. What remains is a hybrid configuration whose budget is dominated by the human-review step, whose throughput is multiplied by AI-generated initial fixes, and whose audit trail produces the documented record OCR resolution agreements consistently require.

That is the conclusion the math forces. Institutions evaluating their own programmes should run their archive size, per-page cost, and reviewer throughput numbers through the same arithmetic — and bring the result to procurement before the next budget cycle. The broader business case for accelerating remediation under the new deadlines is covered in the business case for accessibility remediation in higher education from earlier coverage; this article focuses on the deadline-math conclusion that makes that case unavoidable.