Four weeks out: five areas, and they are not equal
The ADS Body of Knowledge publishes its weighting across five knowledge areas, and the shape of it changes how you should spend the month. Creating documents is 25 per cent. Remediating them is 25. Auditing and testing is another 25. Planning and training is 15, and policy, advocacy and advising is 10.
Three quarters of the exam is hands-on work, split evenly three ways. The remaining quarter is organisational: process, training, procurement, advice. That last quarter is where most candidates lose marks, and the reason is simple. If your job is remediation, you have done thousands of tag trees and almost never written a document accessibility policy, so the three big areas feel covered and the small ones feel like somebody else's job.
They are a quarter of the paper. Treat them that way.
- Week one. Creating. The standards landscape, writing inclusive content, type and colour, structure through styles and lists and tables, universal design and export.
- Week two. Remediating. Alternative text and artifacts, correcting structure, the PDF tag tree and EPUB, remediating in the authoring format, and where OCR belongs in the order of operations.
- Week three. Auditing and testing. The order of an audit, automated checkers and how they differ, manual inspection of the tag tree, and testing with assistive technology.
- Week four, first half. Planning, training, workflow, policy and procurement language. Do not skip this. It is a quarter of the exam and it is the part you have never had to revise.
- Week four, second half. No new reading. Timed mocks only.
Against the A11ytek ADS course, that is lessons 1 to 5, then 6 to 10, then 11 to 14, then 15 to 17.
Remediate something every day
ADS questions describe documents. A stem will give you a table that reads wrongly, or a scanned file, or an export that lost its headings, and ask what you do about it. Reading about tag trees does not build the instinct those questions test.
So keep a few deliberately broken files on your desktop and work on one a day. Tag a table with merged header cells. Set reading order on a two-column layout and then listen to it. Export a Word document with real styles and then export one with hand-formatted bold text, and compare the tag trees. Run the same file through two different checkers and note where they disagree, because that disagreement is itself an exam topic.
A useful target at the end of each knowledge area is a practice score in the low nineties before moving on. That is above any pass mark on purpose. Document questions are easy to answer correctly for the wrong reason, because the right fix often sits beside two fixes that would be right for a different file format. A margin above the line is how you find the ones you guessed.
One week out: four kinds of wrong
By this point the only activity is full mocks under real conditions. Seventy-five questions, two hours, closed book. That is roughly ninety-six seconds each, which is comfortable until a stem describes a document you have to picture before you can answer.
Sort every miss into one of four piles, because each needs a different response.
- You did not know it. A real gap. Go back to the lesson rather than the card, because what is usually missing is the relationship between two standards rather than a single fact.
- You confused one standard with another. The signature ADS mistake. WCAG, PDF/UA, the Matterhorn Protocol and ISO 32000-1 all govern different things, and a surprising number of questions turn on knowing which one answers the question in front of you. The first recall list below exists for this.
- You gave the right fix in the wrong place. Remediating in the authoring application and remediating in the final file are different jobs with different tools, and several questions check that you know which situation calls for which.
- You knew it and second-guessed yourself. Nothing to study. More timed practice fixes it, and that is the week's work anyway.
If the middle two piles dominate you are in better shape than the score suggests, because both are recall rather than understanding. If the first pile dominates, take a reading day back out of the week.
The last two days: the parts that are pure memory
Judgement cannot be crammed. Exact names, numbers and boundaries can, and that is where marks leak. Read each list, cover it, write it out, check.
Which standard answers which question
- ISO 32000-1:2008 is PDF itself, the file format. It is not an accessibility standard.
- PDF/UA-1 is ISO 14289-1, the accessibility standard for PDF. It says what a conformant file must contain.
- The Matterhorn Protocol is not a standard at all. It is the checklist for verifying PDF/UA, organised into 31 checkpoints, and roughly half of its failure conditions can only be judged by a human. The version cited by the ADS Body of Knowledge is 1.02.
- WCAG 2.1 is the W3C guidance the Body of Knowledge works from, and the PDF Techniques are the W3C's documented ways of satisfying WCAG in a PDF.
- A file can satisfy PDF/UA and still fail WCAG, because PDF/UA says nothing about contrast ratios or reading level. The two overlap; neither contains the other.
Tags, and the ones that get mixed up
- A list is L containing LI, each holding an optional Lbl for the bullet or number and an LBody for the content.
- A table is Table with TR rows and TH or TD cells, optionally grouped as THead, TBody and TFoot. Header cells need a Scope of Row or Column, and complex tables need explicit IDs and Headers.
- A meaningful image is a Figure with alternative text. A decorative one becomes an Artifact, which removes it from the tag tree entirely. Empty alternative text on a Figure is not the same thing.
- Alternate text describes an image. Actual text replaces the characters themselves, for a dropped cap or a decorative glyph.
- Headings are H1 to H6 and must not skip levels. Bookmarks are navigation only, and they never substitute for heading structure.
- The tag tree is what assistive technology follows. The content or order panel is a separate thing, and fixing one does not fix the other.
Document-level settings that are easy marks
- The document title must be set in the properties, and the initial view must be set to display the title rather than the file name.
- The document language must be set, and any passage in a different language must be marked at that passage.
- Tab order must be set to use document structure, on every page.
- The file must be a tagged PDF and must not be an image of text, which is why OCR comes before tagging. Tag a scan first and you have tagged a picture.
- Security settings must not block assistive technology from extracting text.
What no checker can tell you
- Whether alternative text is accurate or useful.
- Whether the reading order matches the visual order.
- Whether a heading is at the right level, or whether a heading is a heading at all.
- Whether the cells tagged as headers are the right cells.
- Whether colour alone is carrying meaning.
- Whether the language tag matches the language actually used.
- Different checkers test different things, so a clean report from one is evidence rather than conformance.
The night before
Do not open a topic you have never studied. A standard you meet for the first time that evening will not settle by morning, and finding the gap costs more in nerve than the question is worth in marks.
Twenty minutes on the recall lists is plenty. Then stop.
Exam morning
If you are sitting remotely, test the proctoring software on an earlier day. Clear your desk, because a remote proctor can ask you to show the room and anything within reach becomes a discussion. Have your identification to hand.
Keep moving once you start. Flag anything that makes you hesitate and return to it. Four minutes on one tagging question is three other questions you skimmed.
Read the stem to the end, and notice which format it is describing. The same question has different answers for a PDF, a Word file and a presentation, and the format is often named once, early, in a sentence that is easy to skim past.
When two answers both look right, ask which one fixes the cause rather than the symptom. Adding alternative text to an image that should have been an artifact is the classic trap, and it appears in several shapes.
Additional time is available as an accommodation, and arranging it is an ordinary part of the process.
Three things not to do in the final week
- Do not start a new question bank. At this stage a fresh one mostly shows you what you have not covered, which is discouraging and not actionable. Finish the one you have.
- Do not chase a pass percentage. IAAP uses scaled scoring and publishes no fixed pass mark, so any number you have heard is a guess. Knowing the Body of Knowledge is the only approach that survives that.
- Do not write off planning, training and policy. Together they are a quarter of the exam, which is about nineteen questions, and they are the easiest quarter to lift in two days because so much of the content is definitional.
Practise the way the exam is weighted
The A11ytek exam prep app holds a large bank of ADS questions with written explanations, and its mock draws from them weighted 25, 25, 25, 15 and 10 across the five knowledge areas, so a practice score is comparable with the real thing rather than with whatever mix you happen to pull.
One hundred questions across the three certifications are free, with no sign-in and no email address, if you want to see the format before committing to anything.
Start practising
Back to the ADS guide