Workflow template for design and user experience
Document accessibility review
Before a designed report, deck or publication goes out, the designer gets a findings list on its headings, reading order, image descriptions, tables and contrast, with every check that could not be made named. The designer verifies each finding; the publication owner approves the fixes.
Listed under
Add it to a workflow in the Fully Powered Planner, beside the teams and roles that will run it. Book a fit call
The outcome
A finished document that more of its readers can actually use. Annual reports, client decks, press packs and campaign brochures are designed to be seen, and the parts a screen reader relies on are easy to miss: a heading that only looks like one, a photo the file hides, a sidebar read before the story it sits in, a table whose row labels are announced as plain cells, grey text a little too pale on a shaded panel.
This workflow template takes the designer's own lists of how the file is built and turns them into one findings list in five fixed areas: headings, reading order, alternative text, tables and contrast. Every finding names the page, the element, the rule it breaks and the rows it comes from, and ends with the same plain question for the designer and the publication owner.
It is just as clear about what it did not check. Whether an image description says the right thing, how merged table cells sound, text over a photograph and anything never measured are listed by name, and so are the things this review never covers. A clean result from an automated checker is never treated as a pass. The list certifies nothing and declares nothing accessible: a person verifies, and a person decides.
Who starts the work, and when
The designer who built the document, or the production lead, starts it when a designed document, deck or publication is ready to hand over or release. For an agency, that is before the client handoff; for a publisher or a charity, before the file goes to readers.
The designer prepares the lists from the file and verifies every finding. Where the team has an accessibility reviewer, that person verifies instead. The publication owner, on the client's side where there is a client, approves the corrections and decides when the document goes out.
What they supply
- The review settings: the document's title and file name, its kind (PDF, word-processor document, slide deck, or web page or email), the number of slides for a deck, the audience, the as-of date, who the designer and the publication owner are, and the contrast thresholds if the team has its own.
- A structure list: one row per element in the order the file presents it, saying how the file marks it, what it looks like on the page, its heading level, where the designer means it to be read (every element a screen reader reaches needs this, or its page's reading order cannot be checked), its first words and its alternative text.
- A table list: for each table, whether the header row and the row labels are marked, and whether it has merged cells.
- A contrast list: each text and background pair measured with a contrast tool, with the text size, whether it is bold and the ratio the tool reported.
- The checker report, if an automated checker was run: its items, pasted as they are.
The lists come from the views the document tool already offers and a contrast tool. Preparing the lists is the main effort, and it grows with the number of elements in the document. If time is short, start with the pages that carry images and tables, and run the review on those pages; it covers only what is supplied. No connection to the design tool, the file store or any account is needed, and no sign-in details are asked for. When a header in a list does not match the expected column names, the assistant stops and asks.
What comes out
One findings list in a fixed order: a summary with every count, numbered findings under headings, reading order, alternative text, tables and contrast, other observations, everything that could not be checked, and the checker's items placed against the findings.
Good looks like this: the designer can open any finding, go to the page and element it names and see exactly what to confirm in the file. Every figure in the summary adds up from the lists, nothing is added that the lists do not say, and no correction is written for anyone. The designer proposes the fixes and the publication owner approves them.
An invented example
Illustrative example with invented data. Client K is an invented regional arts charity and Studio M an invented design studio. The annual review, its pages, the funders named only by letter, the tools named only by letter, the roles and every row below are invented. The designer at Studio M listed how the finished PDF is built and what each element looks like, measured contrast and pasted the checker's items. The assistant compared the lists and returned the findings list. Nothing here describes a real document or a real review.
The inputs
1. Review settings
Document: Annual review 2026
File: annual-review-2026-v3.pdf
Kind: PDF
Organization: Client K (an invented regional arts charity)
Audience: members, funders and the public
As-of date: 2026-11-10
Designer: the designer at Studio M
Publication owner: Client K's head of communications
Accessibility reviewer: none
Thresholds: the defaults
2. Structure list (structure.csv)
seq,page,element_type,visual_role,heading_level,intended_order,content,alt_text
1,1,heading,heading,1,1,Annual review 2026,
2,1,artifact,decorative,,,Background wave pattern,
3,1,figure,image,,2,Cover photo of a summer workshop,IMG_0042.jpg
4,1,paragraph,text,,,A year of open studios,
5,2,heading,heading,2,1,Welcome from the chair,
6,2,paragraph,text,,2,This year the studios opened on more weekends,
7,2,figure,image,,3,Portrait of the chair,The chair standing in the main gallery
8,2,paragraph,heading,,4,Where the money came from,
9,2,figure,chart,,5,Pie chart of income by source,Pie chart
10,3,heading,heading,4,1,Programmes,
11,3,paragraph,text,,2,Open studios ran every Saturday,
12,3,paragraph,text,,5,Pull quote: I found my people here,
13,3,table,table,,3,Attendance by programme,
14,3,paragraph,text,,4,Figures include repeat visits,
15,3,artifact,image,,,Photo of children painting a mural,
16,4,heading,heading,2,1,Thank you,
17,4,paragraph,list,,2,Our funders: Funder A; Funder B; Funder C,
18,4,figure,decorative,,,Decorative swirl,decorative swirl
19,4,figure,image of text,,3,Contact box with address and phone number,Contact us
20,4,table,text,,4,Footer in two columns: address and charity number,
3. Table list (tables.csv)
table_id,seq,page,header_row_marked,first_column_holds_labels,first_column_marked,merged_cells
TB1,13,3,yes,yes,no,yes
TB2,20,4,no,no,no,no
4. Contrast list (contrast.csv)
pair_id,page,element,text_size_pt,bold,ratio,measured_with
K1,1,Title over the cover photo,36,yes,varies,Contrast tool A
K2,1,Subtitle on the cover,14,no,3.9,Contrast tool A
K3,2,Body text,10,no,8.2,Contrast tool A
K4,2,Chart labels,12,yes,4.5,Contrast tool A
K5,3,Pull quote,20,no,3.4,Contrast tool A
K6,3,Table text on shaded rows,9,no,4.4,Contrast tool A
K7,4,Charity number in the footer,8,no,,
5. Checker report (pasted)
Alternate text: page 1, figure alternate text may be a file name
Headings: page 2, text styled as a heading is not tagged as one
Primary language: not set
Reading order: needs manual check
The findings list
DOCUMENT ACCESSIBILITY REVIEW: Annual review 2026, annual-review-2026-v3.pdf
For: Client K's head of communications, with the designer at Studio M. Kind: PDF. Audience: members, funders and the public. As of 2026-11-10. Accessibility reviewer: none named.
Thresholds: 4.5 to 1 for other text; 3 to 1 for large text (18 pt and above, or 14 pt and above in bold). The defaults.
Summary: 12 findings. Headings 2; reading order 1; alternative text 5; tables 2; contrast 2.
Structure list: 20 rows on 4 pages: 4 heading, 7 paragraph, 0 list, 5 figure, 2 table, 2 artifact.
Tables described: 2 of 2. Contrast rows: 7: 2 below your threshold, 3 at or above, 1 varies, 1 not measured.
Not checked here: 5 items, listed under WHAT COULD NOT BE CHECKED. Other observations: 1. Checker items: 4.
FINDINGS
Headings
H1. Page 2, seq 8, "Where the money came from"
Rule: looks like a heading but is marked as a paragraph.
From the list: marked paragraph; visual role heading.
Question for the designer and the publication owner: Does the designer confirm H1 in the file, and what correction does the publication owner approve?
H2. Page 3, seq 10, "Programmes"
Rule: skipped level: level 4 follows level 2 at seq 5.
From the list: marked heading, level 4; visual role heading.
Question for the designer and the publication owner: Does the designer confirm H2 in the file, and what correction does the publication owner approve?
Reading order
R1. Page 3
Rule: read earlier than intended.
From the list: the structure reads seq 10, 11, 12, 13, 14 in intended order 1, 2, 5, 3, 4. Read early: seq 12, "Pull quote: I found my people here" (intended 5).
Question for the designer and the publication owner: Does the designer confirm R1 in the file, and what correction does the publication owner approve?
Alternative text
A1. Page 1, seq 3, "Cover photo of a summer workshop" (image)
Rule: alternative text is a file name.
From the list: marked figure; alternative text "IMG_0042.jpg".
Question for the designer and the publication owner: Does the designer confirm A1 in the file, and what correction does the publication owner approve?
A2. Page 2, seq 9, "Pie chart of income by source" (chart)
Rule: alternative text is only a general word; chart description gives no figures.
From the list: marked figure; alternative text "Pie chart".
Question for the designer and the publication owner: Does the designer confirm A2 in the file, and what correction does the publication owner approve?
A3. Page 3, seq 15, "Photo of children painting a mural" (image)
Rule: informative image marked as decorative, so screen readers skip it.
From the list: marked artifact; alternative text none.
Question for the designer and the publication owner: Does the designer confirm A3 in the file, and what correction does the publication owner approve?
A4. Page 4, seq 18, "Decorative swirl" (decorative)
Rule: decorative image is read aloud.
From the list: marked figure; alternative text "decorative swirl".
Question for the designer and the publication owner: Does the designer confirm A4 in the file, and what correction does the publication owner approve?
A5. Page 4, seq 19, "Contact box with address and phone number" (image of text)
Rule: text set as an image.
From the list: marked figure; alternative text "Contact us".
Question for the designer and the publication owner: Does the designer confirm A5 in the file, and what correction does the publication owner approve?
Tables
T1. Page 3, seq 13, table TB1, "Attendance by programme"
Rule: row labels in the first column not marked as headers.
From the list: marked table; visual role table; header row marked yes; first column holds labels yes; first column marked no; merged cells yes.
Question for the designer and the publication owner: Does the designer confirm T1 in the file, and what correction does the publication owner approve?
T2. Page 4, seq 20, table TB2, "Footer in two columns: address and charity number"
Rule: marked as a table but laid out as text: a layout table.
From the list: marked table; visual role text.
Question for the designer and the publication owner: Does the designer confirm T2 in the file, and what correction does the publication owner approve?
Contrast
C1. Page 1, pair K2, Subtitle on the cover
Rule: below your threshold.
From the list: 3.9 to 1, measured with Contrast tool A; 14 pt, not bold, so other text; your threshold 4.5 to 1.
Question for the designer and the publication owner: Does the designer confirm C1 in the file, and what correction does the publication owner approve?
C2. Page 3, pair K6, Table text on shaded rows
Rule: below your threshold.
From the list: 4.4 to 1, measured with Contrast tool A; 9 pt, not bold, so other text; your threshold 4.5 to 1.
Question for the designer and the publication owner: Does the designer confirm C2 in the file, and what correction does the publication owner approve?
OTHER OBSERVATIONS
Page 4, seq 17, "Our funders: Funder A; Funder B; Funder C": marked paragraph; visual role list. Question: does the designer treat this as a finding, or set it aside?
WHAT COULD NOT BE CHECKED
Descriptions for a person to confirm: seq 7, page 2, "The chair standing in the main gallery". Does each carry what the reader needs from the image?
Pages whose reading order was not checked: page 1 (seq 4 has no intended order).
Tables not described or marked unknown: none.
Merged cells to hear read aloud: TB1, page 3.
Contrast that varies across its background: K1, page 1, Title over the cover photo.
Contrast not measured: K7, page 4, Charity number in the footer.
Never covered by this review: how the file behaves with a screen reader or other assistive software; link wording; form fields; colour used alone to carry meaning; captions, transcripts and audio description for media; the document's language and title settings; and anything not in the lists supplied.
This is a findings list for the designer to verify. It states no conformance with any standard and declares no part of Annual review 2026 accessible.
CHECKER REPORT
Checker items: 4: 2 matched to findings, 0 with no finding there, 1 naming no page, 1 outside the five areas.
1. "Alternate text: page 1, figure alternate text may be a file name": see A1.
2. "Headings: page 2, text styled as a heading is not tagged as one": see H1.
3. "Primary language: not set": outside the five areas of this review.
4. "Reading order: needs manual check": names no page; the designer matches it by hand.
A checker that reports nothing is not taken as a pass for any area.
Every finding traces to a row in the lists supplied, as of 2026-11-10. The review changes no file and sends nothing; the designer verifies each finding, and the publication owner approves the corrections and decides on the release.
What happens next
The designer opens the PDF and confirms each finding, listens to the attendance table with its merged cells, checks the title over the cover photo by eye and measures the charity number. Client K's head of communications approves the corrections. The designer makes them and runs the review again on the next version with fresh lists.
Where it fits
- Design and user experience, where the designer checks the file before it leaves the studio.
- Marketing and content, which owns the reports, guides and brochures that carry the brand to every reader.
- Marketing and advertising, where a campaign deck or brochure is checked before the client handoff.
- Media and entertainment, where a digital feature, press pack or companion document goes to every kind of reader.
- Nonprofits, where an annual review or appeal pack has to reach members and funders who read with assistive software.
- Education, where course handbooks and slide decks have to work for every learner.
- Established businesses, with a steady flow of reports and decks built from the same templates.
- Growth & scaling, when more documents go out than anyone has time to check by hand.
It reviews one document. A task that runs across several screens of a website or app, such as signing up or paying, is a different piece of work (journey accessibility findings), and so is checking print artwork against product facts (packaging artwork handoff check).
What has to change for your company
- Your thresholds. The template starts from contrast thresholds many teams use and a common rule for what counts as large text. Put your own in the settings if your team has agreed different ones.
- Your lists. Decide which view of each document tool your designers use for the structure order, the image descriptions and the tables, and which contrast tool they use. Rename your columns once to match the template.
- Your areas. Five are built in. Add another, such as link wording, only with its own column and its own rule.
- Your general words. The template treats a description like "image" or "chart" as empty. Add the words your documents use, in each language you publish in.
- Your owners. Decide who verifies, who approves corrections and who decides on the release, for your own documents and for client work.
- Your platform. It is written for ChatGPT, Claude, Copilot or Gemini and works from files you supply. Decide which files the assistant may read.
What has been checked
Fully Powered AI wrote this workflow template. It was editorially reviewed on 11 November 2026. It was tried once on its invented example in Claude on 11 November 2026, and the record is published with it. It has not yet been tested in a client environment or as a packaged version in ChatGPT, Claude, Copilot or Gemini.
Testing in your environment means running it on a document your designer has already corrected, with cases written in advance: a heading marked as body text, an image hidden as decoration, a sidebar read out of place, a contrast ratio exactly on the threshold, and a file supplied without its lists. The designer checks every finding in the file before anyone relies on it. This workflow template certifies no conformance with any standard, gives no view on legal duties and writes no corrections.
Who decides
Your people keep
- Whether each finding is confirmed in the file
- Which corrections to approve
- When the document goes out
It does not
- Certifies no conformance with any standard
- Declares nothing accessible
- Writes no corrections
Questions people ask
Who is it for?
The designer who built the document, or the production lead, starts it when a designed document, deck or publication is ready to hand over or release. The publication owner approves the corrections.
What do I need to supply?
The review settings, a structure list, a table list and a contrast list, plus the checker report if an automated checker was run. No connection to the design tool or any account is needed.
What do I get?
One findings list in a fixed order: numbered findings under headings, reading order, alternative text, tables and contrast, everything that could not be checked, and the checker's items placed against the findings.
Does it say the document is accessible?
No. The list certifies nothing and declares nothing accessible. The designer verifies each finding and the publication owner approves the fixes.
What has been checked?
It was editorially reviewed and tried once on its invented example in Claude on 11 November 2026. It has not yet been tested in a client environment.
The full workflow template
Everything above is the outline. The complete workflow template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in Fully Powered’s public repository.
Free with an account. Sign in by email: no password and no card.
Next step
Put this workflow template to work for your team: on your plan, or built with you in a sprint.
Build this in a sprint
In a sprint, Fully Powered adapts this workflow template to your rules, proves it on your own work and puts it in the hands of the roles that own it.
Two ways to begin: