The team has proofread and approved the copy. Then the designed version comes back with a sentence cut off at the end, a table row paired with the wrong figure, or a caption sitting beneath the wrong image. The approved copy can be flawless, but the asset your audience receives can be wrong.
That gap is where a post-design QA process starts. It reviews the version that will ship against the original sources, approved copy, brand template, and final-format requirements. A final proof involves a side-by-side comparison of the approved file and the designed version. A quick read of the PDF won’t catch the same issues.
The risk is easy to miss because the finished asset usually looks polished. Errors that the design team introduce often blend into that polish: a missing phrase still leaves a complete sentence, and a wrong figure still looks at home in a chart. You need checks that test the content rather than the appearance alone.
The same principle applies earlier in an AI-assisted workflow. Our AI editing checklist covers the AI tells we catch on autopilot. The checklist below covers what can change after those words move into design.
Post-design QA isn’t a single check. Even after copy approval, the design process can introduce errors or change how people understand the content. The five checks below help you catch those problems before publication:
Let’s look at how each check works in practice.
A document can read perfectly and still be factually wrong. Proofreading checks the copy in front of you. Checking the sources confirms whether the names, dates, figures, and formal statements in that copy match the evidence behind them.
We saw the difference in an urgent public-record filing that had to go to print under time pressure. Before release, the reviewer still had to check every name and address in the finished document against multiple reference files. A spelling check could confirm that a street name looked plausible, but it couldn’t confirm that the address belonged to the right person or entity.
Run this check against the fields that would cause the most damage if they were wrong. Mark each fact in the final asset, locate the original source, and confirm an exact match. If a chart rounds a figure, check the approved rounding rule. If a claim depends on a footnote, make sure design hasn’t detached the note from the claim.
Decide which sources take priority before the review begins. A signed data sheet may take priority over an earlier email, while an approved legal statement may need to remain exactly as written. Note the source beside each important detail so another reviewer can repeat the check. If two references disagree, stop and ask the person responsible for the content which version the team should use in the final asset. Don’t choose the version that fits the layout best.
Keep this step separate from proofreading. A clean sentence and an accurate sentence aren’t always the same, and design can move or crop the information that made the approved wording correct.
Character limits, narrow columns, and tables rearranged to fit the page create editing decisions that weren’t present in the approved copy. If the layout would change meaning or emphasis, flag the problem for the person responsible for the content. Don’t make that decision without approval.
In one recurring report we worked on, strict character limits collided with a data-heavy layout. The approved copy used parallel wording across sections, but that structure broke when the design team placed the text into the template. If we had reworded each instance to fit, we would have produced a tidy report and a different editorial rule every time.
The reviewer logged the problem instead. The person responsible for the content then approved a consistent solution and made it a standing rule for that report format. The client said the edited reports were better than the versions they had been using. The layout problem became a repeatable editorial decision instead of a series of unapproved changes.
Designers still need room to solve genuine layout problems. The check simply makes the decision visible when the solution changes the wording, meaning, or order of information. A short note can prevent the same issue from recurring in every edition.
For each flagged problem, record where it occurs, what the template allows, and the wording or design change the team approved. That gives the next instance a clear rule to follow.
Once the layout is final, put the approved file beside the designed asset and compare them line by line. This is the final proof. A PDF read on its own may catch a broken line or an empty text box, but it rarely exposes a missing word that still leaves a grammatically correct sentence.
We run this comparison even when we didn’t proofread the original copy. It regularly catches the project’s first typo because it’s the first time anyone has reviewed both versions together. The page can look finished, and the error can read naturally in isolation. The difference becomes obvious only when the approved wording sits next to it.
Move through both files in the same order. Compare headings, body text, captions, tables, labels, links, footnotes, and legal wording. Check repeated components individually when separate fields populate them. Record every difference, including deliberate design edits, so the content owner can distinguish an approved change from an accidental one.
Use a fixed rhythm so your attention doesn’t jump ahead: read one unit in the approved file, find the same unit in the final asset, and mark it complete. Zoom in on tables and charts rather than relying on a full-page view. After corrections, compare the revised areas again and scan the surrounding content. A small fix can create a new line break, push text out of its box, or separate a label from its figure.
Some design software can catch problems automatically. For instance, Adobe InDesign Preflight can detect problems such as missing fonts, changed images, and text that doesn’t fit its box before handoff. Those checks are valuable, but they don’t confirm that every sentence still matches the approved file. You need both software checks and a line-by-line comparison.
Subscribe to Beyond the Margins and get your monthly fix of editorial strategy, workflow tips, and real-world examples from content leaders.
Δ
AI can extract text, suggest layouts, and summarize a slide, but it still needs human review when source material has to fit a branded template. That work involves decisions about what someone can shorten and where each point should appear on the slide.
An internal subject-matter expert’s slide deck makes the problem clear. The source may contain useful material in dense speaker notes that also include pasted tables and slides with no consistent structure. To reformat it, someone must decide which detail becomes the headline. When they should rebuild a table, and when to split a crowded slide into two. A tool can put the words into the template and still give the wrong idea the most visual weight.
A brand template also carries editorial rules that may never appear in a prompt. Font size can distinguish a conclusion from supporting detail. Position can show which chart a note belongs to. Repeated slide types can signal when a new section begins. A generated deck may be neat and technically on-brand while breaking those relationships.
Use AI where the task is clearly defined, such as to extract slide text or identify repeated labels. Keep a human responsible for what gets the most attention, whether the content fits the template, and comparison with the source material. This boundary will move as the tools improve. For now, manual review remains part of design QA when you move rough or inconsistent content into a new layout.
Review the asset your audience will actually receive. A script can’t show whether a disclaimer appears long enough to read, and a storyboard can’t reveal a transition that hides the final word of a caption.
For GIFs, check the exported file outside the tool used to create it and wherever the audience will see it. Check timing, cropping, and readability against every background.
For video, use a video review tool so a comment can point to the exact frame where the problem occurs. Then watch the final export from start to finish. Don’t use the project file as a substitute.
PDFs need a separate check too. A multi-column page can look correct while a screen reader reads its content in the wrong order. Check the order of headings, meaningful link text, table headers, and the sequence in which a screen reader reads the page. Before release, open every link in the exported PDF, confirm that fonts display correctly, and check the filename and version.
Test the asset where the audience will see it as well as in the file itself. A video that works in the editing tool may display different captions after upload. A GIF may crop differently in an email template. A PDF link may work on a local drive and fail once the file moves to a content platform. Open the published preview, check the asset on a device your audience is likely to use, and repeat any accessibility check that depends on the exported file. Automated tools can flag likely problems, but a person still needs to confirm that the result is usable.
These five checks work as one post-design QA process. When you check the sources, you protect factual accuracy. Flag layout problems to keep content decisions with the right person. Compare the approved copy with the finished version to catch accidental changes. Human review confirms that the right information stands out, while the final check tests the file your audience will receive. Skip a check and the final handoff won’t change anything important.
If a current project has exposed a gap in that handoff, Proofed can help. Book a free design review for one finished asset, or schedule a free style guide audit to turn recurring issues into clear rules for future projects.
Post-design QA is the review that happens after someone lays out a document or asset. It checks the final version against the original source facts and approved copy, then tests the published format.
No. Proofreading can correct the pre-design copy, but layout can introduce missing text, altered figures, or changes in emphasis. A final proof compares the approved file with the designed asset side by side.
AI can support parts of the task, but it can’t yet reliably turn rough or inconsistent source material into a specific branded template without human review. A person still needs to judge what should stand out and what the space allows. Review the generated asset rather than assume that a correct draft guarantees a correct layout.
Yes. Review the finished video or GIF because timing, motion, cropping, and on-screen readability don’t appear in a script or storyboard. Proof the final exported file before release.
Got a high volume of content to edit?
Let’s talk about the support you need.
We use cookies to give you the best possible experience with Proofed. Some are essential for this site to function; others help us understand how you use the site, so we can improve it. We may also use cookies for targeting purposes.