Key takeaway: A .pptx file is a ZIP archive of XML files. What renders on screen is a fraction of what ships in the file. Speaker notes, entire embedded Excel workbooks, cached chart data, author metadata, logo URLs and recoverable "deleted" content all travel with every copy of the deck.
Last updated: July 2026
What is a PPTX file, technically?
A .pptx is a ZIP container holding XML files, the Office Open XML format. Rename any deck to .zip and open it: slides are XML documents, images sit in /ppt/media, embedded objects in /ppt/embeddings, and relationship files (.rels) map every internal and external link. Anything the file has touched can persist somewhere in that structure, whether or not it appears on a slide.
Sanitising a deck therefore means cleaning the file, not the slideshow. You will also hear this called scrubbing, blinding, anonymisation or de-identification; whatever the label, it fails if it only addresses the visible layer, with redaction of on-slide content being just one part of the job.
Do speaker notes stay in the file when you share a deck?
Yes. Every slide with notes has its own notes XML inside the package. The candid commentary, the "don't raise this with the client" asides, the internal shorthand: all of it ships with every copy, whether or not anyone opens notes view.
Does an embedded chart carry the underlying Excel workbook?
Often, yes. Paste a chart from Excel with default settings and the workbook is embedded in /ppt/embeddings. Every tab, not just the charted range. A single summary bar chart on slide 14 can carry a client's full financial model behind it.
What is the chart data cache?
Chart XML stores a cache of the values it plots. Even a linked chart, with no embedded workbook, retains the underlying numbers inside the deck. Delete the source spreadsheet and the figures live on in the presentation file.
What metadata does PowerPoint store?
The package's core and app properties record author names, the company field, template name, revision count, total editing time and last-modified-by. Custom document properties can carry client or engagement codes. For a consulting deck, author plus company plus a client-branded template name can identify the engagement even after every slide has been cleaned.
Can logos and images give away the client?
Twice over. Visibly, through logos and branded imagery on slides. And at code level, through external image links and hyperlink targets in the .rels files, which can point at client domains or internal SharePoint paths. Cropped images add a third route: PowerPoint keeps the full original picture and stores the crop as a display instruction, so the "cropped out" content is still in the file.
Can deleted or covered content be recovered from a PPTX?
Frequently, yes. A black rectangle drawn over text is a shape sitting on top of live text; the string is intact in the slide XML, and select-all plus copy-paste defeats it. Hidden slides are flagged not to display, but they ship with the file. Off-slide objects sit outside the visible canvas and remain in the XML. This is a large part of why manually sanitised decks are so often still re-identifiable: the visible layer was cleaned, the file was not.
| What hides | Where it lives | The risk |
|---|---|---|
| Speaker notes | Notes XML per slide | Candid commentary ships with every copy |
| Embedded workbooks | /ppt/embeddings |
Full financial model behind one chart |
| Chart data cache | Chart XML | Numbers persist after the source is deleted |
| Document metadata | Core and app properties | Author, company and template identify the engagement |
| External links and logo URLs | .rels files |
Point at client domains and internal paths |
| Cropped image originals | /ppt/media |
Full uncropped picture retained |
| Hidden slides | Slide XML, display flag off | Travel with the file |
| Covered text | Slide XML beneath a shape | Copy-paste recovers it |
What does this mean for consulting firms?
Nearly every consulting deliverable contains confidential client content, and much of it sits below the rendered layer. A slide-by-slide visual review, however careful, only ever inspects what renders on screen. The file layer needs code-level cleansing: stripping hidden traces such as logo URLs and embedded chart data, sanitising the XML itself, and doing it in a way that survives think-cell and the other tooling consulting decks are built with.
That is how Knovari approaches it. It cleans what you can't see, and logs every change so a compliance team can verify exactly what left the file. If you are preparing decks for a knowledge base, this file layer is also why sanitising documents for RAG and AI ingestion has to go deeper than the slides.
Related reading: Why find-and-replace redaction fails · Why PDF redaction tools fail consulting firms
Want to see how Knovari handles consulting deliverables?
Book a demo