Paul.
Try PaulTry Paul

From Excel to PowerPoint: moving your numbers without corrupting them

By L'équipe Paul · · 6 min read

Ninety-four per cent. In Raymond Panko's synthesis of thirteen studies of operational spreadsheets, the reference work on the subject, that is the share of the 88 spreadsheets examined that contained at least one error, at an average cell error rate of 5.2%. The most recent field audits, with better methods, still find errors in at least 86% of the spreadsheets they inspect.

Hold that figure, because the number on your slide came out of one of those spreadsheets. And there is a precise moment in the life of a corporate number when it is most likely to go wrong: the moment it leaves the spreadsheet and enters a slide.

That moment is the fault line between two worlds that do not share rules. In Excel, a number is the output of a calculation, attached to its inputs, recomputed on every open. In PowerPoint, a number is text. It no longer knows where it came from, it does not recompute, and nothing signals that it has aged.

The whole subject of Excel-to-PowerPoint conversion sits in that loss of status.

The starting point: your source numbers are already fallible

Panko's conclusion is the one that matters most here, and it is uncomfortable. Spreadsheet errors are rare on a per-cell basis, but in a large model it is highly likely that at least one bottom-line value is incorrect, and both developers and organisations are heavily overconfident in the accuracy of their spreadsheets.

So the slide is not where the problem starts. It is where the problem becomes invisible: in a spreadsheet you can audit a formula; in a slide you see only a number.

The three conversion regimes

There are only three ways to move a number from Excel to PowerPoint, and their risk profiles differ completely.

1. Retyping by hand

Someone reads the spreadsheet and types the figure into the slide. This remains, by a wide margin, the most common method.

Its risk is twofold. The first is the typo, which goes unseen because nothing in the slide contradicts it. The second, more insidious, is temporal drift: the slide is built on Tuesday, the spreadsheet is corrected on Wednesday, the meeting happens on Thursday. Nobody is lying, and the projected figure is still wrong.

This is the regime with zero traceability. Six months later, when someone asks where that 14.3% came from, nobody knows.

PowerPoint can paste Excel data while keeping a link to the source workbook, via Paste Special → Paste link, or by inserting the workbook as a linked object. Microsoft documents both methods.

This is better, and routinely misunderstood. Microsoft states the key point explicitly: data that is simply copied and pasted does not update automatically when the workbook changes. You have to right-click and choose Update Link. Updating is a gesture, not a property.

And links are fragile by nature. They point at a file path. Move the workbook, rename it, email the deck to a third party, open it on another machine: the link breaks. Microsoft devotes an entire page to repairing broken links.

Linked pasting works well inside a stable perimeter: one machine, one directory tree, one author. It holds up poorly as soon as a document circulates.

3. Regenerating from the source

The third regime stops treating the slide as a container to fill and starts treating it as an output to reproduce. The spreadsheet stays the source of truth; the presentation is regenerated from it as often as needed.

The benefit on first production is modest. It lies in the subsequent iterations. During close or budget season, a set of figures moves three to five times before the meeting. In the first regime, every revision is a fresh opportunity to get it wrong. In the third, it is a command.

This is where generation tools change reporting economics. A handsome slide is a side effect; cheap reproduction is the point.

What has to survive the crossing

A number that crosses correctly keeps four properties. If one is missing, the conversion is degraded.

The first is scope. "14.3%" means nothing. "Gross margin, France perimeter, Q3 2026, excluding exceptional items" means something. Scope lives in the spreadsheet's headers, and it is what conversion loses first. The rule: scope goes in the slide title or subtitle, never in an 8-point footnote.

The second is unit and scale. Thousands, millions, percentage of what. Scale errors by a factor of a thousand are the most frequent and the most embarrassing, because nobody catches them in the room.

The third is the as-of date. A financial table without an extraction date is an assertion without a timestamp. It belongs on the slide; the filename does not count.

The fourth is comparability. An isolated figure does not inform; the variance does. Against what: budget, prior year, previous forecast? The slide should answer that without being asked.

The most common design error: transposing the table as-is

The natural reflex is to paste the Excel table into the slide. It is almost always wrong.

A reporting table is built to be consulted: you look for a row, read it, compare. A slide is built to be looked at: it carries a message in seconds, in a room, from a distance, often badly lit.

Those functions are incompatible. Pasting a 40-row table into a slide produces an object nobody reads and nobody dares remove, because it "looks rigorous".

Useful conversion is a deliberate reduction: on the slide, the three to five figures that carry the message, with the variance highlighted; in the appendix, the full table for anyone who wants to check. It is the same discipline as a board pack. The detail exists; it simply does not occupy the main thread.

How to choose your regime

Three questions settle it.

  • How often do the numbers change? Once a year, retyping is acceptable. Every week, it is not.
  • Does the document circulate? If so, links will break. Either freeze deliberately, or regenerate.
  • Must every figure be justifiable six months later? If so, source traceability is not optional, and retyping is disqualified.

On a recurring reporting cycle, monthly or quarterly, the third path almost always wins. Paul takes your workbooks and notes as input, builds the slides on your corporate template, and regenerates when the figures move. The full mechanics for a quarterly cycle are described on the quarterly reporting use case.

One last point, and it is not technical. None of these regimes corrects an error present in the source table. Automating a conversion industrialises whatever you feed it, the good and the bad. Panko's 94% remains the real subject; conversion only decides whether the error spreads quickly or slowly.

Sources

  • Excel
  • PowerPoint
  • reporting
  • data
  • automation