Turning a days-long PowerPoint build into a one-click, 660-slide deck

The challenge

Every retailer review at this business meant building a large PowerPoint deck from scratch, running to hundreds of slides before a word of commentary had even been written.

The process was entirely manual. The analyst pulled the figures out of Power BI by hand, dropped them into slide templates one by one, refreshed everything, then tidied the file so it wouldn't break, and repeated the whole exercise for every part of the range.

What should have been analysis time was being swallowed by hours of copying, pasting and formatting, and the templates themselves were old and prone to breaking, adding a layer of fragility on top of the manual workload.

A full review across the range took days of building before the analyst could even begin writing commentary or having the client conversation the deck was meant to support.

Our approach

1. 01

   Understanding the existing decks

   We reviewed the analyst's current review decks in detail to capture exactly what each one needed to show, across the different audiences and parts of the range the business reported on.

2. 02

   Building the generator

   We built a Python tool that reads a simple scope from a front sheet, queries the client's existing Power BI data directly, and assembles a complete PowerPoint deck automatically, then tested it against live data and refined it with the analyst's feedback.

3. 03

   Branding and going live

   We matched the output to the client's existing branding and put the tool into everyday use for a live retailer review, so the analyst was producing genuine, client-ready decks with it from day one.

The outcome

The analyst now sets the scope in a simple front sheet, clicks once, and a fully branded deck of around 660 slides is built automatically, drawing straight from the client's live Power BI data so the numbers are always current.

The tool tailors what it includes to the audience, so a supplier version leaves out commercial detail a retailer version includes, without any extra manual work from the analyst.

Because the finished deck is self-contained rather than a fragile template file, the breakage problems of the old process are gone, and the analyst's time now goes into analysis and the client conversation rather than assembling slides. The same foundation is set up to be extended further, whether that's scheduling decks automatically or adding AI-drafted commentary down the track.