It is perhaps unnecessary to point out that there are problems with spreadsheets – others have done so better than I can – for instance here or here.

But if spreadsheets are so dangerous, why do managers continue to use them?

The obvious answer is that they have no alternative.   All the fancy applictions that gather up business data from different sources and show snappy graphs of the number of people who buy jumpers on sunny days are no substitute for a spreadsheet that allows the manager to construct his/her own model.

But it’s not only that.   The one skill that can be assumed in any analyst is an ability to use a spreadsheet.   In this group of people it’s even more important than their ability to use a word processor.   But all the fancy applications need to be learnt.

Which is why, if spreadsheets are really so dangerous, the solution is not to try to replace spreadsheets with other applications, but rather to make spreadsheets safe.  The only way that this can be done is to extract the data from the spreadsheet, so that it can exist independently from the spreadsheet,  and to introduce true audit-ability into the data.

This logic is so obvious that it’s very surprising that nobody (not even Microsoft!) has managed to think of a way of achieving this.  Yet the problem virtually defines the solution

First, it is clear that the ‘database’ behind a spreadsheet cannot be a relational database, unless it is a single table analysts simply do not have the skills to create and maintain relational databases.

Second, it is evident that individual items of data need to be tagged with meta-data – the source of that item of data.   In accountancy terms this can be called an ‘audit’, but I would prefer to resort to antique-dealing terms and refer to ‘provenance’.   Provenance, it seems to me, is exactly what we want when we refer to any fact rather than simply a number.  Numbers are a subset of facts in general.

Third, we need to ask how people ‘read’ a spreadsheet.   If I point to any particular item in a table, how do you know what that value means?  The logic is also inescapable – we understand the value in terms of its ‘context’.   By ‘context’ in this instance, I mean the row hand column headings in the sheet as well as any values in the headings which narrow down the options.

I’ll leave these thoughts here.  We at Azquo have long believed that we have the answer to all these problems.   What’s more we’re about to Open-Source our project.   It will be the dawn of a new age for spreadsheets which can be used without blowing up the organization.

For more info on our solution, see here.



Comments are closed.