I started a new project this week, doing a piece of business modelling for a client considering a new venture. As a result it was quite a different week to normal: very few Zooms or calls, and several days of intensely focused time researching and sorting the data I could find on the subject and building the first version of the model in Excel. Then testing the logic with a known set of data, before building a set of possible scenarios (base case, downside, upside) and running the model past a couple of senior stakeholders at the client for some useful feedback. Next week I’ll be working on improvements to the basic model and then moving from the unit economics of a transaction to forecasting multiple transactions over time.

As a result of doing this, I returned to two essential reference texts, one obvious and one a little less so. The obvious one was my worn, heavily annotated copy of the Economist Guide to Business Modelling. The most recent edition was published when Excel 2010 was the state of the art, so a lot of the detail is out of date: for example, when it was written, synchronous, multi-user access to a spreadsheet wasn’t possible, and I’d love to see an updated edition that addressed recent Excel developments (and Google Sheets). But the underlying logic of the book is still spot on, and the structure and approach it prescribes has been invaluable to me for nearly a decade: I certainly can’t think of a single book with a higher ROI for me.


The author of my other reference text was Jorge Luis Borges. Not perhaps a name generally associated with Excel data. Stick with me. The short story in question, On Exactitude in Science, is a single paragraph in length—short enough to quote here (or you can hear it read by Will Self here:

...In that Empire, the Art of Cartography attained such Perfection that the map of a single Province occupied the entirety of a City, and the map of the Empire, the entirety of a Province. In time, those Unconscionable Maps no longer satisfied, and the Cartographers Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided point for point with it. The following Generations, who were not so fond of the Study of Cartography as their Forebears had been, saw that that vast Map was Useless, and not without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.
—Suarez Miranda, Viajes de varones prudentes, Libro IV, Cap. XLV, Lerida, 1658

Setting aside how extraordinarily evocative and economical it is, the story is an elegant reminder that a completely accurate model is wildly unrealistic and doomed to failure. Representing every variable (let alone any stochastic element) would be as unwieldy as a map at 1:1 scale. Every model requires a degree of abstraction to make it usable, and an Excel model accessible to non-specialists requires quite a high degree of abstraction.

The art of modelling is choosing which variables matter—a combination of experience, research, critical feedback and a little trial and error. For the model I built this week, there were at least several dozen potential variables, but in the end I used only eleven. Abstraction doesn’t necessarily mean inacccuracy: The Economist guide makes the point that 80% of a business’s cost base can be modelled from as few as 15-20 data points. It relates to a philosophical point often misattributed to John Maynard Keynes (in fact, coined by Carveth Read): it’s better to be roughly right than precisely wrong. If the model I’ve been working on turns out to be commercial, usable and roughly right, I’ll be very happy.


My current reading and listening pile: