13 January, 2020

Memo: Communicating business numbers

March 2019

To: Management Team, X GmbH

Having just received and reviewed your documents in advance of Thursday’s board meeting, I have a particular bugbear that I am now going to ventilate in the hope that it will be helpful in designing the way in which we communicate with each in numbers. The following aspects inform my thinking:

  1. Data is only ever useful when it allows us to extract pertinent information;
  2. In order to extract information from data it must necesarily be prepared in such a way as to facilitate not hinder comprehension;
  3. Responsibility for organising the data is always with the person who has collected the raw data, unless he or she specifically states that they are not. In other words the moment you decide to collect and share data, you are responsible for the quality of the presentation and the ensuing information mining activities:
  4. Intelligent discussion can only be based on uniform data – if everyone is working with different data or data formats then the energy of the discussion will be on the differences and not on the substance and subsequently a waste of everybody’s time;
  5. The quality of an organisation is reflected in the quality of the data and its dissemination because only when everyone has access to the same quality and integrity of data can correct, speedy and substantiated decisions be made.
  6. Learning organisations can use well-prepared data to reference decision-making quality and constantly to improve. Bad, messy and disorganised data is both inaccessible for future learning and – being available in different and sometime contradictory formats – useless for the purpose of intelligent back-referencing and post-mortem analysis.
  7. Ask yourself always how you would respond to a presentation of your data if you were on the receiving end. Would you be able to tell after a short study what story the data is telling you and is that the story you want to tell?

Using the data I have received from various sources this morning there is very little uniformity or conscious presentation design from any of the participants. Everyone has shared bits of data (different time frames, different data types, different extrapolations, different formats and presentations) so that after numerous mails there is

  • no uniformity
  • no structured design to capture the information
  • probable duplication of effort in collating the data (with commensurate waste of management time) and
  • very little focused information extraction.

Your last spreadsheet is a case in point. It contains all the data – so thank you for that – but the data is organised in a way that actively hinders the extraction of insight, as the reader has to jump from month and year up and down the spreadsheet, when a different ordering of the data would have made the analysis much easier.

The key question to ask is „What is the purpose of the exercise?“ In this case it has to be to achieve the highest possible degree of accuracy in forecasting the expected demand for our product from one of key customers over the rest of the year. In order to do this we need to be able to compare the months individually and cumulatively over the years and to create a rolling 12 month revenue curve. Presenting numbers horizontally instead of vertically is not helpful in the best of cases and in this case definitely so.

I would ask you to rework the data orientating your presentation on the above guiding principles. We can then use that presentation to draw management conclusions.

I have no wish to appear schoolmasterly, but in my experience and from my background, data integrity is crucial in arriving at intelligent decisions whether individually or (especially) in a group. If it is any consolation, most people and businesses are bad to awful at this.

Best wishes


PS For inspirational reading around this subject please refer to Edward Tufte’s outstanding book “The Visual Display of Quantitative Data” especially the chapter on the Challenger 13 shuttle disaster and how it might have been prevented if the data had been displayed more intelligently