Speechless not numberless
What’s the ideal number of attributes in Controlling? Zero. That is, at least, if they are judgmental, superlative or polemic. And that occurs more often than you might think.
In addition to routine reporting, many controllers today feel that their true calling is to consult managers. And the products of their work are individual reports that almost mirror the tomes of charts from large strategy consultancies. They contain an “action title” that should include an assessment, a warning, a conclusion or a recommended action because only then will they be understood as a consultation. The actual content of the report provides the evidence for the statement made in the headline. In this type of scenario controllers should then, at the very latest, ask themselves a few questions about language. May, should – or must – a number be accompanied by a judgmental attribute? After all, consulting is an assessment and, therefore, a judgment.
| Attribution | number | Source |
| “profits collapse” | -71,4 % | [1], p. 21 |
| “profits dropped sharply” | -32 % | [2], p. 18 |
| “profits fall significantly” | -32 % | [2], p. 18 |
| “profits clearly sagged” | -14 % | [1], p. 17 |
| “revenues on a downslide” | -12 % | [2], p. 18 |
| “unprecedented crash of tax revenues” | -8,8 % | [3], p. 3 |
| “stock is up considerably” | +8 % | [2], p. 14 |
| “respectable quarterly results” | +29 % | [2], p. 21 |
The translations above show different ratings used by German business newspapers. Do we want that? Not in controlling, at least. (Sources: [1] Süddeutsche Zeitung from July 24th 2009, [2] Handelsblatt from July 24th 2009, [3] Handelsblatt from July 13th 2009.)
If you take a quick look at today’s papers (i.e. July 24th 2009), you can see that the press generally answers these questions with ‘yes’. As a result, the press often exposes itself to the criticism that they are more concerned with selling papers than presenting the facts. Today, things are collapsing, sliding, free falling, sinking dramatically, and crashing at all-time rates. Positive trends, however, are commented with caution.
About a week ago, I stumbled upon a headline that translates to “Tax revenues collapse” (Handelsblatt Nr. 131, July 13th 2009, p. 3). The article said that the Finance Ministers in Germany’s state and federal governments have witnessed an “unprecedented crash” in tax revenues. “The state collected 8.8 % fewer taxes in the past month.” Hmm. As the accompanying chart showed, the 2009 tax estimates were on a level somewhere between 2006 and 2007, which were both relatively strong years. “Collapsing” is something serious. Someone who collapses and doesn’t receive immediate medical treatment could become unconscious. Is that the case when the government takes in 9 % fewer taxes?
For the record, doctors also have a similar problem – they use attributes as well. These are even officially defined for the leaflets that come with packaged medications. The numerical risk of side effects uses the following scale. Does that make things clearer? I think just the opposite.
| Very often | More than 1 in 10 patients |
| Often | 1 – 10 patients out of 100 |
| Sometimes | 1 – 10 patients out of 1,000 |
| Rarely | 1 – 10 patients out of 10,000 |
| Very rarely | Less than 1 patient out of 10,000 |
These ratings are used by Germany’s Federal Institute for Drugs and Medical Devices. Personally, I think we can pass on those, too.
So what should we use in controlling? I personally feel that our situation is similar to studying the package leaflets. What do we imagine when someone says very often, often, sometimes, rarely or very rarely? What advantage do we have by not naming the numbers that we know and can prove? The same situation applies when something rises or falls strongly, clearly, considerably, significantly, massively and dramatically or in an explosive or unparalleled manner. It’s a luxury in controlling that we have access to actual numbers, whereas others can only make estimates. If someone reports a 30 % change, everyone knows exactly what that means. With judgmental attributes, however, that is definitely not the case. The controller’s job is to bring the necessary information together and to ensure that the reader can verify it as well.
Jean-Christophe Babin, the managing director at TAG Heuer, probably shared this opinion. In an interview he combated the brash allocation that his industry was able to “win over fewer and fewer customers” with numbers:
You know the statistics of Swiss watch exports. But we’re still doing fine. 30 percent less in our industry means that we are back at the level of 2006 – and that was a very good year in our industry.
Conclusion: People state their numbers and don’t replace them with polemics if they have their facts together and know how to interpret them.
I have been contemplating this same topic. Verbal descriptions can be misleading, even among people of the same background. For another data set exploring this issue see Psychology of Intelligence Analysis by Richards Heuer. Figure 18 of Chapter 12 compares several statements of probability against what various people thought they meant. The terms, probable, probably and likely are particularly interesting.
Monday, August 10th, 2009, 9:02 pm