Benchmart secret #4: Old-school benchmarks are not anonymous

Some benchmarks are never intended to be anonymous, they are intended to be shouted from the rooftops. "the Top 3 supply chains in the world are ..."

These benchmarks are based off perception or publicly available data (such as financial reports) and they are of very little use in driving targeted performance improvement. They are quite good to show your boss if you hit the top 3 list.

When you want to get something useful, something that let's you see what is really happening, you must share more sensitive information but, of course, you need to do this without giving all your data to your peers, you need anonymity.

The old-school style benchmark report shows your company's data across multiple metrics in comparison to your peers' data. You, of course, know which of your competitors and peers were involved but, to keep it all anonymous, in your report, every company other than yours has a false name (Company A, Company B, etc.) Now the fun starts !

Assuming you know something about your own industry (and you should) and you know what other companies are in the study (you do) can you guess which company is which? It's like going to a masquerade ball and wondering whether the one and only guy with red hair is the guy, that you know is coming, that has red hair! Hmmm....there are 3 taller women, but only one is brunette and slim - easy ! Actually, there are typically more people at a party than companies in a old-school benchmark study. Relatively speaking, guessing who is who at the masquerade ball is hard.

If you want anonymity, your data should never leave the benchmarking organization. A crude way to do this is to take all the company labels (other than yours) off all the benchmarks and switch up the sequence you present them in. You may still be able to spot an individual company on a single metric but you can't find that company anywhere else in the report. No harm done, right? Perhaps not, but now you can't see the trade-offs a company is making to get an individual result (e.g. low inventory and very low service levels).

For true anonymity and useful benchmarks you must use a model and compare performance to that model. The model captures, quantifies and explains the relationships between performance measures, environment and decision variables so you can see the trade-offs being made. With a model-based benchmark, each company compares their own data to the model and never to the the raw data from other companies. Now that's real anonymity.