Tuesday, December 1, 2009

The data-driven organisation

As I may have mentioned before, I worked a good number of years for DHL (a multinational, wholly owned by Deutsche Post, with an annual turnover of AUD $7bn), in various European senior marketing roles. Since moving to Sydney, I’ve worked on the other side of the fence, both in research and advertising agencies, and, having worked with a number of clients, I’ve come to the view that DHL was a highly data-driven organisation.

I believe many Australian businesses would benefit from instilling a stronger culture of analysis within their organisation, by adopting some of the following practices employed by DHL:

DHL used business goals to drive decision making.

One of the common problems with analysing data is that companies look at their numbers without putting them in the context of their overall business. As a result, when they receive an analytics report (often in a crisp, new binder with colourful, attractive charts) they sift through it without knowing what it means to their bottom line. Data-driven organisations make sure that goals and metrics are defined and agreed on, and they communicate them to everyone according to role.


Data-driven organisations never rely solely on gut feelings.

For sure, data and research will never give you definitive answers. However, making educated decisions based on analytics-driven insight will help you meet your goals. This doesn’t mean you should throw your experience out the window. But you should be honest with yourself about what you really know and don’t know. I have observed many clients mistake their own personal likes and dislikes for insights in customer behaviour.


Successful data driven organisations spend money in the right places and in the right way. As a result they can justify every marketing dollar they invest and tie it to business goals and KPIs.

For example, instead of putting millions of dollars every year into a full website redesign, they target their spending according to where it will be most effective. To do this, you first need to identify which aspects of your website are most important in driving your business. Let’s say you have a lead generation site that tries to get visitors interested in you offerings so they will request a meeting with a sales rep. Every upgrade you make to the site should improve the way it converts your visitors into high quality leads. Your efforts may involve highlighting calls to action or streamlining request forms so that it’s easier to separate good prospects from bad.

Companies like DHL that are serious about data also use analytics to maximise their ROI for online and offline initiatives. If you have an underperforming marketing campaign, you should reallocate resources to areas that have a better probability of success. That way you’ll always be sure you’re investing wisely and strategically in areas that drive your business goals.


In a data-driven organisation, every team, business unit and individual operates under a unified, global set of standards.

To do this, you need to set overall business goals and metrics. Then you assign different groups in the company their own targets and metrics based on how their works impacts the top-level goals. This process continues down the chain of command until it reaches individual employees. In the end, everyone is aware of how their actions contribute to the success of the company. I’ve found that once people and departments have clear and specific metrics to define their success, they tend to have an entirely different (and often much more motivated) approach to their work.


Whenever a data-driven organisation launches an initiative or campaign, it has already put together a forecast of its potential impact on the business and bottom line. When these projects are complete, the organisation also wants to know how the outcome of the project compared to earlier estimates. For this, you need to include a full post-launch analysis in your process. You should not only focus on the outcome but also use the opportunity to look at the forecasting process. Are you making accurate predications and if not, why?