Technically we are able to follow most web users wherever they go. Technically this gives us the chance to craft a profile as unique as the person that it is attributed to. Technically we could then go ahead and complete this profile with data from real life. And technically we then know exactly what the person in question wants, needs and longs for.
Only that than the person is not really a person any more but has developed into being a target. And a target is not a subject, but an object you use to drive up your ROI. And thus we forget where the boundaries are, and how far we can go without actually infringing the right of those targets. And one day the target finds out about the set-up (which in today’s interconnected world is not surprising), connects with others and takes action. And suddenly the target has become a person again, but one which we cannot connect any more.
Thus, when implementing metrics and gathering data, we need to be very clear about what we need for optimization instead of what is technically feasible to do. Anto Chittilappilly on imedia connection has written a column about how to leverage cross channel attribution through the following steps
- define your goal: why do you want to do this?
- secure executive buy-in: be sure that this is part of the strategical plan of the company
- understand what data you have: this is the starting point for marrying data together
- find the right technological provider, if you cannot provide the technology yourself
- put the results into action: make sure that the results reach the people in the company that make decisions for using those results
- measure the results and optimize.
However, there is one important step missing:
- check which data are absolutely necessary and which are not: make sure you only collect what you really need.
In data mining the pareto principle of 20 of the data will yield 80% of the result is just as true as in most other domains. The remaining 80% of data may still aggregate a little bit to the overall performance, but most likely the time spent is better invested somewhere else. We should use the data we have wisely, and we should only collect what we really need. Just because it’s doable it doesn’t mean that the advantages always outstrip the risks.
Taking the customer serious implies to treat their data with respect, instead of collecting everything in the hope to target a tiny little bit better. We should keep the balance between knowledge and usefulness.