Marketing Mirage

Marketing is a very interesting domain. You know, you hardly ever get the credit for what you achieve but you get loads of s*** when things go wrong even without your help. ;-)

Presently my main focus is online marketing and I am glad that it offers quick analysis of achievement or performance improvement. But with experience and after several discussions, I have come to realize that the metric on which efficacy of online advertising should measured is elusive and always under argument.

Lets see, if we have a site that sells sports equipment, should the online adv. be measured on the traffic coming to the site or on leads generated through online adv. or on the online sales through traffic source. This is a case where you can atleast target a particular segment and maybe show your ads to relevant clients.

First, suppose I choose traffic to the site as the metric, then problem could be that the SEM manager starts driving irrelevant traffic to the site and the overall sales remain stagnant but the cost of marketing goes up. ---> ROI moves down, so not a good metric.
* traffic to site could be a good metric for SEO campaigns but not SEM campaigns

Now lets select leads generated as the metric. This means that no. of forms filled on site through online traffic is the new metric. This metric could be a good metric if your conversion rate of leads is stable and not very random. But this requires cooperation from marketing, design and tech teams as various forms in different colors, formats, position and length have to be tried on before settling for the best.
---> ROI might move up inc ase of stable conv. rate, could be good metric.

Now consider the online sales as the metric. This is the best metric business wise, but could be difficult to target. You need to bring in people who will buy from your store. How do you target them? You can use SEM account to get you focused traffic but the intention to buy is not conveyed in the searches. This only gives a vague parameter on which to improve performance.
---> could be a great metric, but difficult to employ and optimize.