Understanding marketing dynamics

This article, written by Neil Mason, was originally published on Clickz.com on 06/11/09 and is republished here with permission.

At the Emetrics conference in DC I picked up a copy of a white paper by Eric Peterson on marketing attribution. In the white paper Peterson outlines the challenges with marketers relying on the traditional “last click” attribution models that are built into most web analytics systems, at least as a default setting. I’ve highlighted some of these issues before in this column and it’s good to see the subject getting more attention from other commentators in this space.

Peterson outlines a process he calls “Appropriate Attribution” to get some further insights into the way online marketing dynamics work and the roles that different channels play in the acquisition process. It’s a good approach in the absence of getting more granular with the data and looking in detail at the individual click streams and analysing the individual clickstream data to understand the patterns in marketing touch points that seem to lead to successful conversion outcomes.

Work we’ve done in the insurance market in the UK has shown how you can come to quite different conclusions about what marketing is working and what is not. In this case we looked at the behaviour of individual visitors buying insurance policies (or not) over time and the various marketing channels that were used in the process. In one case we looked at over half of the policies bought online required two or more visits to purchase a policy. This is understandable as people will generally be doing a certain amount of research before buying a policy, making sure that the policy meets their needs and looking at the price. In the UK market “price comparison” sites are a major feature of the way that people buy certain types of financial services and travel products. When we looked at all those policies where the buyer had taken two or more visits to buy the policy, price comparison sites were the last channel used in a significant proportion of cases. So price comparison sites come out very strongly when using a “last click” attribution approach to look at marketing effectiveness.

However, when we analysed the channels which people used to first come to the site, the picture was quite different. Only about half of those who completed the buying process via a price comparison site had started the process using that channel. Other channels, such as search and affiliates, were far more prominent when looking at sales attribution on the “first click model” than on the last click model. This evidence supports the notion of some channels are better for acquiring prospects and others are used to close the deal. When we also looked at the channels used between the first and last click, then visits that came “direct to site” featured more strongly. In the case of insurance products in the UK a core pattern of behaviour that seemed to be emerging was that people found the site through search or affiliates, bookmarked and returned to the site direct to do more research, and would then complete the purchase after having checked the price on a price comparison site.

Uncovering these sorts of patterns is not easy and we are limited by the constraints of the measurements technologies available. Peterson’s concept of Appropriate Attribution is a good approach for working within those constraints and will work better with some of the mainstream web analytics technologies than others. However, if an organisation is spending significant sums of money on digital marketing then it may be worth looking at the potential return from investing in getting a more granular view of marketing response and extracting the relevant data from the web analytics system and analysing that data in something else.
At last month’s Emetrics Summit in Washington DC Expedia explained how they were doing just that, in order to ensure that they fully understood how to attribute the effect of different digital marketing channels on sales. Getting the data out of the system through a data feed or similar has other benefits, as it allows you to create your own view of the world and to think differently about how to look at different channels, particularly Search. More about that next time. Till then…