A Better Model for Web Analytics Revenue Attribution
A regular day on the Web...
A website visitor finds your site in the Google search results. Excited by what you have to offer, she does some searches for similar products over the net, clicking on a range of pages from both you and your competitors. A few times she comes across some comparison shopping sites that include your listing and she clicks through to see if it's the same or perhaps a different offer. Eventually she decides that yours is the site that she's going to buy from and mentions your site to her husband. Later on he types your domain into Google and clicks through on your Adwords link that appears right at the top of the page. He loves your product too and buys it for his wife.
To which channel do you attribute the sale? Organic Search? Comparison Shopping? Adwords?
There is no "correct" answer. The fact is, all played a role. Organic search introduced your site and through further research on organic search the brand and offering was reinforced, assisted by comparison shopping listings. In this scenario, Adwords played the lesser role since if the Adwords listing wasn't present, more than likely your site would have ranked highly in the organic listings for the search on your domain name. But in the end, according to typical web analytics packages, Adwords would have been attributed the sale.
Of course there are any number of scenarios, with each channel potentially playing a different role with varying levels of contribution to the end sale. Multitouch Analytics provides a greater insight into what is really going on.
Google Analytics and Last Session Attribution
Google Analytics uses an "attribution model" (a way of attributing conversions to visits) based on "last session initiator", that is, the channel through which your site was found that initiated the last session of browser activity for the given visitor before the sale. A session is (roughly) any browsing activity on your site that happens within a 30 minute time period. So, for example, if a visitor enters your site via organic search, then an hour later visits again through Adwords and then makes a purchase, the organic search entry is forgotten, Adwords initiated the last session before the sale, so Adwords is attributed with the sale.
Clearly the last session model is flawed, as at best it gives an incomplete picture of user behaviour with respect to your marketing channels, but more seriously it gives a very biased picture. Some would argue that it doesn't matter how you measure providing you measure consistently; unfortunately this is not true, as illustrated by the scenario described above. Organic search is often used more as an early introduction and buyer research tool, whilst comparison shopping engines and paid advertising tend to be used more after the decision to purchase has been made, so "last session" attribution favours those channels.
Google Analytics also offers first session attribution, but that is no more helpful for the same reasons. What's more, you have to choose up-front whether you want to use first- or last-session attribution, and once committed it is a difficult and long process to change over.
Multitouch Analytics integrates with Google Analytics and offers additional reports that give you the full story on which marketing channels have been touched by your buyers.