…on its own
Pretty much every SEO or online marketing campaign I’ve ever been involved in has used last click as the defining metric for success. In part this is due to the limitations of most analytics packages, but it is also a convenient method of attributing value to different channels.
Of course, the problem with last click is that it simply doesn’t tell you a great deal about the effectiveness of your overall marketing mix – all it tells you is that on a particular date, one of your customers bought something from you after arriving at your website in a particular way.
At the 1986 FIFA World Cup, a fleet footed Argentinian player called Diego Maradona danced round the whole of the England Football team and scored one of the greatest goals in the history of the tournament. It looked like this:
What made the Maradona goal so special was the fact that he did all of the work himself. Compare it to this goal:
In this one, David Beckham played in a phenomenal ball from just over the half way line, and Ronaldo tapped in the goal.
A look at the score line would give equal credit to Maradona and Ronaldo for scoring the goal but give no account to the method of the goal. That is how last click wins attribution works. and it is flawed.
The danger of ignoring the “assist” is that the importance of different channels within the mix can be heavily understated.
The recent TagMan Study that was published in eConsultancy covered their analysis of actual sales and measured attribution based on influence suggested that generic SEO actually influenced around 14 times more sales than it was credited with on a last click model.
The challenge for marketers will actually be in creating a model that is reliable and specific to their business and actual marketing blend. It is also important that the model needs to be independent of preconceptions: A company that has invested heavily in one particular channel might attribute more influence to that channel than it actually has.
More Complex Systems
It is important to consider that a typical company sells more than one product, and a typical product range will cover a lot of different price points. The purchaser behaviour across a range of products will be completely different. If you think about a company like Wiggle Cycles, they have products that range from low cost components like handlebar tape through to high end time trial bikes that cost upwards of £6,000.
If Wiggle were to place a single attribution model across all of their product set, the actual influence of different channels for different types of product would be skewed. Too much weight would be given to influencer channels such as display for impulse buy products where first click is often the same as last click, while too little weight would be given for the high end products where user research is a much more important part of the process.
Unfortunately, for most companies, there is a big disconnect between the analytics data, and the ecommerce data about specific products that have been sold, meaning that under the most popular attribution technology in use at the moment, it is not plausible to create different attribution models across different product groups without compromising the website in some way.
Next generation analytics will need to be tied more closely into the actual fulfilment data base in order to provide marketers with more information about who is buying what, and what is influencing them. If you were able to determine that purchasers of one product were generally unaffected by anything other than PPC, whereas another product might be bought only after a consumer was heavily influenced by multiple visits to the website and exposure to different experiences.