Cracking the attribution code for digital marketing

If you work in marketing and you’ve not been bombarded with the term "attribution" in recent years, count yourself lucky. Everywhere, the talk has been on "attribution challenge" and how they can bring some science to bear on the "attribution puzzle".­­

Analytics 2015 takes place on 18 June at the Suntec Convention and Exhibition Centre Singapore. Visit the website here for more details. 

The problem modern marketers face revolves around one key challenge; how do you measure the effectiveness of marketing activities where consumers are exposed to? In particular, people are using this term in relation to the digital world (though, as marketers would know, this is not a digital challenge solely).

In the recent past, attribution was about helping digital marketers know what was effective and what was not in the areas of display and search. It now is laying claim to a wider sphere of interest covering email, affiliates, mobile, video and social media, and other digital marketing channels, such as price comparison and aggregator websites.

That might seem like an essential part of any marketer’s toolkit. Yet a mid-2012 study by Forrester Research indicated that 44% of interactive marketers did not have any processes at all in place to assign credit to their efforts.

Among those who do, by far the most common approach used for digital attribution is known as "last click". This involves crudely awarding the whole credit for a sale or other desired "conversion event" to the last click that preceded conversion. It has obvious limitations and inadequacies, and is frequently held up to criticism, not least because it usually underestimates the role played by many online influences such as display advertising.

For branded campaigns where the call to action is not immediate, "last click" clearly fails to give a realistic picture of the full effects of the various strands of marketing activity.

Happily, it is possible to improve on this state of affairs, using analytical techniques that are commonly used to measure the effectiveness of offline marketing activities.

An online consumer electronics retailer I know improved on last click by using econometric analysis to tease out and examine the influence on consumers of display, search and affiliate marketing activity.

They found some interesting results.

When looked at on a last click basis, it seemed as if almost all the value of their online spend was attributable to online search and affiliate activity. Given the company was investing in banners, videos and site takeovers, it was important for them to know whether online display activity was worth the effort.

The company came up with a clever approach. They grouped the click data into five time segments for each of three categories – display, search and affiliate – and used these five time segments as the basis for their analysis.

This allowed them to isolate the impact of each channel at a given point in time prior to purchase and prove, for example, that display advertising was having a positive effect.

Display advertising was exerting this influence early in the purchase cycle. And it was actually having nearly twice as much impact on purchasers as other methods of analysis had suggested.

This is good news, obviously, for media owners. It is also good news for marketers, who had previously more or less had to take it on trust that display advertising was worth paying for.

What analysis like this proves is that, although simple to measure, unfortunately last click is not good enough. It substantially overrates the value of affiliate and paid search activity (by margins of 21% and 28% in this example) and will mislead marketers into sub optimal investment decisions.

The author is Glenn Granger, CEO for marketingQED.  marketingQED is the Gold Sponsor for Analytics 2015.

Glenn will be speaking at the Analytics 2015 conference along with speakers from Luxola, OCBC Bank and more. 

To register, please click here or contact Joven Barcenas at jovenb@marketing-interactive.com or call +65 6423 0329.