“Half my advertising is wasted, I just don’t know which half.” – John Wanamaker (1838-1922), Retailer, Philadelphia
Attribution has long been a ‘hot topic’ for anyone concerned with understanding how their marketing spend, particularly their digital investment, works. However, for a method that promises to better measure marketing performance and improve ROI, adoption appears to have been slow.
Driven by the output of digital measurement systems (typically the ad-servers) most businesses continue to operate using the ‘last-click’ model. However, this ignores all prior events along a consumer’s increasingly complex path to purchase. The sentiment of Wanamaker’s observation remains as pertinent as ever.
Clients are now beginning to embrace multi-channel attribution (MCA) to fully understand the effectiveness of their marketing budgets along the consumer journey. MCA has a number of different interpretations; the most common being the analysis of digital only channels towards a digital only outcome. However, as attribution evolves in its level of sophistication, so does its scope and definition. We explore a number of key themes, such as cross-device attribution and attribution integrated with econometrics.
At its simplest, MCA attributes a fractional credit to each event within the consumer journey that has an influence on the target transaction. This enables a value for each event to be calculated. Budget allocation can then be optimised accordingly for future plans. These basic attribution scenarios can be run using the software from ad-servers; for example path to conversion reports within DoubleClick For Advertisers (DFA) reporting. These approaches are reliant on subjective human input to establish the ‘rules’ for fractional distribution of credit.
More advanced MCA is based on a model where the data talks: an algorithmic approach, where an engine, not a human, generates the coefficients for each event’s influence. This approach works on a more granular level, taking into account the impact of specific attributes for each touchpoint, such as site, placement, creative and keyword.
Those advertisers who will reap the greatest benefit from attribution to drive smarter, data-driven investment decisions, are the ones that are exploring, learning and acting upon the following five developments:
Smartphones and tablets are delivering more and more actions in the path to conversion. However, most advertising tracking mechanisms rely on a cookie-based approach that does not exist on mobile.
A better solution is needed to accurately track impression and conversion data across all screens, including mobile devices. ‘Bridging’ technology is being developed but the longer term win may well lie with cookie-less alternatives. A single view of the customer is what Google’s AdID sets out to achieve and others (e.g. TagMan) also have their version for a single identifier.
Other platforms such as Facebook and Amazon present the opportunity to track and connect users and their actions across multiple devices through a connected login. Data management platforms can also come to the fore with more advanced mapping techniques employed to ‘join the dots’ for a single customer view.
Connecting a ‘bottom-up’ digital attribution to ‘top-down’ econometrics will be a further win on the journey towards a total, holistic view of marketing activity performance. Econometric market mix modelling (MMM) can highlight the ‘lift‘ from base sales that digital activity achieves and hence sharpen the attribution output towards what drives incremental gains.
Additionally the coefficients for digital performance can be fed into the MMM at a much more granular level, which will provide a more robust budget allocation recommendation against all conversions.
Efficiency gains will be maximised when as many parts of the model can be automated as possible. This applies to data collection and organisation, through to the connection of output seamlessly into buying platforms such as AOD or search bid-management tools. With automated implementation performance results stimulate a dynamic learning engine in the model.
4. Viewability & Fraud Detection
A further benefit of attribution may be the ability to expose fraudulent activity and verify that ads served have been seen by human eyes. It is widely acknowledged that online bot traffic exists and it is difficult to detect fraudulent from legitimate impressions, clicks and conversions. This has an obvious impact on measuring the success of campaigns. Whilst there are a number of specialist companies establishing their credentials in this area (and the ad-servers are also making progress to tackle the issue), all and any focus on this topic is a positive for the industry.
5. Attribution acquisition
In the last month, both Google and AOL have strategically acquired leading attribution specialists (Adometry and Convertro respectively). This will certainly help place attribution even further up the agenda and should be a positive for the industry where the focus is on qualifying the value of marketing activity.
We will no doubt see integration with their existing analytics offerings; for example Google Analytics Premium leading to deeper consumer understanding through the journey to site and the experience on it. The jury is out regarding how ‘independent’ these models will be in the future when commanded by a media owner with vested interests.
However, the future of attribution modelling is certainly positive thanks to its rapid development, with technology helping us achieve much more than is possible from simple rules based attribution models. Advertisers will require skilled specialists to navigate the way forward. WE can expect exciting times ahead.