At the turn of 2011, the buzz word ‘attribution’ began to garner attention as marketers started to realise value in customised attribution modelling. Understanding the true value of individual channels in a final conversion has always been an ideal strived for.
Reliable insight data allows marketers to understand the interplay of each channel, better enabling them to influence their end goals and increase efficiencies. What’s key, however, is that they can do this by allocating budget across those channels (or channel combinations) that drive results. With this in mind, it’s easy to see the flaws in last-click: a legacy solution with no real concept of applying credit where credit’s due.
With just over half of online marketers using custom attribution (72% of which use first or last-click),I’m sometimes left scratching my head wondering what the other half are doing with all their data.The old cliché ‘knowledge is power’ seems fitting about now; there are some justifiable reasons, the biggest being politics within marketing departments (different teams for different channels) and technological limitations (particularly of larger organisations that are heavily reliant on older tools). Of those who are using some form of attribution, only a tiny proportion had any faith in a totally accurate form of modelling being available. This is the result of the huge variables that go into assigning credit. What is the value of a social media click vs. a brand search vs. video engagement to the end conversion? Do these values change with product or category, time of day or season?
With the insights currently available it’s no wonder that once marketers begin to tackle the issue of Big Data, it presents big fears. Tools and teams are available to marketers through agencies and technology providers who work together to try to alleviate these problems by investing large amounts of time and money in statistical modelling.
Marketers, however, can approach the problem from a different angle. IgnitionOne coined the term ‘attruebution’ at a Jump event hosted by Econsultancy in 2012 when they spoke of the pioneering approach they had taken to tackle the above issues. What if you were able to assess channel engagement by assessing user behaviour online and linking that to the referring traffic source?
IgnitionOne uses standard online media metrics as well as an audience scoring algorithm to determine a prospect’s buying propensity by evaluating the past behaviours of converters and non-converters. This is a big step forward, as many technologies only consider the average 3% of converters typically seen on a website. The data is then used to evaluate the effectiveness of media interactions and provide customer-driven scientific weighting that eliminates the guess work in delivering suitable attribution profiles. With the change in focus from latency and exposure models, it provides a greater insight into the value that is generated with each media touch point.
The graph above demonstrates that although PPC renders the highest conversion rate, it also produces the least amount of user engagement. With this in mind, how valuable was this channel towards the end conversion? Highest engagement was driven by the social channels so what would be the result of reducing budget in this area to increase PPC spend?
Do you really need to attribute your data? The short answer is yes. If not now, sooner rather than later. It’s a bigger priority for certain verticals, and management teams will need to evaluate the percentage of their conversions that are the result of multi-channel interactions. From this they will have a better idea of the ROI from any media investment they make.
When you finally do decide to take the leap and move off the legacy last-click, have a real think about what you hope to achieve. By thinking outside of the box marketers can take advantage of potential insights and use this to drive their hard earned budgets into the areas that they know are producing real results. No more squabbling between internal teams about who should get what, when the data speaks for itself.