Ebiquity’s Head of International Effectiveness, Mike Campbell and Principal Effectiveness Consultant, Patrik Sahlin recently co-hosted a World Federation of Advertiser’s webinar on the topic of Total View Attribution.
The majority of major advertisers are again forecasting growth in digital ad spend in 2018. Yet the proof that digital media is providing a better return on investment (ROI) than other broadcast media channels, TV, radio press etc is not always present despite the fact that we now live in a world with an abundance of data.
There is significant confusion regarding the optimal way of measuring the sales impact of digital media. There are two principle techniques in the industry that are tackling this challenge: Digital attribution using cookie tracking is measuring digital media touchpoints and is linking these touchpoints to online sales conversions, however this approach is typically overstating the impact of digital media because it is not factoring in the impact of other marketing and sales touchpoints including offline media and it is also very often not possible to link the digital media to offline sales.
Meanwhile, Econometrics is scaling the impact of digital media’s contribution to sales but is not providing the necessary detail of understanding formats and publishers and search details
This webinar explains how you can develop a holistic attribution system that takes the best of these approaches to provide the correct overall scaling of different media channels whilst providing the necessary detail of digital media performance that digital marketers crave.
A recording of the webinar is available to download via the form below.
Mike and Patrik also answered a range of post-webinar questions from attendees…
How do you ensure the attribution modelling is unbiased?
It depends on the choice of the underlying mathematical methodology. Each methodology is surrounded by its own validation framework. We have chosen to use Bayesian statistics because of the statistical output that it generates. We are always happy to share this output with clients. I would suggest that if you are working or discussing attribution with any company who is not willing to share statistical outputs then the suggestion is that they are hiding something.
Shouldn’t Data Collection come before Modelling stage?
That is correct.
Could conversion pixels from Google and Facebook that are placed on the same page and fire concurrently be used to de-duplicate the converting audience?
In theory, yes, but that would create a bias in the data as the non-converting audience wouldn’t be de-duplicated.
What share between online and offline sales do you recommend should assess total view attribution?
As a rough rule of thumb if digital media is less than 10% of your media budget and your sales pattern then we would advocate starting with econometrics and seeing what that reveals. It may show that your digital media is inefficient and warrants a reduction in which case you may not wish to carry on to look at digital attribution in detail. Another way of thinking about this is to consider that the likely upside from digital attribution is generally above 10% ROI improvement, then a project costing $50,000+ would be easily warranted on a $1m digital spend.
Can you provide an actual range of costs in dollars?
The fees are highly dependent on the scope of the project, the number of conversion criteria, the scale of the data sets etc., but typical ranges are from $50,000 upwards for developing the first models and insights larger projects can easily run into six figures.
You were talking about usage of cookies for the Digital Attribution. How do you see considering upcoming GDPR in 2018?
Ebiquity welcomes GDPR as it enforces advertisers to take ownership of data collection. Media agencies and publishers are currently taking the self-appointed responsibility for data collection, but not all tracking is in consumers’ or advertisers’ best interests. There are a number of key responsibilities within GDPR that need to be adhered to. There has to be a customer opt-in agreement and there also must be no ability to identify the individual in any analysed dataset. I.e. the customer address/web identifier must be removed from the data set that is under analysis. These factors are both technically possible thereby enabling the approach to adhere to GDPR legislation.
Do you really evaluate the efforts to assess the data as one week long?
The Data Assessment is an audit of the advertiser’s current data tracking. The purpose is to verify that data tracking is in place before modelling and data collection commences. It can take longer than a week, but with many organisations, our experience is that it can be completed quickly. (Please note that the initial data collation stage typically takes longer than a week)
What would you recommend to an OEM in the automotive industry with a predominant share of Sales offline?
The starting point would be econometric modelling as from experience this will reveal the greatest optimization upside from your media planning. Digital attribution can then be used to give detailed results on digital media campaigns. There are technical solutions to linking online customer journeys through to test drive requests or configurator pages etc. through to offline sales as there is in many industries.