Digital Strategy

Giving better attribution to your Marketing Measurements

Going into the 2018 Spring and aspiring more than before to make disruptive marketing campaigns ? As a marketer, I understand the excitement of being at the front row of the frenzy scene of voice recognition, programmatic ads, working with micro influencers and testing AI powered solutions to create content and provide tremendous website experiences to increase your conversion rates…But yet,  are you still relying on the last click model ? It’s time to reshape the way you measure your overall marketing spendings across all channels to evaluate your real ROI and take strategic decisions accordingly.
What it is said on the field on this specific topic ? Listing of existing marketing attribution models and tools as well as some examples of companies who decided to implement models measurements will be provided to you in this article, leaving besides detailed demo calculation which rely later under your jurisdiction.

Figures and quotes to begin

Back in 2015, a Google blog’s post entitled The path to better measurement : Analytics and attribution already set the tone  « Think of attribution as the peanut butter to analytics’ jelly. Yes, each is great on its own, but for many, they’re even better together ».

As stated by a French source (L’ADN), 2017 was the year of « Predictive Marketing » with the culmination of testing and development of new algorythmic attribution models.
In fact, changes in digital spendings have been observed in the past year among large-scale companies : P&G cutting off $100 million dollars eliminating unprofitable ads and seing no decrease in sales, whereas L’Oréal did the opposite increasing its expenses to maximize its digital returns (as detailed in the Fospha’s article : Top 6 Changes in Marketing Measurement that will Impact 2018).

How will the 2018’s measurement landscape look like? It will be the year to change your digital budget, because according to the Demand Gen Report’s 2017 Marketing Measurement And Attribution Survey Report : « 91% of B2B marketer agree that marketing measurement and reporting is a top priority for their organizations » and « almost a third of marketers do not even measure their contribution to revenue, and fewer « 14.8% say they contribute to more than half of company revenues ».


Existing Attribution models and brand measurement examples

So based on these facts and as detailed in my previous article : Is Customer Data Platform the new DMP ?, we can confirm that the marketer’s graal is to get an holistic view of the consumer journey by adopting unified plateform technologies leveraging structured and unstructured data and integrating attribution modules (measurements based on trackable online interactions like clicks) to exact audience targeting and optimize the overall media ROI.

The last click model was still used by 79% of french advertisers for assessing their marketing performances (2014’s IFOP study) . This single touch model, gives all the credit just to the last touchpoint of the consumer and tends to overweight the paid search lever among other things. Only a part of your real marketing cost is evaluated, you may not have a margin but instead an overall loss. To understand the limits of the last click model here a thorough and clear explanation given by Jerome Sutter’s slideshare : Le Last-click, c’est paléolithick ! 

New attribution schemes called Multi-touch Attribution Models are the measurements to be used in 2018 ! 47% of marketers using MTA said that they were using the right model (according to the 2016 State of Pipeline Marketing Report). Here you will find the different models that you can setup if you have a Google Analytics account :

Linear : assign an equal share in conversion to every channel. It is often used for long branding exposure

Time Decay : time-based giving more credit to recent interactions than older ones : shorterm.

Position based : greater conversion value is given to both first and last click (40% each) with 20% that is attributed to the rest in between middle

Custom attribution model : to create your own hypothesis and test it

For instance, AccorHotels has declared in June 2017 that the group was tending to not use anymore the last click model to be in line with there ecommerce acquisition strategy. They were using at the time the Last Click Model with an attribution window of 30 days and on the Display part the 48 hours Post View Model (Think with Google’s article Comment AccorHotels pilote sa stratégie d’acquisition e-commerce ? »).

The International Jewelry brand Pandora decided to increase its digital investment by 46% in 2016, therefore in order to follow that new strategy, the company had to change the way ROI was calculated to optimize media spending. They have accordingly started an attribution project in collaboration with the Mazeberry’s Attribution Solution : performing A/B testing, activating 18 levers, and getting a 360° vision of their marketing mix. They multiplied their profitability by 6 for the Retargeting, by 2 for acquisition emailing and by 4 for Display RTB media ! (Pandora’s business case available here).


About metrics and publisher solutions

Ready to use these models ? There are different web publisher solutions on the market that offer attribution sections such as Commanders Act, known as Tag Commander with its Mixcommander product, AT Internet with its Channel Optimizer tool, Convertro providing MTA and mobile attribution solutions, Impact Radius platform giving omnichannel insights with its attribution reporting. And of course the Google free and easy to use Google attribution modelling solution whom has been launched in may 2017 and still is in beta version (the advanced and paying version is Google Attribution 360).

As already explained in 2016 in the Forrester Wave™: Marketing Measurement And Optimization Solutions, Q4 2016 Forrester’s research recommends to adopt a unified marketing impact analytics UMIA » it goes further as it includes the impact of all marketing media and efforts as well as other more economic variables such as variation in prices.
So there is a need to still use independant metrics that can be complementary to not solely rely on a one side vision of your performance metrics !

Additional challenges to consider

Measuring your cross-chanel and cross-devices performances is challenging as it’s difficult to fully track print left by consumers on their journey to purchase your product. Brick and mortar businesses with online and mobile application for instance, can face difficulties linking their online and offline activities to know exactly which online channel was used to drive in-store revenues.  Wifi, bluetooth beacons but also.. word of mouth, reputation and other factors come into play to influence the final purchasing act. Google and Salesforce have recently joined their forces to enable GA360 clients to « integrate online and offline customer touchpoints » as Mark Benioff, CEO of Salesforces (remark compiled in the previously cited Fospha’s article : Top 6 Changes in Marketing Measurement that will Impact 2018).

Organizational structures can still slow the implementation of these new metrics, as according to the research undertaken by the Data & Marketing Association and the Winterberry (stated in another Martech’s article Forward thinking: Reshaping how Marketers View Campaign Attribution) : « More than 35 % of surveyed marketers believe inherent organizational structure issues the tendency for companies to deploy little to no unified strategy to oversee marketing performance complicate attribution efforts ». Furthermore, GDPR is more than ever here to tickle you with its restriction on consumer data and the way we collect them (clicks collected through cookies), a reminder provided in the The disruption of attribution is coming MarTech’s article. And on top of that, which place will take AI and intelligent agents on the customer journey ? How will machine learning will attribute a sale when the purchase will be made by algorythms and not by the consumer itself ? (mentionned in the following MarTech’article).


Takeaways for marketers : There is a multitude of attribution performance measurements.(tons of articles like Beginner guide to Google Analytics Modelling or how to do Conversion Attribution modelling are exploring the subject extensively). Either, you have more the detective Del Spooner (Will Smith) profile in I robot mistrusting innovative technologies or on the contrary you are too much engage falling in love with an AI as Theodore Twombly (Joaquin Phoenix) in Her, do not go nuts and take time to think about what its really bests for your business to increase conversion. Tests measurement models and choose the one adapted to your audience and your strategic objectives to be better data-driven focused. And then time for testing, implementation and constant experimentation ! A wonderful journey that never ends 😉

About the author