With the advent of digital and e-commerce sites, new performance indicators have been created, which make it possible to measure the online performance almost in real time.
For slightly more than one decade, and since the emergence of site-centric and ad-centric analysis technologies, companies have been able to assess and judge the instantaneous impact of a marketing interaction, such as, for example:
This procedure for evaluating a marketing action as a performance measurement is a deterministic attribution procedure. With time, even offline media are becoming enriched with new performance indicators (KPIs), the duration of customer service conversations is measured, and the broadcast of an advert on TV channels is compared with the Internet traffic it generates. All these new indicators have been developed in order to assess the performance of marketing actions.
Far from being sufficient, the challenge is to succeed in interpreting them overall in order to answer the following questions:
Attribution, therefore, needs to be capable of comparing the data related to the exposure of an individual to an advertising asset, and the action resulting from it. In marketing terms the aim is to reconcile the ad-centric data and the site-centric data.
In the cases illustrated below: the journey was undertaken with a single digital identifier, and the journey can, therefore, be reconstituted:
In the analysis tools, we find a conversion after 2 paid touch points: a display advertising impression followed by a click on a paid search.
Conversely, if the user was accustomed to using private browsing or frequently cleaned their cookies, this would be a case of a journey with multiple irreconcilable digital identifiers.
In a digital environment the evaluation of the performance of the channels in relation to the actions recorded on a website (arrival at site/lead/purchase/etc.) is accomplished using two indicators:
The timing of the action accomplished after impression of the advert is an important element to take into account. It is called the attribution window. Setting an attribution window at 30 days means that, for a purchase made on 30 September, only interactions with the brand which took place after 1 September will be taken into account by the model. The duration of the window depends on the type of interaction considered, and on the buying cycle specific to the company wishing to analyse its data.
For example, an operator selling motor vehicles or real estate will need to collect data over a period which may range from 6 months to one year. Whereas an operator offering consumer goods involving less commitment may obtain significant/ relevant results with a much shorter window of up to 7 days post-click and several hours post-view. This ad-centric and site-centric reconciliation depending on the attribution window requires the use of a dedicated attribution solution, or the journeys to be reconstituted in a full-stack Adtech/Martech ecosystem.
There are three goals of attribution for an advertiser:
An attribution model common to all parties is used as a basis for comparing the results of each partner in a standardised manner. Deduplication of the conversions and calculation of the performance of the marketing system incorporating several channels can break "silos management" views and provide the opportunity for an overall strategy to be managed.
An attribution model is above all a solution which enables one to better understand the role of each lever in the journey of a user, and their performance, and to identify tangible routes for optimisation. Various analyses can be made with the aim of understanding the behaviour of an individual in accordance with their interactions with a brand. These analyses enable the role of a lever in the journey to be identified, and can give rise to intra-lever optimisation solutions (for example an increase of the bids for a keyword, generic search) and inter-levers optimisation solutions (the discontinuation of a branding format in order to pre-qualify individuals).
The lessons resulting from the analysis enable decisions with a direct impact on the budgets to be taken subsequently. Indeed, conducting an analysis of attribution of the various marketing channels can enable the distribution of the investments within a media mix to be rationalised according to the real contribution of each channel. They also enable more enlightened arbitrages to be undertaken at the end, aiming to maximise the ROI of the activation campaigns.
This introduction is an extract from the chapter dedicated to Atribution & Measurement of Media Performance in the Yearbook 2019 (Converteo ADLPerformance) ; click below to download the long version: