If you manage complex digital marketing activities or even simple Facebook Ads or Google AdWords campaigns, you probably know how important it is to be able to correctly evaluate the contribution of each traffic source to the conversions and turnover of your company or your client’s company .
The rule used to assign one or more traffic sources the “merit” of having generated a sale or, more generally, a conversion is defined as the attribution model .
One of the classic attribution models is that of the last click (or last interaction). In this case, all credit is given to the last source of traffic from which the visitor accesses the site (before making a conversion).
For example, if the visitor accesses the site X today from Facebook, tomorrow from an AdWords announcement and in two days he will make a purchase going from an organic search on Google, the latter will be considered as the source that generated the conversion.
It is easy to understand therefore that, by adopting this attribution model, all the traffic sources that focus on the final part of the funnel (see image below) will have an advantage, as they are aimed at an audience that has potentially a greater tendency to convert.
Instead campaigns instead of awareness , more rarely lead to an immediate conversion (purchase) by the visitor, but they can help to “transform” a visitor into a future client.
Making decisions based only on the “last click” attribution model is usually a rather short-sighted choice. This model, for example, could lead you to believe that all sales on your site are generated by direct access and therefore all the money you are investing in advertising is useless.
The attribution model “Last interaction” is obviously not the only one. Here are some popular alternatives:
First interaction :
it attributes the merit of the conversion to the first source of traffic that leads the visitor to the site;
Last non-direct interaction :
it attributes the merit of the conversion to the last source of traffic that leads the visitor to the site but excluding direct access;
the merit of the conversion is evenly distributed on all the traffic sources that contributed to its realization;
Temporal decay :
the traffic sources temporally closer to the conversion receive most of the merit;
U-shaped model (or based on position):
40% of the merit is attributed to the first and last interaction and the remaining 20% is distributed among the other sources that are in between.
However, the trend towards algorithmic attribution models , which are based on the automated analysis of the collected data, is becoming increasingly popular . This evaluation method does not use rigid evaluation rules, but adapts based on the analysis of the behavior of users who convert and those who do not convert.
Attribution models based on an algorithm assign a different value to each click that generates a visit to the site, based for example:
- the number of clicks needed for a conversion;
- to the pages displayed for each click;
- at the bounce rate of each click;
- to the micro conversions made for each click
Here are some of the data that emerged from the report:
As can be seen from the graph, the most common attribution models are those that attribute merit to a single source: the last in the case of Last-click, or the first in the case of First-click. This is not a particularly encouraging fact.
Except in rare cases, in fact, every time you attribute 100% of the merit of a conversion to a single source of traffic, you are probably doing something wrong.
From the table, which reports the opinions of the 145 company managers interviewed, it turns out that the algorithmic attribution models are those that receive the greatest consensus in terms of effectiveness , followed by the post-click.
Post-click attribution is not to be confused with Last-click. In fact, with the Post-click attribution it is possible to set the so-called “attribution window”, ie a time interval within which to attribute the conversion to that source. In this way it is possible to consider the conversions that take place within 1 day, 7 days or even 28 days (just to give examples) after the click on that banner or on that ad has been made.
In this short post it is impossible to completely exhaust this topic, but it is clear that the question of attribution models is at the same time complex and fundamental to correctly analyze their digital marketing activities.
The use of one model to the detriment of another can lead you to waste your budget on the wrong sources and maybe reduce it to those that instead bring you new customers.