Read in a row, or jump to a section of interest:
It’s rare that a user visits a site, completes a shopping cart, and makes an order in one session. Especially when it comes to a complex product with a long sales cycle. Customers google the product and enter the site from the search, see ads, subscribe to social networks, read partner newsletters or blogger reviews – and all this together forms the client’s attitude to the brand. And therefore – influences the decision to buy. In this article, newage. will show you how to evaluate the channels that bring a customer to a brand.
What is attribution in advertising
In copyright, attribution refers to identifying the author of a work; In psychology, attribution is the attribution of a person’s behavior to some cause. In marketing, we are interested in conversion attribution.
Attribution in marketing is the attribution of user conversions to points of interaction with the brand. Based on the chosen attribution principle, the analyst or analytical system decides what role each individual source, channel, and campaign played in the conversion.
Touchpoints must be attributed in order to optimize marketing activities – to strengthen channels that lead to targeted actions and remove those that do not produce results.
There are several principles that allow you to evaluate the role of individual marketing activities in conversion. Which one to follow depends on the specifics of the company, niche and product. Consider several attribution models in marketing.
First Click Attribution
First interaction attribution assigns all value to the channel that first brought the user to the site.
This model is useful when entering the market and launching campaigns aimed at increasing brand awareness. If your goal is to fill the top level of the funnel and reach as many people as possible, use first-click attribution.
It is especially important to consider delayed conversions here: post-view and cross-device. For example, a user saw your ad on Youtube but didn’t click on the ad. And a couple of days later I went to the site from a branded query in organic search. A user wouldn’t google a brand if they didn’t see the ad, but the standard tools won’t take into account the view and will consider organic as the most valuable source.
We use Comprehensive Analysis to track pending conversions and correctly attribute first contact with a brand.
Last Click Attribution
In this model, all the credit goes to the last source of the user’s transition to the site. Last-click attribution is useful for businesses that rely on impulse purchases or during campaigns focused on simple conversions.
Let’s say your online magazine launches a mailing list and the target conversion of the campaign is to leave an email in the subscription box. This is a simple, impulsive action, for which the user does not need to be warmed up and courted in advance – once on the site, he will either fill out the form in the pop-up or not. In this case, you can close your eyes to past interactions and evaluate which channels led to the subscription.
Last Non-Direct Click Attribution
This is an improved, slightly cleaner version of last click attribution. If a person bookmarked the site and the last visit before the conversion was from direct/none, this attribution model will assign value not to the “direct” source, but to the one from which the user came before it. If the last click before the conversion was from any source other than direct, then the value will be assigned to it.
Linear model of attribution (Linear model)
Here, the conversion value is distributed among all touchpoints. This ad attribution is useful for companies with long sales cycles, where marketing accompanies the lead as it moves through the funnel.
The standard situation in the sale of primary real estate. The client talked to the sales manager and is slowly moving towards signing documents: periodically communicates with the manager, consults with relatives, collects some documents, looks at competitors’ offers. This can last for months, and at any moment the client can say: “You know, I changed my mind about buying an apartment.” To prevent this from happening, marketing continues to warm up the already brought lead, sends “signs of fate” to him in the form of remarketing, invites him to subscribe to the newsletter or social networks, releases PR materials that remove objections, etc.
And when the conversion has already taken place, the client signed the documents and made the first payment, you can look back at all the activities carried out and say: “Thanks to all these actions, the client did not break loose.”
The main disadvantage of such an attribution model is “leveling”. Both contextual advertising, after which a person left his contacts, and a meme that he simply liked on Facebook are considered equally useful here. But this approach allows you to notice and cut off activities that pass by customers and do not give a result at all.
Attribution based on interaction age (Time decay)
The closer an interaction is to a conversion, the more valuable it is. This model is used to take into account which touch leads to a conversion, but do not want to lose sight of the previous customer journey. This can be useful when organizing promotions.
For example, a candy brand is hosting a UCG content contest — take a photo with the product and get a prize. Participants have two weeks to publish, but it is immediately clear that the peak of activity will be on the day of the deadline. It is logical to give the most weight to the touch that motivated the client to participate, but at the same time track how he learned about the competition and warmed up to participation.
The focus of this model is the first and last points of contact. We find out which channel introduced the client to the brand and which motivated the conversion – they get 40% of the weight. And the remaining 20% is distributed between intermediate stages.
Such an attribution model is useful, for example, when introducing a new product to the market. If you are building product knowledge through media channels and at the same time running a conversion-focused performance campaign, use this model.
Data-driven attribution is available in Google tools. The system analyzes account information, interaction chains, conversions and draws a conclusion about the contribution of each channel to the conversion.
From the Google help it is not entirely clear how exactly the algorithm works and how it distributes the value of the conversion. But if you trust the system, you can test such a function. This attribution model is available to advertisers who have had at least 3,000 ad clicks and 300+ conversions in the last month.
Custom attribution model
If all of the above models don’t fit your business model and don’t fit your current campaign goals, you can set up your own value distribution. This option is for advanced marketers who understand what type of interaction is most valuable to the company. Also, custom attribution models can be useful for experiments and hypothesis testing.
Google Attribution: Tools and Data
Most of the attribution models listed are available in Google tools. Recall where to set up and test attribution models in each of them.
Attribution models in Google Ads
To view the attribution report in Google Ads, open the “Tools” menu, and in it, in the “Tracking” column, go to the “Attribution” section.
Attribution Models in Google Analytics
Attribution models and conversion value are configured in the Conversions section.
Here you can also compare the available attribution models and see the data in several sections. To do this, go to the section “Conversions” — “Multi-channel funnels” — “Model comparison tool”. And then select the models you are interested in.
This is a new service currently in beta testing. Surely many have already noticed it in Google Analytics and tested it.
This tool is all about conversions, touches, and attribution models. Here you can:
- view data in the context of all the above attribution models;
- estimate the time from the first touch to the conversion;
- find out the length of the conversion path;
- compare data from different attribution models, etc.