The Authoritative Guide to Marketing Attribution

Insight

The state of marketing attribution in 2023 is dynamic, to say the least. 

Marketers remain dedicated to the quest of demystifying customer journeys and optimizing the allocation of their marketing budgets, using more sophisticated tools and analyses than ever. 

And, everyone is kind of waiting for Apple’s next hypocritical move in the name of privacy (but is just about chipping away at Google Ads’ supremacy and driving use of Apple Ads) to make marketers’ jobs harder. 

What a time to be online. 

In this article, we’ll provide some SEO’d fun. But we’ll keep it interesting. What’s to come:

  • Understanding and leveraging marketing attribution models (and when to actually use each)
  • Attribution in GA4
  • Marketing attribution windows + commentary
  • The marketing “halo effect”
  • Discussion of Marketing Attribution, ROAS, and LTV
  • Custom modeling and assisted attribution

[If you're just looking for an attribution solution, click here.]

What is Marketing Attribution?

Marketing attribution refers to the various logic used to assign a specific order or group of orders to a specific channel, vendor, subchannel, or media type. The ultimate purposes are to identify the marketing efforts and customer touchpoints that lead to sales (or conversions). In practice, marketers look to identify those marketing efforts and customer touchpoints that lead to the most sales. 

And, ideally, those that lead to the highest LTV

Marketers can use a variety of tools, metrics, and analyses (specifically, models) to identify their best-performing efforts and budget allocation. In doing so, they can answer questions like:

  • What channels do we spend the most (and least) efficiently in?
  • How can we better allocate our spend?
  • What is our ROAS by channel?
  • What channels contribute the most (and least) to revenue?
  • What channels contribute the most (and least) to LTV?

Understanding and Leveraging Marketing Attribution Models

An attribution model is a framework that assigns varying levels of “credit” (in practice, revenue or margin) to different touchpoints in the customer journey that have contributed to a conversion or purchase. Fundamental to accurate attribution modeling is having sufficient and accurate data.

Important Notes

  • Google offers multiple attribution models, but we caution against fully trusting an attribution model or attribution data offered by ad vendors. They have a tendency to favor and give more credit to ads they host. However, certain of Google’s out-of-the-box attribution models become quite reliable with data enrichment and/or further analytics.
  • Certain models aren’t as reliable or possible without advanced tracking software. To build more robust models of your customer journeys, we recommend investigating tools such as Rockerbox (if you run eCommerce on Shopify or Magento), Blotout.io, and fingerprint.com.

The Basic Marketing Attribution Models

The three simplest attribution models are first-click, last-click, and linear attribution: we’ll explain all three models and their use cases.

First-Click Model

What it is: A first-click attribution model assigns 100% of the credit for a sale to the first channel a customer who purchased interacted with. 

When it’s useful:

  • Understanding the efficacy of awareness-building or brand-building campaigns: a first-click is an important metric because it helps track reach. 
  • If a brand is in a customer acquisition mode, it is a simple way to track which channels are reaching customers who ultimately purchase. 
  • Answering the question, “How are customers first getting to my site?”

Last-Click Model 

What it is: A last-click attribution model assigns 100% of the credit for a sale to the last channel the customer interacted with before they purchased. 

‍When it’s useful

  • Understanding which channels are most effective to deliver the messages, offers, and content that convert a customer to purchase. Specifically, it can be useful to evaluate CTAs.
  • Often used after brands have a more substantial customer base.

Linear Model

What it is: A linear attribution model considers the multiple touchpoints in the customer journey and breaks down the credit into equal parts. 

What it’s useful for

  • For brands who want to consider every touchpoint and not bias touchpoints in any direction.
  • Providing a high-level indication of marketing strategy success.
  • When adding additional channels to support and nurture the customer journey, it is helpful as a baseline to track which combination of channels drive more customers toward a purchase.

Other Attribution Models

Of the following three attribution models—time-decay, U-shaped, and W-shaped—the first two are offered by Google (which, as we mentioned, should be taken with a large grain of salt). The last model, W-shaped, tends to only be possible via other software. 

Time-decay attribution model

What it is: A time-decay attribution model is a more nuanced version of linear attribution, giving more credit to the touchpoints closer in time to the sale. 

What it’s useful for

  • Fine-tuning your marketing strategy and figuring out where to shift or invest more budget to increase conversions. The time-decay model can help you understand the customer journey at the bottom of the funnel as the customer becomes closer to buying.

U-shaped attribution model

What it is: A U-shaped attribution model (also called a position-based or bathtub attribution model) gives more credit to the first and last touches, with less credit to touchpoints in the middle of the customer journey. Generally, 40% of revenue is assigned to the first and last channel, and the rest of the credit is divided among the remaining channels. 

What it’s useful for

  • Understanding which top-of-funnel and bottom-of-funnel touchpoints are most effective in the context of a longer customer journey with multiple interactions. 
  • Used if you place more importance on top of funnel and bottom of funnel channels.

W-shaped attribution model

What it is: A W-shaped attribution model is another version of position-based attribution that gives more weight to first- and last-touch and also credits a mid-journey touchpoint in the funnel. 

What it’s useful for

  • Understanding more complex campaigns and longer customer journeys with many different touchpoints. It helps you test which mid-journey touchpoints may be most effective in nurturing a lead to drive customers to purchase. 
  • Useful for brands with longer sales cycles that may require more touchpoints before customers decide to purchase. 
  • Answers the question: “Which mid-funnel touchpoints are most effective in moving customers faster through the funnel?”

Note: W-shaped attribution is more likely to be accurate with more advanced software.

Attribution + Models in GA4

Note: We have a dedicated article on GA4 data-driven attribution, which you can read for the full story. And, it really is a full story!

In GA4, Google introduced a new marketing attribution model, called data-driven attribution, while keeping previous models (i.e., first-click, last click, linear, time decay, and U-shaped).

What it is: Data-Driven Attribution is a dynamic attribution model that looks at your users’ source/medium/campaign combinations over their customer journeys and attempts to determine which are most crucial to the conversion. 

It uses a weighting system to distribute credit: combinations that were more important toward driving a conversion are weighted more heavily, and less important combinations are weighted less. 

What it’s useful for: Data-driven attribution is a powerful attribution system (that still shouldn’t be 100% followed blindly, given that it’s coming from Google), however it requires a thorough understanding of GA4’s Channel Grouping, Source, Medium, Campaign system, which is significantly more complex than Universal Analytics’. 

Defining Marketing Attribution Windows + Example + Analysis

What is an Attribution Window?

Most ad/marketing vendors have the concept of an attribution window, which is a particular time period when they are able to take credit that some interaction led to a conversion.

  • Google calls an attribution window a lookback window.

Vendors may take credit for orders even if a customer did not click through an advertisement on their site directly before they made a purchase. Most vendors have different attribution window choices along with a default option.

  • Facebook allows users to choose between 1 and 7 day view and click windows with a default of 1 day view + 7 day click. 

  • However, Google’s default window is a 30 day click window for acquisition conversion events, and other conversion events may have lookbacks of 30, 60, or 90 days [see above].

Attribution Window Example + Analysis

To illustrate how attribution windows works (and how vendors claim conversion credit), consider the following customer journey:

Fredmatilda Lunkgrove (we’ll call them “Fred”) came to your site on January 1st after clicking through a Facebook ad. Fred did not buy on that initial visit. 2 days later, Fred came back after clicking through a Google ad, but still did not purchase. Finally, 2 days after that, Fred typed in your URL directly and made a purchase.

Based on their default settings, both Facebook and Google are going to count this as an order/conversion that can be attributed to their ads. 

  • For Facebook, this individual clicked through an ad, and then purchased within the next 7 days. 
  • For Google, the purchase was also within the 30 day window.

But what if you were to change the windows?

If you have changed your settings to a 1 day click + 1 day view window for Facebook, then Facebook would NOT count this order as a Facebook "conversion." 

So which attribution window is “correct”? It depends on your business. Do you have a short customer journey that lasts only a few days? Or does your customer journey tend to take a week or longer? 

For lower priced items, you will typically find that a shorter window (1 day click + 1 day view) gives you the most accurate results. 

In the example above with Fred, it's likely that seeing and clicking on an ad in Facebook likely did contribute to their making a purchase. After all, it was their first interaction with your brand. But how much did Google contribute? And what if the Facebook click came after they visited your site through 3 other vendors?

Setting your attribution window comes down to product type, price points, and other factors that require a real understanding of the path customers take to finding and purchasing your products. 

The Marketing “Halo Effect” (Can It Be Quantified?)

Quite relevant to the topic of marketing channels’ impacts on customers is marketing channels’ effects on each other, and ultimately, on sales across selling channels. It’s a challenging topic: really, being able to quantify this “halo effect” is considered to be one of the “Holy Grails of Marketing.”

Brands, for instance, may know that if they’re spending across Google Ads, TikTok Ads, and Amazon Ads, and selling via Salesforce Commerce Cloud and Amazon Seller Central—but, they likely won’t be able to quantify the impact of, for instance, Google Ads and TikTok Ads on Amazon sales.

This lack of clarity on the interplay among channels leads brands to spend more on all channels in the end. 

Ultimately, the only way to accurately quantify the halo effect is with a holistic understanding of all brand performance, which you can gain with an omnichannel data + analytics platform.

Marketing Attribution, ROAS, and LTV

Fundamental to attribution for marketers is tracking ROAS and the revenue driven by each channel. 

While we agree that ROAS and Revenue (a little R&R?) are useful guiding posts for attribution modeling and analysis, we would like to put forward an important point in the overall attribution discussion: to track gross margin per customer by acquisition channel. That is, LTV by acquisition channel.

Based on the attribution modeling you’re working with, you will see revenue contributions per channel, and ROAS will show how efficiently you’re spending, at least on that same revenue basis. 

Neither revenue nor ROAS, though, are profitability metrics. This means that deductions are not factored in, which in turn means that channels that generate a ton of revenue may be barely profitable or in fact unprofitable. 

By leveraging attribution models, checking in on ROAS + Revenue, and looking at your LTV by acquisition channel, though, you will gain a significantly more complete picture of your marketing performance—and profitability:

LTV by First Order Channel Visualization from the Daasity app

To ensure that your investments in certain channels are ultimately paying off on a profitability basis, track LTV over time. 

Last Marketing Attribution Notes & Advice

When considering your approach, it helps to have the right mindset. That is, an open mindset. That’s because there’s no one “right” way to “do” attribution, and it varies by strategy and time.

Ultimately, it comes down to two major variables: 

  • Understand how the model(s) you choose work.
  • Ensure the information you can get from the model supports the sales cycle, your business goals, and your marketing strategies.

While you could argue that attribution is an “imperfect science,” it’s still important to use models to give you a data-driven marketing basis. Otherwise, without any or one-sided data (e.g., vendor-reported only), you’re determining your marketing allocation in the dark. 

Also, our experience shows that attribution models don’t have to be super complex, even if your customer journeys are.

You can get a lot of useful information from a relatively simple model if you understand how the different models work, your customer journey and your business goals. 

Marketing Attribution with Daasity

Daasity is a data and analytics platform purpose-built for consumer brands selling via eCommerce, Amazon, retail, and/or wholesale. We offer best-in-class marketing attribution analytics, by offering eight models out of the box:

This includes what we’ve called assisted attribution, which will attribute an order to each of the channels that had a touchpoint on the path to conversion but did not get last-click credit. Through extensive testing with our brands, we believe that last-click + assisted attribution provides the fullest, most unbiased picture of a channel’s impact. 

Brands can evaluate their attribution via a variety of metrics, including gross margin, gross sales, net sales, and more.

Furthermore, brands can customize their attribution logic, and rank order the attribution methods they value most:

…As well as what is considered to be a marketing channel vs. an ad channel:

For more information about marketing attribution analytics in the Daasity platform and the dozens of ways to slice and dice your data, learn more here.

And, to learn more about Daasity and how we can help you build a single source of truth around your data, please reach out to us!

May interest you