Attribution Management and Bidding in a Multi-device World

Attribution Management and Bidding in a Multi-device World

Customers are no longer on just one channel. A user might read about your products on their mobile phone, download your app to compare products, and then make a purchase using their laptop. And it’s not unusual today for a purchase to involve at least six trips to your website before the customer chooses to buy. How do you best allocate your bidding under these circumstances? With a thorough understanding of attribution management.  

In a Simplilearn webinar, Brad Geddes, author of “Advanced Google AdWords” and co-founder of AdAlysis explained the steps involved in implementing an effective attribution management strategy for your company. You can read a summary of the webinar below.


Imagine a person walks down the street and looks into a store window but then walks away. The next day, that same person comes back and goes into the store, picks up the item on the window that caught their eye the day before, but then puts it down again and leaves. On the third day, the person shows up and walks into the store and picks up some other items. Then on the fourth day, the person walks into the store and buys something. On which one of these days would you say the person converted into a customer? The first time they saw the store, or the item on the window? The day they went into the store? Or only when they finally carried something up to the cash register and paid for it? 

That might seem like a silly story to tell, but it’s essentially what happens with our e-commerce websites—although drastically simplified. And in the case of the brick-and-mortar store, the shop owner is at a distinct disadvantage because she doesn’t know what drove all of these interactions with her store. Was it the sign out front? A flyer in the neighboring coffee shop? A referral from a friend? A Facebook post? The colorful window display? She can’t be sure which marketing effort to attribute the sales to. 

Not so in the world of pay-per-click (PPC) advertising in which marketers can usually attribute user actions to conversions, to help them make informed decisions about where to invest their efforts and their budgets. 

Why Does Attribution Matter?

Attribution is a key component of successful PPC. Attribution helps you understand how to bid, especially when you have many different keywords and display ads, plus multiple channels and devices. Attribution enables you to sort through that clutter to determine what a visit is worth to you in a certain path. It helps you to understand what clicks are worth to you, which ones to bid on, and how to evaluate the situation when a user comes to your site multiple times. 

Attribution also enables you to dig deeper to determine the value of awareness, meaning that first touchpoint vs. the closing, i.e., when the customer makes a purchase. And it helps you to know how each channel is contributing to sales. It can, therefore, help you to figure out the budgets for each channel, as well as when it’s time to remove a channel. Most importantly, attribution helps you manage your marketing budget wisely.

The Attribution Models and Their Uses

Making the most of attribution—and therefore your marketing budget—requires an understanding of the various attribution models and how to use them. In his webinar on Attribution Management and Bidding in a Multi-Device World, PPC expert Brad Geddes spoke about the various aspects of attribution, starting with the various models and then delving into other factors you must take into account when making attribution decisions. 

There are several different attribution models, and each is useful in its own way. Most companies, however, pick an attribution model by letting the CMO choose, but that pick is usually based on opinion and not data. It’s worth your time to learn about the various attribution models so you can select the one that will best serve your goals. The models include last click, first interaction, linear, time decay and the position-based model. 

The last click model is a popular attribution model. Last click interaction means someone clicked on your link and then converted. Anything they did previous to that is not counted. So last click gets a 100% conversion value. In a long customer journey, it might not be a good bid method because it tells you nothing about the customer journey. What it does tell you is what closes the sale, the final interaction point that gets someone to convert. There’s a lot you can tell from this because these touchpoints are the ones you want to be selling on. This tells you what happens in the buy phase. Even if you don’t bid on it, it tells you what is closing sales. 

The opposite of this is the first interaction model. First interaction answers the question, what is the first time we have a brand touchpoint with this user who then later converted? This is the awareness stage, the top of the funnel. Geddes said that very large companies give a lot of value to the first touch because it’s the opening of the conversation. So if you’re focused on generating awareness and getting people into your funnel, the first touch is very useful. 

If the last click is at the end of the journey and the first interaction is at the beginning, what’s in the middle? The linear attribution model. The linear model tells you when someone comes to your site multiple times before they convert. It values each touchpoint the same which is not as useful as the first interaction or the last click, but it is useful when asking, is this channel even contributing? You can see if a touchpoint is moving people forward…or not. 

The time decay model is an attribution model focused on what closes the sale. It values the most recent interactions to closing more than first interaction, so it can help you focus on what is closing sales. 

The position-based model is focused on both awareness and sales. It’s a good model to start with if yours is a new company. It values the first and last touchpoints the most but doesn’t forget about the middle journey. 

Of these five models, some are better suited to certain types of businesses than others. Geddes gave a thorough explanation of why the last click model might be best suited to a market leader with little competition striving for conservative growth, while a first click attribution model might be the best choice for a new company that’s growth-oriented and facing stiff competition as the newcomer to the marketplace. This webinar will help you in determining which model is the right choice for your business. 

Using Data-driven Attribution Instead

In addition to these attribution models, you can turn to machine learning to determine attribution for you–if you have enough data–using Google’s data-driven attribution (DDA). If you have a lot of wandering pathways and a lot of conversions, this can work well for you vs. choosing or customizing your own model. It takes a lot of data to make this work, however. If you don’t have a lot of data, you’re better off choosing your own attribution model to start. 

Device-based Attributions

You also want to get attributions for different devices. If you look at attribution modeling, you’ll see how users are behaving across all devices. The mobile experience can be a poor one, leading marketers to see poor results and want to give up on mobile. But if you segment your channels by devices and then look at your attribution modeling, you might see you’re getting last clicks on desktop view, while most of your first clicks are on mobile. And that tells you customers are finding you on mobile but converting on the desktop–not that your mobile experience is poor. With this kind of insight, you can know how users are interacting with you on a device and can bid appropriately. And until you do a channel device comparison by model, you won’t fully understand the customer journey. 

Additional Benefits to Attribution Insight

Just like the person walking down the street who makes multiple stops in a shop before buying, users are typically going to your website multiple times, generating data you can use on audiences. If you’re doing attribution bidding, you have enough data to come up with a really good audience strategy with AdWords. In addition, you can use CPA or ROAS to automate your bidding, with Google handling it for you, but you need to have a thorough understanding of your attribution and modeling first. 

Not only that, but you must track everything for attribution to work well, even beyond devices. That includes assisting keywords and cross-channel influences, as well as offline touchpoints such as phone calls, in-store visits, chat and more. And once you are tracking all of this data, plus using the best attribution model for your situation, you’ll soon be using attribution models to maximize the efficiency and contribution of your channels. You can watch the webinar to learn how.

Recommended For You

Leave a Reply