The Fundamental Guide to Attribution
Written by Kristin Vick, Account Manager and Stephanie White, Sr. Account Manager
There are more ways than ever to track and measure where your audience is coming from and how they are converting. However, are you tracking all performance fairly? Do you have enough data to make actionable decisions? We all know attribution is key to understanding and optimizing your PPC methods (and beyond), so in this whitepaper we’re going to walk you through key tips and tricks to perfecting your attribution strategy.
How to Use AdWords Attribution Models
First, we’ll discuss the recently introduced AdWords attribution models and how you can give the right amount of credit to each ad, click or keyword that led to a conversion.
With all the information available to consumers these days, it goes without saying that many will usually visit your site more than once, and will undoubtedly click more than one of your ads before they convert. How do you know which of your marketing tactics are working?
We will talk about building and applying attribution models so you can give credit, and know how much credit to give, to each click utilizing the following models:
- Last click
- First click
- Position based
It is safe to say that a majority of advertisers are most familiar with the “Last Click” model in AdWords, and probably already use this model without even realizing it. Using “Last Click” gives all the credit to the last clicked ad and corresponding keyword that led to a conversion.
For all my sports buffs out there, it is basically saying it doesn’t matter who passed the ball. It doesn’t matter how many people assisted in helping you get the winning touchdown. He who scored the point gets all the credit.
The benefit is that it is the easiest model to apply, however, you might be ignoring all of the other clicks your customer has made along the way. Using a more integrated attribution model will not only allow you to fit a better model to business, but will also assist you in reaching customers earlier in the conversion path and improve your bidding strategy.
Our own Cassie Oumedian said that Last Click model is dead. “If this advertiser made decisions based solely on the last click CPA by channel and cut back on Display, Social and Paid Search, their conversion traffic would tank. Last click is dead, people. Straight up, “Red Wedding” dead. Let’s move on.”
Adwords would also agree that they are moving away from Last Click and suggesting Data Driven Attribution (DDA) or Position Based if DDA is not available. Your account must have enough data acuuminated before DDA is available as an option. In that case you might see a message stating “You currently don’t have enough conversion data to uses “Data-driven.”
If last click goes to the very last click, then first click is the exact opposite. First click attribution gives all the credit to the first clicked ad and corresponding keyword. Going back to the sports metaphor, whoever started the play, gets all the credit.
For example, if someone started searching for your product or service and clicked on one of your Non-brand ads and then leaves your website. Later, they come back and search for your brand name and click on a Brand ad. This would mean that the Non-brand keyword would get all the credit. This may be useful for giving Non-brand credit, but it doesn’t really do a great job of telling the complete story.
If first and last click models are too finite for your business, you might consider linear attribution. In this model, all clicks that lead to the conversion are given the same amount of credit.
Linear attribution allows for a team oriented model. All who played in the game get equal credit for the win. For example, if when you reviewed a linear attribution model and 30% of the middle of the funnel converts, you should consider building a +30% revenue into your model for assisted conversions.
Time decay offers a unique perspective, especially if your business has a longer conversion cycle. Time decay allows a weighted attribution, meaning clicks that happened closer to the conversion are given more weight than clicks that happened further away from the conversion.
AdWords has a 7-day half-life built into this model. If a click happens 8 days before a conversion, it is given half the credit as a click would receive if it happened the day before a conversion.
Position-based attribution does not mean the average position of an ad, but rather the position of click on a timeline. As an example, the first and last click ads and corresponding keywords are both given 40% credit, while clicks that happened in between the conversion path window are given 20%.
As I mentioned earlier, if your account does not have enough data available you might not be able to use the Data-driven model. In that case, this would be your next best model. This would likely give additional credit to your Non-Brand campaigns.
One thing to consider if you use this model, you will begin to see partial credit given to campaigns and keywords. For example, in the screenshot below, you will see the total conversion volume for one Non-brand campaign shows 170.43. This partial credit shows that this campaign is part of the conversion process, but does not always result in a conversion. However, this great example of why Last Click may not be as accurate next to the other models.
While this model is still in beta, it is worth mentioning. Data-driven attribution models automatically give conversion credit based on past data performance. If you have a new account or don’t have a ton of traffic, this model may not be available to you yet.
As mentioned, If this model is unavailable for your account. You might consider testing the Position Based model instead. However, if this model is available you may want to use it instead of Last Click. Since we often find that Non-Brand drives Brand conversions, this model may offer valuable insights into how the higher funnel terms are impacting your paid performance.
Using these attribution models will help shed light and identify if certain keywords, ad groups or campaigns are contributing to overall revenue even though they may not be bringing in direct conversions.
It is definitely time to rethink the attribution models you are using for your conversion tracking in Adwords.
How to Use the Attribution Model Comparison Tool
Understanding your customers’ buying journey is a complex task, but it is worth exploring to understand how they typically shop. Discovering how long your customers take to convert can help you verify that you are measuring your conversions properly. Customer behavior may vary depending on product prices.
Inside Adwords, under Tools > Measurement > Search Attribution > Attribution Modeling you will be able to select different attribution models to compare against your existing model.
For example, in the above screenshot, you can see that Data-Drive is available for this client and the conversion volume would increase for Brand and Shopping Desktop. The Shopping Model campaign would lose 21% of the conversion volume. This this current model, the mobile devices would get 100% of credit for the purchase. Losing conversions after switching to Data-driven just tells us that customers may be starting their journey on Search or Desktop before converting on their mobile device.
Multi-Channel Funnel Top Paths
Reviewing the Top Conversion Paths is a great way to determine if customers take longer to make a decision. In the first path in the image below, two visitors started with paid search and then visited the website directly six more times before converting. This data can be important in learning if your paid search is assisting conversions through other channels.
In one account, we can see more often than not the customers visit the site multiple times before converting. In fact, approximately 76% of the customers placing transactions visited the website more than one time.
One thing to keep in mind, is the Multi-Channel Funnel (MCF) is the only report attributes credit differently for Direct traffic compared to the other reports. In the above example, we can see the Direct has been given full credit for several different sessions. In other reports, the Direct visit is dropped and credit is given to the channel preceding that visit.
In the Multi-Channel Funnels reports, if direct traffic (i.e., a user used a bookmark or typed your site URL into his or her browser) converts, the conversion is attributed to the Direct channel. This differs from other Analytics reports in which the conversion is attributed to the previous non-direct campaign or source, if there is one.
For example, in other Analytics reports, if a user enters your site via a referral, then returns “direct” to convert, the “direct” source is ignored. Instead, the referral gets credit for the conversion.
In Multi-Channel Funnels, the “direct” source is not ignored. The Direct channel gets credit as the last interaction before the conversion, and the referral is counted as an assist interaction.
How long does it take your customers to complete a purchase or submit a lead? Understanding your customers’ behavior can help you determine which attribution model you should be using for your paid advertising.
In the above screenshot, we can see that only 41% of revenue is attributed to the initial click. The rest of the revenue is dispersed between day 2 up to 90 days out. The products prices are high for this client and so it takes customers many visits before they choose to make a purchase.
This type of data can also help you determine if your conversion window is set correctly in AdWords. In the MCF and Time Lag examples above, the credit for most of these conversions would not be given to the Paid channel. Without this attribution information, you could make decisions that negatively affect PPC decisions or hurts performance in other channels.
In AdWords, under “Tools > Attribution” you will be able to see additional details about how long it takes your customers to convert. The default view includes all types of conversions. You can select a specific type of conversion, so in this case, we wanted to see additional information about how long it takes customers to make a purchase. You can adjust the history window to include 30, 60, or 90 days’ worth of data.
This client’s customers take on average 9.37 days to make a purchase. Only 45% of revenue is attributed to that Last Click. The reason for this length is because their products are expensive and require adequate space to install. Purchasing these products requires thought and would not be considered an impulse buy. These customers take longer to complete a purchase. If our conversion window was set at 30 days, some of our conversions would likely fall outside our window and would not show in our data.
In contrast, one of our lead generation clients’ customer journey is much shorter. Nearly 97% of customers convert immediately and there is very little time lag. Their services are job oriented and customers are likely to convert right away after clicking on their paid search ads. If you are tracking more than one type of conversion action, like phone calls and transactions, you might review each conversion action independently because they can vary. It may take customers less time to pick up the phone to ask a question than moving forward to submitting a lead or making a purchase.
You may want also look in Analytics, under “Conversions > Multi-Channel Funnels > Time Lag.” This view can show you the average for all channels or you can specifically select Adwords. In this view, we changed the conversion action to transactions and increased our window to 90 days.
In this example, the data shows us that it takes the customers longer to convert than what AdWords attribution had shown us. Only 32% of customers complete a transaction within the first 10 days. Over 37% of the customers who completed a transaction, did it with between 61-90 days after they initially clicked on our ad. It could take customers between 2-3 months to make a purchase from the initial click on the paid search ad.
Another item to take into consideration, if customers take longer to complete a purchase, sometimes their path is more complex because they visit the website multiple times through different channels. Many times, we have found the direct channel is getting credit for paid-initiated conversions. For one client, we found that 75% of our paid initiated conversions were converting through the direct channel.
Another issue we discovered was our revenue and CPA data were not lining up. In Analytics, we could see that Texas and California were in the top locations for revenue.
Our client had noticed that he was receiving a high volume of orders from California, Florida, and Texas. However, when we reviewed AdWords for non-brand geography, we found these locations had the highest CPAs. Normally, we would pull back on these locations, but we had a theory that AdWords was not telling the complete story.
Our theory was that non-brand was assisting in driving brand and direct conversions. So, we decided to increase our location bid adjustments in AdWords for the warmer states and monitor Analytics performance. What we found was that transactions and revenue increased for most locations.
We decided to test switching the attribution modeling from last click to first click. Two weeks later, when we compared the performance we found that brand saw a 62% decrease in conversion volume while non-brand saw a 45% increase in conversion volume.
This data tells us that non-brand was likely contributing to the brand conversions. Our goal with switching from last click to first click modeling was to gain additional data in order to optimize our non-brand campaigns more efficiently.
This is why we recommend digging deeper into your customers’ buying cycle to make sure you are not making decisions off incomplete data from AdWords. This customer had such a high volume of direct conversions in Analytics that were initiated by paid search.
Just When Google Attribution Couldn’t Get Better…
Google is stepping up their already impressive reporting game and combining their reporting tools to give you one solid platform that incorporates Analytics, AdWords, and DoubleClick.
If you own a PPC strategy for your client or company, chances are you use more than one channel to reach customers online. Google has said they understand this and want to make Google Attribution to be a one-stop shop for evaluating marketing campaigns. “By creating a tight loop between strategy, ad spend and feedback, Google aims to make the tool attractive to marketers that feel last-click models don’t sufficiently explain customer behavior,” says John Mannes of TechCrunch.
In the real world, most marketers know that conversions should be credited to more than the last touch point. A customer probably saw video ads, banner ads, emails, social ads, and other materials. All of these worked in consonance to drive conversions. Why, with all of the data available would you still attribute the sale to the last ad seen?
“We really focused on simplicity for Google Attribution…you basically connect your accounts, import your data, choose the attribution model that works best, then you can export the results through AdWords. You can immediately optimize your campaigns through the attribution models. Something that took months to do, we’ve reduced it down to minutes,” said Babak Pahlavan, Senior Director of Product Management, Google.
Using Google Attribution allows digital marketers, especially PPC marketers who are looking for ways to simplify their reporting and insights, an easy way to evaluate and make changes faster for clients.
Google Attribution aims to solve three problems for marketers:
- How to properly credit marketing sources and medium
- Easily see upper and mid-funnel interactions
- How to easily inform bidding decisions based on full-funnel attribution data
Marketers can easily see the conversion impact of keywords and ads based on multichannel and multi-device conversion paths data. By automatically sending modeled conversion data into AdWords, automated bidding can take upper and mid-funnel contributions into consideration. Again, this obviously has a benefit to Google as well.
One major question to answer is how is this different than what’s available now?
Currently, attribution within AdWords doesn’t offer any insight into the connection between ads and marketing efforts on other channels. Google Attribution provides the cross-channel context that is missing in AdWords attribution. In Analytics, Google Attributions offers more touch-points and more in-depth analytics.
Key benefits of Google Attribution coming to marketers include:
- Data-driven multi-touch attribution models
- Measure the full journey using Google’s privacy-friendly cross-device graph
- Ability to act on insights through smart bidding integrations
- Measure value of clicks/impressions with data-driven attribution through machine learning
Currently, Google Attribution is in beta and will be rolling out over the next quarter to accounts in the US. While marketers are waiting, advertisers can look at data-driven attribution modeling in AdWords, if available.
Supercharge Your Brand’s Bottom Line with Social Attribution
Social advertising can be difficult to measure, but is a key part to any integrated digital marketing campaign. With the continued push towards pay-to-play social postings, here we will focus on paid social advertising and why it is important and how it may not always drive direct results, but ultimately improves the bottom line with multi-channel attribution.
ATTRIBUTION AND THE MULTI-TOUCH JOURNEY
Attribution is a way of knowing the value of any campaign and its involvement in the buyer journey. Whether B2C or B2B, multiple steps take place in the journey and funnel that the user took to make a final decision. Social Advertising is no different, but is not always the direct driver of sales.
However, throughout the buyer journey, various ads influence users to either learn more, become a lead, or ultimately make the purchase. The typical buyer journey looks different in each industry, but more often than not, a buyer’s journey has more than one stop, and will be affected by more than one ad, one campaign, and even more than one marketing source.
Let’s assume a typical buyers journey in the health and fitness industry. By looking at the path and overall interaction model, we could see that social is the first touch. This is usually because someone is looking at how they should work out, looking up beauty and fitness articles and seeing what their friends are doing to stay fit and healthy.
This can make it difficult to understand or drawback direct correlation to social campaigns having a direct effect. But in this case, without social they would not have taken the next steps of being more willing to engage with display ads or do further research on a brand, through organic searches, and then moving to generic or brand paid search based on their mid-funnel research.
In this model, let’s assume there are 7 total steps a user takes before completing a purchase:
- Social Interaction
- Display Click
- Organic Search
- Referral/Paid Search
As we can see, there are a lot of steps that define the path to purchase, or in this case, membership sign-up. Social is not only the first interaction but a multi-assisted interaction put together by our own internal resources and overlapping the Google buyer-journey. With the continued move towards micro-moments, each individual touch point creates value towards a potential customer interacting and purchasing from your brand.
HOW DO ATTRIBUTION AND SOCIAL WORK TOGETHER?
Multiple social platforms can be attributed to a final conversion/sale. Pinterest, for example, could be used for workout ideas and health articles, so by creating Pinterest awareness campaigns, as an example, you can direct readers to a website blog that outlines how muscle group classes can tone or help them lose weight. It could also be articles specifically referencing diet and the influence and partnership with scheduled workout routines.
Each social platform has campaign options based on buyer journey and attribution opportunities:
- Website conversions
- Website conversions
Now that there is an outline to social attribution, let’s look at the path in the journey and how social interacts with other channels and the bottom line. How is accurate reporting achieved and how is that reporting conveyed to show value in all campaigns?
First, educate your client or team on indirect sales and integrated marketing efforts driving towards the bottom line to help get this process started.
Second, use platform specific reporting to show each campaign types contribution and attribution that the social advertising campaign drove towards a conversion.
- This can be from a lead sign up, share, link clicks, etc.
Third, use Analytics either through Google or another analytics platform to assess both direct and indirect sales attribution from paid social using two multi-channel funnels report types:
- Assisted Conversions outlines all channels and their respective influence on both direct and assisted conversions to the website. By singling out our social campaign, in this example, we can see that the assisted conversion dominates the Facebook campaign influence. This shows that social initiates the conversion, but does not correlate to the direct sale.
- Top Conversion Paths portrays, based on that assisted to direct conversion value, where in the journey different campaigns touched a user who did ultimately make a purchase. As we can see, paid, shows that users touched a paid campaign early in their respective journeys, but also may have had multiple touch points.
As seen in the examples, having multiple sources is an integral part of any paid advertising mix. It is crucial to the long-term success of other campaigns and the final website sales.
We’ve looked at a specific example for the interaction path based on beauty and fitness and seen that the journey starts with social and has a long-term impact on the final conversion. We have also seen that social advertising can show up in multiple spots within the conversion path, based on analytics, but typically assists in success versus directly driving it.
By using this information, think about how your business can achieve even greater results by looking at the larger, integrated advertising picture that will improve results and the bottom line by using social attribution to find success.
Now that you have Google and social attribution under your belt, it’s important to know how attribution is driving the future of PPC.
Data Attribution and the Future of PPC
While reading a blog post, a fellow reader’s comment on the blog caught some attention. In so many words, the comment said something to the effect of “I feel bad for PPC. It will soon be irrelevant as no one clicks on ads anymore. Ads are annoying.”
Wait. What? No one clicks on ads? Are you kidding me? Everyone clicks on ads and there is data to prove it! Here’s the thing: it’s getting a bit crowded in the marketplace. Facebook ran out of Newsfeed ad space. Google removed sidebar ads. Competition is stiff. And it isn’t because ads are becoming obsolete. Quite the opposite.
It is because of the increase in competition and decrease in traditional ad space that we as digital marketers are called to be wiser in who and how we target as well as in how we allocate our budget. Enter the new era of attribution.
Google’s Data Driven Attribution Model
Google AdWords’ Data Driven Attribution Model has been available for qualifying accounts since May 2016. If you have spent any amount of time talking with your Google rep over the last 18 months I bet attribution has come up at least once. Possibly 50 times. We get it, Google. Last-click is so 2007. If your client has always used last click, and you have always used last click, then switching starts to feel a little scary.
Well, time to get over it. The path to purchase is no longer a straight line, (it hasn’t been for some time) and if we have machine learning available to give us insight into that path, we are fools to ignore it.
Making the Switch
The first step to switching to data driven attribution is to just look at it. In the below example, you can see things don’t shift drastically.
In this case, I’d say it is a no-brainer to go ahead and cut the cord with last click. Just remember: you won’t get historical data driven attribution numbers. Keep that in mind as you are reporting and looking at MoM and YoY data. But eventually, you can use Excel Solver to maximize your budgets and properly allocate money to contributing keywords. Voila! We are done, right?
Everyone Who Has Mastered Data Driven Attribution in AdWords Take One Step Forward…
Not so fast, [your name here].
What about display? What about other channels? When we are taught to diversify our ad budgets across multiple channels because the consumer’s path is less a staircase and more a bowl of spaghetti, what good is a search/shopping campaign-only attribution tool in AdWords? I want the whole picture. I want to see search, display, social, programmatic, direct mail. I want to consider the lifetime value of a customer. I want to trust that when I look at data in one platform, I am not accidentally counting it twice when I look in Analytics. I want it all. I want the truth!
Yeah, yeah, yeah. I know, Jack. I can’t handle it.
I know what you are thinking. “Google has Analytics 360 and Attribution 360. Just use those tools.” Good point. There is much buzz over Google Attribution. So when I am ready to personally invest my life savings into having access to those programs, I’ll let you know. In the meantime, what are digital advertisers to do? Do we ignore DDA if we don’t have access to more robust (and expensive) analysis platforms? Do we try to sneak a much higher tech fee into our client’s billings to get access?
No on both accounts.
The other day at Hanapin, we were having a group discussion surrounding DDA in AdWords. And the brilliant Kelly Pollock came to this conclusion: This is just a baby step. We can’t ignore DDA. And we won’t stop at DDA in AdWords. This is a stop on the way to a full picture of the customer journey and optimizing campaigns to give due credit. So for the time being, go confidently in the direction of your data.
Now, Back to the Future
As I continue to mull over the future of PPC I know these three current truths:
- PPC is here to stay
- No attribution model is 100% accurate
- If you want to be an industry leader, you need to be willing to take risks and ask tough questions
Even with the possibility of having perfect vision with cross-channel attribution through Google platforms, there are still other things to consider: if the client uses a CRM, how do we mash together that data with DDA? And if a conversion happens and we don’t see the touchpoint, did it really happen?
My point is we can take this step forward with DDA in AdWords, marry it with multi-channel funnel data in Analytics, and start to provide our accounts with more than just “let’s bid up on keywords that have low CPA” (by the way, if your agency is providing you with that as a strategy and nothing more, you should really push them for more). Go ahead and take the baby step of using attribution models in AdWords to your advantage. See what insights the right model can provide and know that there is more on the horizon.
So, I’m sorry, blog post commenter. PPC is here to stay. And the next time a well-placed, well-targeted ad directs you to exactly what you need (a Nicholas Cage mood board tee shirt maybe? I’m not speaking from experience or anything), thank your lucky stars that it is.
Now you can see how vital attribution is to your success. Take these strategies and run, because without fully understanding your customer journey, you’ll find yourself stuck behind the industry as it moves full-speed ahead. Whether it be in Google Attribution or on the social path, we hope these tips help you take hold of attribution and use it to your brand’s success.