Google Ads Attribution: Find your North Star and Place the Right Bid

Google ads Attribution Models and KPIs for optimization keep getting more complex. What is the right bid for your business? How do you set up the right budget for your Google ads campaigns? Where and when do you move this budget?

Ideally, we want to spend money on where our sales are coming from. But where our sales are coming from depends on the attribution model.

When it comes to choosing an attribution model, it feels like you are picking an entire ‘belief system.’

In this article, you’ll learn how to reverse engineer what the right bidding systems are for your campaigns based on the status of your operation and how this can determine which  Google ads attribution model you should use for your bids and budgets.

With more and more attribution models becoming available, and customer journeys not getting any simpler, the seemingly simple questions we ask ourselves as marketers on a day to day basis are becoming less and less clear.

And to be fair, these questions can be daunting at times.

Much like picking the right type of pasta for your sauce, it takes a lot of thinking, pondering, and holding internal philosophical discussions to identify which attribution model and KPIs are best to optimize your operational maximize your Google Ads efforts.

Let’s try to make it easy.

Picking the right North Star

The first element to take into consideration is picking the right North Star (which does not refer to Kanye West’s daughter, but to the overall goal of a marketing operation.)

So “Picking the right North Star” means defining your goal and ambition, and the metric that you want to use to optimize your business. There are a few options.

#1) CPA: Cost Per Acquisition

The most common “North Star” is Cost per Acquisition (CPA or CAC).

It is easy to implement because it is simply the cost sustained to acquire a new customer. Most businesses that are focused on lead generation use this metric – imagine if you are generating leads for an insurance company or a credit card application.

It is, however, a very crude metric that ignores the quality of the acquisition.

When analyzing your business in terms of the cost per each lead, you are taking on a big challenge: how to assess the quality of each lead. 

Let’s say you measure the cost for each credit card application. You don’t know how many of these applications will be successful or not.

And here is where the more sophisticated Cost per Marketing Qualified Lead (CPMQL) comes into play.

#2) CPMQL: Cost per Marketing Qualified Lead

What you’ll consider in this case is the cost of a lead that has been deemed to be good enough. In the example above, when someone applies for a credit card, they go through a process where the sales team assesses whether the lead is good enough and then injects this information back into the CRM.

This process analyzes whether your contact is a real person and if they are on target with your business.

In the credit card example, you may define location and credit card score, so a company would need to be located in the UK and have a designated credit card score to qualify.  With this approach, the granularity level is fine, but you rely on the assumption that qualified leads are more likely to convert into paying customers, which is not necessarily true.

If you want to go deeper into KPI granularity, you can use the Cost Per Paying Customer (CPPA), sometimes referred to as CPAC – Cost Per Acquired Customer.

#3) CPPA: Cost Per Paying Customer

With this approach, you are analyzing the actual value of a paying customer, but there are some setbacks too. Let’s go back to our example.

First, you issue the credit card to a much lower number of people than the number of qualified leads. Then, there may be a delay between when the lead arrives, gets qualified, and becomes a customer. Not to mention the challenges of getting all information in one place.

All of the above makes it difficult to make the right decision when it comes to bidding correctly and allocating budget for your campaigns.

There are going to be tradeoffs, as with pretty much everything in life. You know the rule: if it’s tasty, it either makes you fat, or it’s illegal! The more granular is the metric, the lower is the data. In other words, you have two options. A rough metric that will give you enough information or a sophisticated one with a smaller volume of data, which will make it more challenging to come to a decision that will help you optimize your campaigns.

For a credit card company, CPMQL might be the right balance between volume (higher than CPPA) and quality (deeper than CPA).

I know you have been waiting for it, so here it comes! The next one is ROAS (Return On Advertising Spend).

#4) ROAS: Return on Advertising Spend

This approach will give you a highly accurate level of optimization.

Most e-commerce businesses optimize for ROAS. However, there are a couple of problems with this approach.

First, revenue can be scattered across the different elements of a marketing initiative, such as multiple keywords. Secondly, outliers are always around the corner, and that could lead to bias when analyzing your campaigns.

You may also find that cheaper items sell better online, making ROAS look better but potentially reducing your bottom line.

For example, not everybody takes delivery costs or low margins into account. If you are not an eCommerce business, you also need to consider offline conversions.

So despite ROAS being the most used KPI by e-commerce as it provides a high level of optimization, it is not immune from detracting factors.

So far, we have covered the most common KPIs. Let’s take a step forward and have a look at LTV (Lifetime Value) and particularly LTV/CAC ratio (Lifetime Value/Customer Acquisition Cost.)

In this case, you’ll optimize using a smarter version of ROAS. That’s because while ROAS only captures revenue on advertising spend, LTV can capture the likelihood that a specific customer will come back and buy again. Further, LTV doesn’t simply look at revenue but also considers your margins on the first transaction. These considerations make LTV a more accurate level of optimization.

You’ll have the same cons as ROAS and yet another one. The likelihood of a customer coming back is based on an average number. So, it will not always be clear and might take a long time even to get a rough outline.

You picked the right North Star, now it’s time to discuss attribution models.

Google Ads Attribution Models

Attribution models are so important in marketing today that you could compare them with the experience of going out for drinks.

Let’s say you start with an Appletini, then a regular Martini (stirred, not shaken), then a glass of red wine… and can you say no to “just” a couple of pints of beer? And to end the night with class, you add one flute of prosecco (or two). At this point, you’ll probably feel “officially”  tipsy… But which was the drink that made you drunk?

If we were in a world using the last-click attribution model, the prosecco would be entirely responsible for your hangover….but that wouldn’t be a fair representation of what happened.

The attribution model conversation is about how you attribute the value of a conversion to the right element in the user journey that entices the user into buying a product or signing up to a newsletter, or whatever your business goal is.

You could go with multi-event attribution. This approach aims to distribute the credit of a specific conversion to all the advertising touchpoints that were influenced by that conversion.

The position-based model is within the so-called heuristic models (off-the-shelf models offered by Google).
In most cases, this is the lesser of several evils as you’d assign different values across positions in the chain, regardless of the actual impact on the completion of the sale.

With this model, you are simply assigning the attribution based on the position in the chain using a top-down approach. This is a very simplistic way of seeing the complexity of reality.

Let’s go beyond this.

The next step is algorithmic attribution, which is a complete analysis of the data you have so you can determine the relative impact of a given touchpoint on conversions.

Rather than “shortcutting” and applying a blanket statement with a position rule, the algorithmic attribution involves having a custom model and weightings for each touchpoint based on every single specific user dynamic.

There are several algorithmic models. One of the most accurate and reliable models is based on Markov Chains.

As well as the position of the touchpoint, you could think about the type of touchpoint, especially if you are advertising outside of the Google Search network.

Should you consider ads that drive conversions from clicks equal to ads that drive the same value of conversions from impressions? It will depend on your business model and how and where you are reaching your customers.

Typically, for Search campaigns, you would focus on click attribution. When it comes to Display, video, or social campaigns, you would look at view attribution – but take it with a pinch of salt.

You could assume that a click is more likely to influence the behavior of a customer than a view. However, you could prove this assumption using an “uplift” test where one group of people sees your ads, and another group does not.

To pick the right attribution model, think about the level of sophistication of your business.

Despite being largely used, last-click attribution is possibly one of the worst solutions you could adopt. You should consider the position-based approach.

Don’t trust me; test it. Go to Google Analytics at least once per quarter and use the model comparison tool to check last-click against position-based, linear attribution, and first-click. You will see some interesting differences.

Even if the position-based approach is preferred to last-click, it is a very simplified view of life.

If you run a more sophisticated business, you should embrace non-heuristic models and move towards algorithmic models, and Markov is one of the best options out there.

Just like when you are wondering which drink caused your hangover, the right attribution model will give each drink the appropriate credit. It will help you decide whether next time you should start with red wine and finish with beers, or whether it’s best just to skip the Prosecco.

Pick the Right Google Ads Bidding Model

Let’s recap what we’ve been doing so far.

The first step was picking the right North Star (in short: pick the goal you want to optimize for); then, we went ahead and analyzed the attribution models (in short: pick the system you want to use to attribute value across your marketing operations).

Now we are ready to define the right bidding model.

When it comes to bidding models, the best analogy you can think of is that of driving a car.
You have three options: you either use a manual gear (stick-drive), an automated gear, or you have a self-driving car (also called autopilot.) You can use a similar approach to bidding.
The manual gear is a very resource-intensive way of driving. You are the one who needs to change the gear every single time, and it is comfortable for everyone only if you’re a good driver.

This is very similar to manual bidding: you are bidding by hand, keyword by keyword; you are bidding manually on keyword view operations, or the placement, or the ad assets on Facebook.

The second approach is automated bidding: you can adjust a set of rules that define the bidding for you.
In this case, you are still in charge of the bidding because you are defining the rules which your bidding system should apply to each keyword, and you are not relying on Google.

The third approach is going entirely on autopilot: use the Google Automated Solution.

Manual, Automated, or Google Ads Smart Bidding?

So there are three ways you could bid: Manual non-Google Bidding, Automated non-Google Bidding, Google Smart Bidding – and even then, it depends on the KPIs that you are using.

There are pros and cons to each one of these bidding models.

• First, let’s see Manual non-Google Bidding.

The benefit of this approach is that you can customize it to consider disruptive events (e.g., seasonality, special discounts, weather conditions, and more.)
However, it is not sustainable for large accounts, as it is very time-consuming.
Also, it ignores Google audience-related information (i.e., Google knows more about what’s behind an auction, and has access to signals that are not available otherwise, such as user location and operating system.)

• Alternatively, you can use Automated non-Google Bidding.

With this approach, you are the one deciding the bid by using a set of rules. These rules are in addition to the manual bidding model and are also customizable in case of disruptive events.
This model can also be used for larger accounts as it is implemented programmatically through APIs or Google scripts.
However, it still lacks Google audience-related information.

• The third option you have is Google Smart Bidding.

It does require a limited amount of data to be activated, and it knows the auction from the inside out. There’s a caveat tho.
Usually, Google Smart Bidding takes time to be trained, it relies on historical series and tends to ignore disruptive events. Furthermore, this model may have problems when dealing with a crisis, such as the COVID-19 pandemic.
Very often, it is a full-on black box, and it is challenging to read if it will bring positive or negative results.

When it comes to choosing your bidding model, it is vital to understand the size of your operations.

If you are a large business, there is no way you can bid manually. At the very least, you should use automated non-Google bidding.

Depending on whether it is crucial for you to be in the driver’s seat and it is not possible to sustain a black-box type of approach, you should know if you need to steer away from Google Smart Bidding.

Wrapping It Up

If you are not sure what’s the right bid for a keyword or if you’re spending the right amount on your Google ad, you don’t need to guess. You can follow the North Star process to make sure that your small, day to day (often very granular) decisions are linked to your high level, long term business goals (i.e., making money).

First, make sure you set the right KPI for your goal. Either choose ROAS, CPA, or whatever. Just make sure your marketing North Star is navigating towards what you care about.

Next, you need to find a usable attribution model. It won’t be perfect, but it won’t be worse than last-click. It will tell you which ads are helping you get to your business goals.

Finally, you can make sure that your bidding is helping you hit your KPIs based on the best possible attribution model.

Remember to review this over time. As you get more data, you learn more about your customers and can better predict their value. You can begin to slowly start moving further down the funnel and building a more sophisticated growth engine.

But perhaps most importantly, and never forget this, cheese never, ever, ever, goes with fish.

Gianluca Binelli is the founder of Booster Box, a performance agency specialized in scientific marketing. In 2019, he was voted 2nd Most Influential PPC Experts by PPC Hero. He worked 6 years at Google as a Product Specialist and online marketing manager for Google’s products in EMEA. He also served as Capital G’s Advisor, helping startups with their online marketing in Google’s own late stage investment fund. He is a regular speaker at international PPC conferences and teaches digital marketing at the University of Pisa (Italy). You can get in touch with Gianluca on Linkedin, Twitter, or visit Booster Box website.