le’s new attribution model: Three solutions for advertisers

In September, you’ve likely heard that Google Ads 4 and Google Analytics 4 are retiring the first-click, linear and time-decay attribution models.

The last-click and data driven attribution models, as well as external attribution, will continue to be available.

Some PPC marketers are unaware that Google will not just stop using these attribution models in terms of bidding. These models will be removed from reporting and comparison tools.

You can no longer use attribution models to analyze customer journeys in Google Ads or Google Analytics. You need to find alternatives.

A look at attribution models

Attribution models are used to link a conversion, such as a sale or lead, with an impression or click on an advertisement. It is a way of determining which ads, networks or audiences perform the best.

In the past, we have used different attribution models and rules to make this connection.

Here’s an example of each model using a football analogy:

• Last Click: All credit goes to the goal scorer.
• First Click: All credit goes to the first player who touches the ball in the course of the goal-scoring action.
• Linear : All players involved in the play that led to the goal should be given equal credit.
• Time decomposition: Last players to touch the ball in the action that led to a goal should be given more credit.
• Position based: Each player gets 40% credit for the goal scorer, and the first person to touch the ball in the event that led to the goal. The remaining 20% will be distributed evenly among the other players.

Google’s preferred model of attribution is a problem

Google Ads will now default to data driven attribution.

Google does not share the rules for deciding which ads are linked to conversions. I assume DDA uses a combination of these attribution models.

DDA can be tailored to you.

• Each data-driven model is unique to each advertiser,” a href=”https://support.google.com/google-ads/answer/6394265?sjid=17232203734043974084″ rel=”noreferrer noopener” target=”_blank”>according to Google. According to Google , each data-driven model has its own specific advertiser.

This is theoretically perfect.

An attribution model custom-made just for you. You didn’t have to think about the rules at all!

It sounds too good for it to be true.

DDA is customized to your account. What criteria are you using? We don’t really know.

It doesn’t really matter as long it works.

We could compare it with other models to make sure.

What happens when Google discontinues the “old” attribution model from its reporting section?

Is fewer attribution model a sign of poor performance?

What’s really the question?

As much as we may hate losing control as the years pass, it shouldn’t matter as long as performance continues to increase.

As we have seen, bid management has a minor impact on conversion rates (3%).

The real problem lies at a strategic level.

As Google states:

On the way to conversion, your customers may see multiple ads by the same advertiser. Attribution models will help you understand how well your ads are performing and optimize them across different conversion journeys.

How can we optimize conversions across journeys when we don’t have visibility? Let’s go through an example:

Analyzing the customer journey in action

To illustrate my point, I will use an example from one of our clients who has a simple media mix.

The client uses different tactics, just like in football: midfielders, defenders and strikers. To score a single goal, the whole team is needed.

Tactic	Last click purchases	First Click Purchases	Difference
Search Engines	2,478	1,579	57%
Email	1,978	1,184	67%
Search engine paid	1,621	2,796	-42%

When using the first-click attribution model, paid search does “score” well. When using the last click attribution model, paid search does not perform as well. When using this attribution model, organic search and email marketing are the winners.

It is not surprising that this is the case.

• Non-branded paid searches are the first step in the conversion process. Leads are generated.
• Lead nurturing is essential to mature prospects. This is done primarily through email marketing.
• Organic and paid brand search can lead to qualified prospects buying through the organic and paid search.

In football terms, this is:

• Search engine paid for non-branded brands = Defenders
• Email = Midfielders
• Organic and Paid Branded Search = Strikers

Does DDA suffice?

Would you understand this conversion funnel without these attribution models?

Probably. This example is very simple.

What if you start working on B2B projects where sales can take months, or B2C projects where repeat purchases are critical?

That’s a different story. I’ve seen many examples of DDAs that didn’t perform well.

Validating DDA conclusions using old, rigid attribution models is still valuable to me. You expose yourself to harm if you don’t use benchmarks.

Machine learning can only be as smart as the data you feed it.

Advertisers looking to adapt to changes will find three solutions here.

Solution 1: Next-level tagging plan

The first step in identifying interactions with customers is to develop a solid data-driven program.

You can confidently use DDA and last click models with complete tracking… but you’ll need to substitute first click for all the customer journey steps.

It’s not perfect, but it is the first step. In my example, we would attribute the last click to non-branded searches and the last click to branded searches. It’s not ideal, but works.

This requires that you track the entire journey of your customers. You can’t just rely on a simple tagging system. You need to micro-convert.

Solution 2: Integration of CRM data

Do you only track sales when tracking conversions?

You need to feed back the entire customer journey, yes even post-sale (through external attribution) into your ad platform.

The tool can be used to increase visibility, similar to lead scoring but this time with client scoring.

You can influence your bids in a different way than the “data-driven model” if you notice performance discrepancies.

The CRM is a tool that advertisers must use to understand the customer journey and determine the right media mix.

Solution 3: Alternative attribution methods

This project is more complex than others.

In essence, incrementality is exposing a certain audience to an ad, then hiding it from another similar audience. Then, you can compare the performance of both audiences.

This method, as you can imagine is cool but also prone to error. This method is only possible if your budget allows for it.

Customer surveys are a great way to find out what your customers want.

You can, for example, use an exit-intent Popup (asking visitors leaving where they came, what they did not like, etc.). You can capture more information by adding additional fields to your lead/purchase journey.

Be careful when using such data, as they can be skewed.

No perfect attribution model exists

In this article I have been searching for the best way to measure performance.

Don’t fall down the rabbit hole. There is no perfect attribution.

You want a strategy that is reliable and orientated.

It’s not for me to make business decisions. I am a ad-geek. Prioritize accordingly.

The post Google attribution model shakeup: 3 solutions advertisers first appeared on Search Engine land.