You’ve probably heard the news: Google Ads and Google Analytics 4 will completely retire first-click, linear, time decay, and position-based attribution models in September.

Last-click and data-driven attribution models will remain available, along with external attribution.

What some PPC marketers don’t realize is that Google won’t just discontinue these attribution models from a bidding standpoint. They will also be removed from the reporting and comparison features.

This means you can no longer analyze your customer journeys within Google Ads and Google Analytics using attribution models. You need alternatives.

A look at attribution models
Attribution models help connect a conversion (i.e., a sale or a lead) to an ad click or impression. It’s a way to determine which ads, audiences or networks perform best. 

Historically, we’ve used several attribution models with different rules to make that connection.

Using football analogy, here’s what each model represents:

Last click: The goal scorer deserves all the credit.
First click: The first player who touched the ball during the action leading to a goal deserves all the credit.
Linear: All players who touched the ball during the action leading to a goal deserve an equal share of the credit.
Time decay: The last players who touched the ball during the action leading to a goal deserve more credit than the first players.
Position-based: The goal scorer and the first player who touched the ball during the action leading to a goal deserve 40% of the credit each. Other players will get the remaining 20% evenly.

The issue with Google’s preferred attribution model
This shift leaves data-driven attribution (DDA) as the default attribution model in Google Ads.

Google doesn’t share the rules that decide what ads to link to a conversion. I personally assume DDA utilizes a combination of the aforementioned attribution models.

There’s one very cool bit, though: DDA is tailored to your account.

“Data-driven attribution is different from the other attribution models because it uses your conversion data to calculate the actual contribution of each ad interaction across the conversion path. Each data-driven model is specific to each advertiser,” according to Google.
Theoretically, this is perfect. 

An attribution model custom-made just for you. And you didn’t even have to bother thinking about those rules! 

Yet, it sounds too good to be true. 

DDA is tailored to your account. But based on what criteria? We don’t know.

This shouldn’t matter as long as it works. 

And we could make sure it does by comparing it to other models.

But what happens now that Google will discontinue “old” attribution models from the reporting section?

Does fewer attribution models necessarily mean poorer performance?
Now that’s the real question. 

While we probably all hate to lose more control with every year that passes by, that shouldn’t be an issue as long as performance keeps on increasing. 

And as we saw earlier, the impact is minor in terms of bid management (3% of all conversions).

The real issue lies elsewhere – at the strategic level.

As Google states:

"On the path to conversion, customers may interact with multiple ads from the same advertiser… Attribution models can give you a better understanding of how your ads perform and can help you optimize across conversion journeys.”
So how do we optimize across conversion journeys if we lack visibility? Let’s walk through an example first:

Analyzing customer journeys in action
One of our clients has a relatively simple media mix, so I’ll use that as an example to illustrate my point. 

Like in football, that client has different tactics: defenders, midfielders and strikers. It takes that whole team to score a goal.

Tactic
Last-click purchases
First-click purchases
Difference
Organic search
2,478
1,579
57%
Email
1,978
1,184
67%
Paid search
1,621
2,796
-42%
Notice that paid search “scores” pretty well when using the first click attribution model. However, not so much when using last click. Organic search and email marketing steal the show when using that attribution model.

This is as expected, though, because:

The conversion journey starts with non-branded paid search. They generate leads.
Lead nurturing is necessary to mature prospects. That is mainly done through email marketing.
Qualified prospects eventually buy through organic and paid branded search.
Or, to put it in football terms:

Non-branded paid search = Defenders
Email = Midfielders
Organic and paid branded search = Strikers
Is DDA enough? 
Would you have understood this conversion funnel without those attribution models? 

Probably. This example is quite straightforward. 

But what if we start working on a B2B project where sales take months or a B2C project where repeat purchases are important?

Now that’s another story. I have seen plenty of examples where DDA did not perform well. 

I think validating DDA conclusions with old and rigid attribution models still has value. Without such benchmarks, you expose yourself to potential harm.

After all, machine learning is only as intelligent as the data we feed it.

Here are three solutions for advertisers looking to adapt to the changes.

Solution 1: Next-level tagging plan
Developing a solid data program is your first step to identifying customer journey interactions. 

Through complete tracking, you can use DDA or last click attribution models confidently… but with all those customer journey steps to replace first click and so on.

I know it's not ideal but this is a first step. If we use my example above, you’d attribute last click leads to non-branded search and last click sales to branded search. Not ideal, but it works.

Naturally, this requires tracking the entire customer journey. You can't rely on your old simplistic tagging plan. You need micro-conversions.

Solution 2: Integrating CRM data
When tracking conversions, do you stop at sales? 

Now you need to track and feed the entire customer journey (yes, including post-sale) back into ad platforms through external attribution. 

You can then use that tool for increased visibility – like lead scoring but with client scoring this time.

If you spot performance discrepancies, this should enable you to influence your bids differently from the "data-driven" model.

In short, the CRM must become (if it isn't already) a central tool for advertisers to better understand and inform the customer journey – and, consequently, the appropriate media mix.

Solution 3: Other attribution methods
I’m venturing into more sophisticated grounds here, which doesn’t apply to all projects. 

Basically, incrementality means exposing an audience to your ad and purposefully hiding that same ad from a similar audience, and then comparing both audiences’ performance levels.

As you can imagine, this method is very cool but prone to errors. (Not to mention only available if you have big budgets in the first place for data reliability purposes.)

Your next best bet is with customer surveys. 

For example, you can use an exit-intent popup (asking leaving visitors where they came from, what they didn’t like, etc.) or additional fields in your purchase/lead journey to capture additional information.

Naturally, be careful with such declarative data since they are often skewed to an extent.

There’s no perfect attribution model
Throughout this article, I’ve been chasing the perfect way to measure performance.

But don’t get lost in the rabbit hole. There is no such thing as perfect attribution. 

What you want is a reliable yet directional input to your strategy.

Getting past that stage is for ad geeks like me, but not useful for business decision-making. Prioritize accordingly.