The black box of Google Ads (+15% revenue)

Yes, ad costs have risen. Yes, running ads is still profitable, with Google Ads giving you $2 for every dollar you spend on average (Google's Economic Impact report data).

Still, if you want to nail paid ads and optimize them to drive success, you need a ruthless prioritization framework, a suite of execution recipes to follow, and, of course, a deep understanding of how each advertising channel works.

Today, Eva Khayrullina shares how they leveraged some of the Google Ads features to achieve a 15% revenue uplift.

We’ll also tell you a story about how an Optimization Manager and a marketing institute partnered up to help Cisco experiment more.

TODAY'S HOST

Eva Khayrullina, Marketing Manager @ SegmentStream

Performance Max is likely one of the most controversial features in Google Ads.

It allows you to advertise across Search, Display, YouTube, Discover, Gmail, and Maps from just a single campaign.

To do so, PMax uses machine learning models to choose ad placements and pick creatives from your provided assets.

Given this heavy reliance on machine learning, juggling creatives and placements across the whole Google fleet, it might seem like the ‘black box’ of analytics just got darker for PPC.

At SegmentStream we decided to put this to the test with KitchenAid, and found out that letting machine learning handle your targeting and delivery can lead to better results.

For KitchenAid, switching to ML and Conversion Modeling for their campaigns brought a 15% revenue uplift, and solved two crucial challenges.

The first one was analyzing the black box that Performance Max campaigns were, given that:

For upper-funnel channels (think YouTube, Discover, or Google Maps), sessions often didn’t end with a conversion and instead just increased the user’s awareness and intent to convert in the future.
Most people don’t buy their products (high-end kitchen appliances) during their first site visits, so the sales cycle can be quite long and exceed the 30-day attribution window of Google Ads.
Cookie restrictions make it difficult to track the whole user journey when they change devices and often spend weeks researching products.

The second challenge was optimizing those campaigns effectively, since PMax needs lots of input and signals. Conversions at KitchenAid are fewer, but more expensive than your average ecom.

We solved both challenges with Conversion Modeling, an approach that applies behavioral and contextual data to fill the measurement gaps left by tracking limitations.

Meta and Google apply Conversion Modeling under the hood to bridge the gap between ad interactions and conversions. For Google, tho, you may need to enable consent mode and enhanced conversions depending on which restrictions are in place.

KitchenAid, however, chose SegmentStream as its Conversion Modeling platform to track sessions, evaluate the user’s probability of converting in the future, and create a modeled conversion once that probability is sufficiently high.

Those modeled conversions can then be used to understand how ad campaigns influence total revenue, even when they didn't result in a directly observed (‘real’) conversion.

The analytics team at KitchenAid then compared data (ROAS, conversions, or CPC) between observed conversions vs. modeled conversions. And the modeled ones were also sent to Google Ads, amplifying the input that feeds the machine learning models and optimizes campaigns.

With the KitchenAid data, we ran a geo experiment and split the largest Performance Max campaign (Germany) into two further campaigns.

The test branch was optimized toward modeled conversions, while the control one was optimized against the observed ones (that is, purchase events).

Here are the results: the campaign that was optimized toward modeled conversion had 7% more purchases, 15% revenue growth, a 15% improvement in ROAS, and 40% more clicks at a 27% cheaper CPC.

Modeling conversions allows you to bridge the measurement gap when your customer’s path to interaction is becoming more and more difficult to track.

Knowing how to solve those mysteries in the customer journey is key to adapting to these changes that make it inviable to rely on just cookies and IDs.

Connect with Eva on LinkedIn.

Get a career leap (no strings attached)

Great news: you can now buy any CXL course individually.

From digital psychology and neuromarketing, through brand and planning, to the highly technical detail of analytics, CRO and PPC.

You can buy any course today and enjoy lifetime access to your training, without any subscription or ongoing commitment.

Check out our most popular courses now:

All our courses include exclusive learning resources, playbooks, and access to our private community of marketers.

It's a great way to get a highly-sought after qualification and increase what you can deliver in your role... while working at your own pace.

Google Tag Manager: Building a control center for your data in 2023

In 2023, Google Tag Manager (GTM) is probably the single most useful tool that your team can use to collect data that informs smart marketing decisions.

Understanding its benefits and how to properly utilize it can be challenging, though. We've put together this guide as a reference point for everything you should know to get started with GTM.

We bet the only piece of content more valuable than this post is our own Google Tag Manager course.

Read the entire article.

Customer Stories: Fostering experimentation in a $51 billion company

How do you even begin building an experimentation culture at a $51 billion company with over 80,000 employees?

Let's talk about how an Optimization Manager at Cisco and CXL partnered to upskill a CRO team and start optimizing on the shoulders of a tech giant.

Read the Cisco story here.

Unpopular opinion: What the job is really about

Remember when people called marketing the colouring in department?

We didn’t like that. So we learned analytics, data science, coding, experimentation, CRO, PPC, CAC, Payback time calculations...

We were masters of data and analytics. But if one thing is going to save your company and your career now, it’s creativity.

Sure, AI can build a graphic for you. Even a video ad. But creativity isn’t a format or file.

Creativity is saying something true in such a powerful way that people stop, hear it, and believe it.

In a world drowning in content, standing out means saying less with more truth, achieving emotional connection and a greater sense of purpose than just shipping stuff.

We might have lost that power in our teams. We’ve hired for technical skills too much.

If you want to use creativity to get results, stop trying to talk about your product in your content. Talk about your customers deepest fears and hopes instead.

Call out the bullshit. Have a point of view and take risks to surprise people.

You can get an AI to write your posts, create your ad set for Meta, and run CRO tests on your AI-produced formats. You can do more, faster, for less.

But the job isn’t producing content.

If you don’t understand that creativity is about saying and showing the world to people in a way that changes how they see it, you won’t stand out, relate, or change what customers think, believe and do.

That’s the job.

Connect with Katherine on LinkedIn.