Past, present and future of marketing measurement

Why I Quit Marketing Metrics
Couple of years ago when I worked as a CMO I decided to quit marketing metrics - why? It's too damned hard. Spoiler, things have changed since then for a good reason.
I love marketing since it can be one of the most rewarding disciplines. You onboard a client, do your magic, and suddenly, it is raining conversions.
The other day, you find yourself checking campaign metrics at 11 pm and nothing moves the needle in the right direction.
Tired of reading all day, we got you!
All the while, your boss and your client say it should be sooo easy to optimize marketing spend and campaigns. "You have all that data. That really makes it easy. A monkey could do this. Back in my days..."
Why is that? Why are some campaigns delivering on autopilot, and some know only one direction (and it's the wrong one), no matter what you do?
It must be that damn customer!
Probably right. And you should know. You have all the data, right? Right?
If it only was so easy. There is an abundance of marketing data, correct? But one could say it's an overabundance.
One could say, there is too much data!
You have Google Ads, Meta, TikTok, LinkedIn, Snapchat, and God have mercy on your soul, if you do TV or print.
The reality is that access to large volumes of data means nothing in the first place. If you are not in the lucky position to have a DMP at your hands or dedicated data engineers you are on your own. And it is a lonely place.
A major problem is too much siloed data
Most platforms on which we run our ads do not play well together. No pineapple.
Smart people call this siloed data. Like a farmer, you have different silos, one for corn, one for rice, one for soybeans. In your case, you have a silo for Google Ads, one for Meta, and a third, dirty, messy one for offline. And they are all rusty, and leaking, and probably you have some pests somewhere in there.
Silos are dangerous
Every farmer will tell you, Silos are dangerous. You can drown in them. Look it up if you do not believe us.
The data silo metaphor is so good because that is what will happen to you if you dabble in data silos without an exit route.
Even if you have a DMP and a data engineer, stitching all that data together is a task not for the faint of heart. Seriously, if you have data engineers, be nice to them.
Lying with data - data lying
If you choose to build your silo on one of the advertisers' platforms, you are in danger.
The dangers of silos are not limited to getting all the silos together and working with the combined contents. Even though rice might pair well with soybeans, the grains just do not fit, no matter what you do.
The biggest danger of silos is if you do not own them, you are not in control.
The advertising platform controls it. And they just charged you a couple of thousand bucks; what do you expect them to say? Sorry, it was all for fun, no conversion, better luck next time?
Obviously not. To say they outright lie in your face might be a little too much, but they lead you to believe their data.
Only if you drill down in documentation and definitions you come to some enlightening sentences.
What all these platforms do is they use attribution to let them look good. The most common way is to take a ridiculous look-back window.
Never look back
What does that mean? Google, Meta, and Co. measure the interactions with your ad. From now on, if that poor soul decides to buy from you in a given period, let's say 30 days (because that's what most use per default), the conversion will be attributed to that single click.
In plain English, this means, that one click is solely responsible for the conversion. No matter what.
But is it? Neither you nor Meta or Google know. Probably not even the customer if you were to ask them. But damn, it looks good in that spreadsheet.
The Original Sin
This fraudulent, erroneous measurement of conversions is the original sin in marketing analytics. We know that a user clicked something at the beginning, and a conversion happened at the end (hopefully). In between?
The messy middle.
That is a term even Google uses in its research (https://www.thinkwithgoogle.com/intl/en-emea/consumer-insights/consumer-journey/navigating-purchase-behavior-and-decision-making/) about attribution. It is a catchy way to say we have no clue what's going on and hope for the best.
Of course, they put it in a lot of fancy words, cognitive biases, and behavioral science principles. C'mon, they're Google. What do you expect?
But where does this leave the poor marketer at the front?
Easy, here is a shiny newish tool: Google Analytics 4 attribution.
Real talk: Everyone hates GA 4. It is en vogue to bash it. Might be based on the hasty launch, unclear communication, and missing core functions. Who knows?
But we had time enough to cope. GA4 launched in 2020, and was mandatory in 2023. We're in 2025. Smooth sailing.
Yeah, not so much. There are events that can mess up your analytics data. And you sure have everything tagged to the max, no social interaction gets recorded as direct, correct?
User ID on fleek. And the Google Ads conversions match up with the ones in GA4.
Ok, that was too much, sorry.
Your GA4 attribution is prone to fail
You have incorrect or missing data, you might not have enough data. You might have the wrong data. Facebook might have changed the referral again (was there a letter left?). So, perhaps you can use attribution. Hey, at least GA4 is not saying you shouldn't.
But who am I, there are multi-touch attributions!
A customer drove by my billboard in Boise, Idaho, and he later clicked on my AdWords ad while in Charlotte.
I know this is true. I have user ID features!
After he saw my ad in a smartphone game he converted.
So I put 20% on the first interaction and ten on the second...
Oh, look! I can choose from models. Time decay sounds grim, better go with bathtub. Feels comfy.
So perhaps, now I am less wrong.
Or am I wronger?
Perhaps a dump truck was standing in front of my billboard and the user didn't see it.
Or they just clicked on my intrusive full-screen banner ad by accident while playing a game?
When in doubt, buy a new tool!
If your standard GA4 is not enough for you, take a look at GA 360, and if that does not work, try Big Query. All for the higher good.
That was a lot of Google bashing. It's not that they are alone with it, but most people start their marketing analytics journey with Google tools because they are free.
As we said above, that does not mean that paid tools are better. The main problem stays with you, the messy middle. We just have no damn clue what and why a user acts like he does.
But what if we just take the two things we are certain about?
No, not death and taxes.
Ok, ok, the four things we are certain about:
- Taxes
- Death
- The first interaction
- The conversion
We cannot do much about death and taxes, but we sure as hell can measure the first interaction and we have a way to know if something happened in the end. I see an order in my store, or I have a new email in my inbox.
That seems to be a good starting point.
Perhaps it is too early to ditch marketing measurement. Let's untangle it and try something new.
What direction can marketing measurement go? We‘ll find out in the next videos on marketing measurement and optimization, so stay tuned!