Over the past 3 years, I’ve helped over 100 businesses achieve growth with digital marketing (Growth marketing, Paid search, CRO)

My goal is to support others to reach their marketing goals with the same dedication and passion that fuels my own journey.

What I offer:

→ Google Ads management for ecommerce brands and SAAS.
→ A proven framework to scale businesses with a systematic Google Ads approach.
→ Expertise in business economics to scale brands profitably, basically your CMO consultant.

My results:

→ Retaining clients organically
→ €1 million in monthly ad spend
→ PPC community of 20,000+ members
→ Scaling accounts with 70% revenue growth in 3 months

Here's how my team or I can help you:

If you want to learn more about how we can grow your business
👉 Go here: https://flowboost.com/growth-plan

If you want to get better at Google Ads specialist:
👉 Go here: https://flowboost.com/flowboost-library

Follow my journey:
👉 Go here: https://www.arjanschoorl.com/ or follow my on Instagram: @arjanschoorl

Arjan Schoorl

CMO at Stella

For the past 8 years, I've immersed myself in the world of digital advertising and performance media. I've had the privilege of managing $38 million in ad spend, which has translated into $100 million in revenue for the clients I've worked with. My expertise spans all performance media channels—whether it's Paid Search, Paid Social, Account-Based Marketing, or the intricacies of CRO and A/B testing, I've been there, driving results.

I've been fortunate to play a part in some significant growth stories, helping companies scale from $10k to $100k in monthly revenue or $1 million to $5 million in monthly revenue. I’ve worked across various platforms like Google Ads, Microsoft, LinkedIn, Meta, and have also dived into programmatic display, CTV, and audio advertising.

In 2022, my work on innovative campaigns for a leading online pharmaceutical company earned me a spot on Performance Marketing World’s 30 Under 30 list—a milestone that I’m incredibly proud of.

Now, I've taken on a new challenge as the founder of Stella, a marketing measurement startup. My focus is on all things advertising, and I’m committed to helping brands navigate the ever-changing landscape of performance media to achieve exceptional results.

Brenden Delarua

Never thought I’d get to collab with THE Google Ads. 

Very very awesome to see. Love how this post turned out.

Co-Founder @ Stella | Accessible Incrementality Testing & MMM for Mid Market DTC Brands

Are you settling for “good enough” data because full-stack measurement seems too hard? 

It's time for a reality check. We connected with three Google Ads experts to bridge the gap between your expectations and what it actually takes to get your data right.  

The reality check: 
• 𝗛𝗮𝗿𝗱 𝗱𝗼𝗲𝘀𝗻’𝘁 𝗺𝗲𝗮𝗻 𝗰𝗼𝗺𝗽𝗹𝗲𝘅.. For Julius Fedorovicius from Analytics Mania, the evolution of tagging has eliminated the wait for IT tickets. Today’s leaders are using simplified interfaces to turn their data from a project into a competitive edge tool. 

• 𝗬𝗼𝘂 𝗰𝗮𝗻’𝘁 𝗰𝗮𝗽𝘁𝘂𝗿𝗲 𝗱𝗲𝗺𝗮𝗻𝗱 𝗶𝗳 𝘆𝗼𝘂𝗿 𝘀𝗶𝗴𝗻𝗮𝗹 𝗶𝘀 𝗯𝗿𝗼𝗸𝗲𝗻. For Dana DiTomaso from Kick Point Inc., there’s danger in relying on last-click attribution for a “neat story.” Because real customer journeys are messy, assigning clean credit to a single touchpoint ignores what actually created your demand in the first place.  

•𝗠𝗮𝘁𝗵 𝗶𝘀 𝗰𝗼𝗼𝗹 – 𝗮𝗻𝗱 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰. For Brenden Delarua from Stella | Growth Intelligence, inherently trusting dashboards leads to rash business decisions. Instead, he suggests marketers move beyond what’s trackable to understand what’s truly causal. 

Swipe to see how to close the gap. 

Start building your data strength today → https://goo.gle/4aA8rLR

Measurement misconceptions

I once presented record-breaking ROAS to our CFO while the business was losing money.

I had no idea.

I trusted Google and Meta to tell me what was working. They told me everything was working. Turns out they were just taking credit for things that were already going to happen.

Branded search “works” because those people already wanted to buy. Retargeting “works” because you’re re-paying to claim a sale you already earned. High ROAS, zero incrementality.

It gets even muddier with Pmax and broad targeting on Meta. Because Retargeting and branded search still make it through even with exclusions sometimes.

I didn’t know the difference. And it cost us.

That sent me into an obsession. Every blog post. Every YouTube video. Every Ben Dutter thread I could find. Then actually running the models myself. Breaking them. Starting over.

Most marketers know something feels off with their attribution. They just don’t have a word for it yet.

Incrementality was mine.

If your ROAS looks great but the business doesn’t, you might need yours too.

So many companies will do anything to avoid paying more.

They'll pay for food trucks, team-building exercises, "culture initiatives" etc.

Anything except just giving people money.

At Stella, my team gets weekly cash bonuses for hitting their goals.

We also give them one optional stretch goal each week. Extra work that pushes the platform forward faster.

Do it, you get another bonus. 

Don't do it, nothing happens. No penalty. No guilt. It's optional. That's why it's called a STRETCH goal. 

The deal I have with my employees is stupidly simple: 

Extra work = extra money.

I love my team. But this is a business relationship. 

They're working for money. I'm paying for outcomes.

I've worked at companies that called employees "family" while barely paying them. 

Companies that threw pizza parties instead of raises. That gave Christmas bonuses some years but not others because "we didn't hit target". 

Those companies lose people.

The fastest way to make someone resent you is to pretend money matters less than "culture".

People aren't stupid and they have bills to pay. Cut the crap and you'll retain your top people.

A data scientist at a major agency got on a demo call with us last week.

Halfway through he goes: “Can you expose the ad stock and saturation priors? Can we configure the model code directly?”

We built Stella. We know exactly what he’s asking. And the answer is no.

Not because we can’t. Because that’s not the point.

Then the next call that same day, a marketer. Smart guy. Halfway through the demo he goes “I’m smart but honestly, you lost me a while back.”


This is our life right now.

Data scientists want to get under the hood and rebuild the engine. Marketers just want to get in the car and drive.

We kept trying to serve both. That was the mistake.

Stella is not for data scientists. If you’re a data scientist, go build your own model. Genuinely. You probably can.

Stella is for the marketer who got wrecked in a CFO meeting presenting record-breaking ROAS while the business was bleeding. Who trusted Google and Meta to tell them the truth. Who knows something feels off with their attribution but doesn’t have a PhD to prove it.

That’s who we built this for. That was me.

We’re still simplifying. Still listening. Still embarrassed by every confused look on a demo call.

Everytime a user gets confused using Stella, we take that as a signal to fix and simplify. 

We are a data science tool for marketers. Not for data scientists.

Holdout tests -> MMM -> Verify Accuracy -> TAKE ACTION -> Repeat. 

Incrementality and media mix models shouldn’t be temperature checks or one-and-dones.

It should be interwoven into your marketing processes.

If you want to cancel Stella, you cancel.

It’s crazy how hard SaaS companies make this.

You click “cancel” and you’re digging through settings, booking a call you don’t want, and explaining yourself to someone whose job is to talk you out of leaving.

At that point, I’m not “considering my options.”

I’m canceling forever.

So...we don’t do that.

Some people think it's risky.

Hard disagree.

If a product is useful, you don’t need retention tricks.

If it’s not useful, forcing customers to stay doesn’t fix the problem.

In our case, if you do cancel Stella, you’ll lose the historical data we’ve collected.

This is a simple business function and almost all software is like this. Data costs money to store so we don't store everyone's stuff for free forever on the off-chance they'll rejoin. 

That’s usually enough.

The difference is we’re straightforward about it.

No mandatory 12 month contracts or insane hoops to jump through if you want to cancel. 

Pretty crazy that this is not the norm.

We had 10 demo calls this week for a product you can use for free.

That’s when I knew we f*cked up.


Here’s what happened:

Stella helps marketers measure incrementality. We spent months building the easiest UI possible. Click a button, get your answer. Done.

But we also knew marketers would need to defend their results to data scientists. So we added every statistical validity metric under the sun.

MAPE. R-squared. Confidence bands. Statistical Significance. Post-treatment windows. The works.

Now our users are drowning.

A marketer logs in, sees “start your first study” and immediately books a demo call.

A data scientist logs in, doesn’t see their pet metric, and questions our entire methodology.


We built a product for everyone and somehow made it perfect for no one.


Yesterday I told Vinay we need to hire someone to handle all the customer questions.

He said: “Or we could just fix the product.”

Brutal. But right.


Every confused customer isn’t a support problem, it’s a design problem.


So here’s the new rule:

If a data scientist can’t poke a hole in it, AND a marketer can’t use it without a tutorial, we’re not shipping it.


Default view: dead simple. One number. High confidence or low confidence. Green or red.

Advanced view: every nerdy metric hidden behind a toggle for the data scientists who live for this stuff.


The best products don’t make you choose between powerful and simple.

They’re simple first. Powerful when you need it.


If you’re building something technical for non-technical users, stop trying to educate them with your UI.

Just give them the answer. Hide the math.

They’ll thank you by never booking a demo call again.


Want to try Stella? It’s free. No demo required. (That’s the whole point.)

CMO: We’ve spent $50k on YouTube ads at a 0.13x ROAS 

CFO: WHAT THE HOLY HELL


CMO: it’s okay, 35% of our post purchase survey responses claim to remember seeing a YouTube ad of ours. 

CFO: okay okay so YouTube caused 35% of sales? 


CMO: no not exactly.. we just finished our YouTube holdout study (with Stella) and we are seeing a 21% lift in test DMA compared to a counterfactual. 

CFO: English, moron. 


CMO: In regions where YouTube was live, we saw 21% more revenue. Meaning revenue caused by YouTube. 

CFO: how can we be sure it’s caused by YouTube? 

CMO: we have validity metrics like MAPE, R-Squared, Stat Sig, Trend, and iROAS range. They are the best way to know if we can trust the statistics other than doing a secondary holdout to verify the lift in different markets. 

CFO: you said 21% of revenue in the tested regions? How much is that? 

CMO: nearly $80k of incremental revenue. 


CFO: from $50k of spend on YouTube?

CMO: yeah 


CFO: and 35% of purchasers remember seeing our YouTube ads? 

CMO: exactly 


CFO: well why haven’t you explained it like that before? Let’s scale this thing. 

CMO: … hold on… let me tell you about Scale Tests….

Probably an insane take but I told a buddy I would literally go broke before I compromise my values at Stella. 

There are a lot of ways to make money in software.

Some of them work fast. Some of them look great from the outside. 

And some of them slowly push you into building things you don’t actually believe in.

You start optimizing for how you can squeeze every last penny out of your users...instead of optimizing for an actual great product. 

You ship features because they sell, not because they’re right.

You explain away problems instead of fixing them...because fixing them would slow shit down.

At some point, you realize you’re spending most of your time defending decisions you wouldn’t make again.

I don’t want that.

So yeah, I’ll leave money on the table if I have to.

I'll price lower than the entire market AND provide a better product. 

I’ll even say no to things that would make my life easier but the product worse.

I’m not anti-money. At all.

But I want to like what I’m building while I’m building it.

Some of the worst marketing decisions I’ve seen were backed by flawless analysis.

No fake numbers or bad data happening; it was good work, done by smart people who cared about getting it right.

But the decision still went the wrong way.

Most of the time, the problem was that everyone was hoping the data would make the decision for them.

So the team keeps asking for one more cut of the data even though they know it won't change anything.

Because once someone makes the call, there’s nothing left to point at if it doesn’t work.

So the analysis gets more refined. It gets easier to defend. And it definitely gets easier to explain to someone else in a room or on a slide.

But the situation doesn’t really change. Eventually, the work stops helping the decision and starts buying time instead. 

I’ve watched teams walk away from momentum because the numbers didn’t feel clean enough yet. Then a few weeks go by, the window closes, and everyone acts surprised.

The analysis did what it was supposed to do. It just wasn’t supposed to carry the responsibility for the outcome.

Sometimes data isn’t there to remove uncertainty. 

It’s there to help you understand what kind of uncertainty you’re choosing to live with.

Agencies are ruining incrementality testing for everyone else.

And they don’t even know it.


Here’s what I keep seeing:
An agency tells their client they do “gold standard measurement.” 

They pick a few random regions, turn off spend, run a model, get a number, and call it done.

They never post about their methodology publicly. Because once methodology is public, it has to survive real scrutiny.


But here’s the part that actually keeps me up at night.

It’s not that they’re doing it wrong. It’s what happens AFTER they do it wrong.


A marketing leader gets a 5x iROAS on YouTube. They triple spend to $300k/month. MER tanks. 

And now they walk into every future conversation saying “we tried incrementality testing, it doesn’t work.”

Or they run an MMM once as an annual report. No forward testing. No back testing. No iteration. And when the model misses, they say “MMM isn’t accurate.”

They didn’t fail at measurement. They failed at the process around measurement. But the methodology takes the blame.


This is why I get on demo calls every single week with marketing leaders who say:
“We tried incrementality before. It didn’t work.”
“We don’t believe in MMM anymore.”

They’re not wrong to be skeptical. They just got burned by someone who treated statistical analysis like a checkbox instead of a discipline.


An MMM isn’t a report. It’s a system.
A holdout test isn’t a one-time read. It’s a confidence interval.

An iROAS without an upper and lower bound isn’t a result. It’s a guess.


If you’re offering measurement services to clients, you either need to be a data scientist or you need to use tools purpose-built to guide the process end to end.

Because “directional” isn’t a flex. It’s a liability.
If you disagree, I’d genuinely love to hear how you’re running these tests. This is uncomfortable, but it’s worth talking about.

Everyone’s talking about #incrementality.
Nobody talks about what you actually DO with the results.

You run a holdout study on Google.
It comes back: 2.1x iROAS.

Cool.
…now what?

Do you cut Google? Scale it? Is 2.1x good? Compared to what?

What do you do with Meta’s number when it comes back next month?

How do you make a budget decision across 6 channels with 6 different iROAS numbers tested at different times, under different conditions?

This is where 90% of brands get stuck.
They have the measurement. They don’t have the system to act on it.

Here is what actually closing the loop looks like:

Step 1: Build the MMM first.
Before any holdout test, you need a baseline model built on 2 years of your spend and revenue data. It estimates each channel’s true contribution. This is your map.

Step 2: Forward test it.
Hide the last 3 months of data from the model. Ask it to predict that period. Compare prediction to reality. If it is within 10-15%, the model is trustworthy. If not, fix it before doing anything else. Do not skip this step.

Step 3: Run holdouts to validate the map.
Geo-holdouts on your top 2-3 channels. You are not just collecting a number. You are asking: does the model’s estimate match what actually happens when we turn this channel off in the real world?

Step 4: Feed the experiment results back into the model.
This is called Bayesian calibration. You are using the holdout result to sharpen the model’s estimates. Without this, most of our MMMs run at about 87% accuracy. With it: roughly 95%. That difference compounds every quarter you use the model to make decisions.

Step 5: Run the budget optimizer.
Now the model has enough signal to tell you: here is your current allocation, here is the optimized one, here is the predicted MER difference. That is your decision brief. That is the answer to “okay now what.”

Step 6: Move 10-15% of budget. Watch for 4 weeks. Run another forward test. Compare actuals to forecast.

If they track, the model is validated. Make a bigger move next quarter. If they don’t, something changed. Diagnose it. Adjust. Repeat.

That is the whole system.


The reason brands don’t do this isn’t because it’s technically hard.

It’s because nobody built them the infrastructure to actually run this cycle. So they pay for a test, get a number, put it in a slide, and go right back to optimizing toward platform ROAS.

That is the gap we built Stella to close.

If you are sitting on incrementality results and not sure what to do with them, DM me. Happy to walk through it.

It's not the $200K annual license that kills MMM adoption.
It's the hours of manual work. Every. Single. Month.

Here's what actually happens:
Your team decides to run MMM. You sign the contract. Great.

The vendor sends you a template: "Just format your data like this."
47 required columns.

Now someone spends an entire day:
Pulling Google Ads data. 
Pulling Meta data. 
Pulling email data. 
Pulling Shopify data. 
Formatting dates the right way. 
Checking for errors. 
Uploading. 
Realizing dates are still wrong. 
Starting over.

6 hours. Every month.
Want to run multiple models? (New customers vs existing customers?)
Now it's 12 hours of manual formatting per month.

So teams stop updating. They run it once or twice a year. The model goes stale. Predictions become useless.

And everyone blames "MMM" for not working.
The problem was never the model.


It was the invisible operational tax that nobody planned for.
This is why most MMMs die after the first quarter.

Not because the math is wrong.
Because moving data around in spreadsheets isn't anyone's actual job.


The DomoMMM (Powered by Stella) solves this for their customers - if your data is already in Domo, it's already formatted. Click to update. That's it.

Because the goal is insights, not Excel expertise.

With all the debate around what halftime show to watch, we are throwing our names into the hat.

Experience an event of culture, excitement and pizzazz with the Stella team. 

Discussing your favorite topics like: 
- What is incrementality testing in marketing
- How to pick the right locations for your holdouts
- What is Multicolinearity and how did it destroy my marriage

All that and more. 

At the Incrementality Masterclass Live Event on LinkedIn. 

Same time as the Superbowl Halftime Show. 

RSVP today. Please do. No one has RSVP'd yet.

taplio

The Battle

Arjan Schoorl

Brenden Delarua

Followers

Posts

Engagements

Likes

Performance ScoreBrenden Delarua's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Arjan Schoorl

Brenden Delarua

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Brenden Delarua's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreBrenden Delarua's LinkedIn Performance