Founder, P33 | Executive Thought Leadership Agency

Helped 50+ B2B tech companies build their Founder/CEO Brand and implement Executive Thought Leadership Programs for LinkedIn. Customers, results & process: project33.io

About me:

I'm Finn, 29, grew up in Bremen, Germany, studied Physics & Philosophy in Heidelberg for 3 semesters, dropped out of college in 2017 because I was doing it for other people, flew to New York with no plans, documented the journey on Linkedin, Instagram and Youtube, and started my first company (Project 33) on the back of having grown a small audience. We're a small remote agency that helps B2B founders & execs create thought leadership content for Linkedin.

In my free time, which I try to have as much as possible of, I read (fav author: Nassim N. Taleb), write, run, lift, code (currently building an app), cook, make Youtube videos, turn books into Instagram accounts, grapple (BJJ), and travel (lived in NYC, Cape Town, Bali, Minsk, CMDX, Lima & Antalya so far). In December 2020 I married the most beautiful woman in the world. In January 2023, I became dad to the most beautiful baby girl. We currently live in Berlin, Germany.

Finn Thormeier

Data Strategy and Optimisation expert - More Revenue, Less Cost

I’m a former investment banker with over a decade of experience in financial modeling, trading, and asset pricing.

Before founding Growth Dynamics, I worked as a Data Scientist at Skai and later as Measurement Manager and Principal Data Strategist at Oracle, helping global brands connect their marketing investments to real business outcomes.

That background taught me one thing: models only matter if they hold up in reality.

In finance, I learned that theory often breaks the moment it meets the market. Marketing is no different.

At Growth Dynamics, we bring that same real-world discipline to measurement. We specialize in long consideration journeys and offline sales, the complex paths where attribution breaks and intuition often misleads.

We’ve been on the side of operators, owning budgets, defending investments, and making trade-offs that theory can’t capture.

Our goal isn’t to build elegant models but useful ones that drive measurable impact and profitable growth.

Charles-Antoine D.

The 3 main questions I get asked all the time have the same answer: brand.

❓ What drives sales in my brick and mortar stores?
❓ What improves the effectiveness of my performance marketing?
❓ What makes us show up in AI results?

Brand. Brand equity. Stored preference.

Yet most budget goes to capturing demand instead of building it.

No, performance marketing is not brand, anything that has a ROI target is not brand.

1. Performance marketing

If people already know you, trust you, and are predisposed to choose you, everything downstream gets cheaper. That is literally how the system works.

Also, brand makes paid media cheaper through the mechanics of auction platforms. Higher expected CTR and relevance improve Quality Score, which improves ad rank and can reduce CPC. 

2. Brick and mortar

Stores are a conversion surface. People walk in when your brand is mentally available, not when you “optimise the door”. (also great OOH by the way)

If you keep pushing budget into bottom funnel, you hit what we call the Deathzone. You keep spending, but you are not building demand, so CAC rises and growth stalls. 

3. AI visibility

The same story, with data.

63% of AI visibility is explained by long term brand equity, and the strongest correlation is branded web mentions (r = 0.664). “GEO tactics” explain about 11%. 

So no, you do not “hack” AI discovery with an AI playbook. You build the thing that AI reflects: brand strength, expressed in real-world mentions and credibility. 

So what should you measure instead of attribution paths?

Measure demand and preference then prove incrementality.

Bottom line
Attribution tells you how the sale happened.
Brand explains why you were chosen.

There is no hack.

Build brand equity, and performance, retail, and AI visibility all compound.

Charlie de Thibault

Turn $25m+ media budgets into confident growth decisions

We have an incrementality framework that has been validated by measurement leads across some of the biggest channels in the industry.

It still surprises me how many teams and agencies rely on methods that routinely fail.

The methods people overuse

Pre-post tests and interrupted time series are everywhere. Easy to run but they do not account for seasonality or confounders. If you can run a design with a control group, you should.

Geo selection is the job

Parallel trends between your control and treatment is key. They are not something you assume. They are something you validate, quantify, and pressure-test. Many teams optimise the pre-period fit and call it done. That is not enough.

You also cannot fully automate this. You need to inspect each geo individually: which ones drive the trend, which introduce volatility, which have different seasonality or market dynamics.

Why we often prefer diff-in-diff over synthetic control

Three reasons.

First, weekly seasonality. Marketing has strong weekly patterns, and synthetic control requires tight alignment across geos that often does not hold for high ticket products, offline sales, and long consideration journeys.

Second, structural geo differences in ROI. A counterfactual that fits the trend can still be conceptually wrong if underlying geo economics differ materially.

Third, impact differences across geos. Each geo behaves differently, and a single event can move one geo in isolation. With synthetic control you are completely blind to this reality.


Three mistakes that invalidate results

1. Testing on 10% of the market gives you weak external validity. This means you cannot assume the results will be the same on the rest of your market
2. Reporting a single aggregate lift obscures how much your geo selection influenced the number.
3. Comparing lift tests across channels is misleading. Short-term channels are mathematically advantaged: more of their impact shows up within the test window, above the noise. Extending the test window only partially solves this, because the only signal you can capture is the part that rises above noise. This is especially punishing for smaller channels and channels with longer-term effects.

Do not let lift tests push you to underinvest in those channels.

The bottom line

Incrementality credibility comes from rigorous geo selection, validated parallel trends, and treating geos as structurally different markets, not interchangeable model inputs.
______
We put all our learnings into one growth playbook.
👇 
 DM me if you would like a copy. 
👇 
You can also subscribe to our One Move to Grow newsletter and private Slack channel. 😊

The Death Zone
A lot of brands hit a level of spend where growth stalls.

 CAC rises. 
ROAS flattens.
 The team optimises harder. 
Nothing moves.
We call it the Death Zone.

The instinct makes sense. 
Push more budget into what is trackable.

 The problem is that trackability does not prove incrementality.
 When we look at the data, a lot of spend that looks efficient is not actually adding growth. And the spend that does add growth often sits in brand, where attribution makes it invisible.

This is why a lot of teams get stuck. 

The dashboard says performance is working. 
The P&L says otherwise.
When we help clients escape the Death Zone, the fix is rarely more optimisation. It is cutting some performance budget and reallocating into brand. 

The cascading effect looks like this: brand investment builds share of search. 

Share of search improves quality scores. Quality scores lower CPC and improve auction outcomes. 
Performance marketing starts working again.
We saw this with one client. 

Share of search rose over three and a half years.
 Revenue followed with around 125% growth. 

The performance metrics improved because the system changed, not because the targeting got better.

If growth has stalled and the team is stuck optimising in circles, the problem is usually upstream. This is a capital allocation question, not a channel optimisation question.

Brand is not a soft investment you fund after performance is sorted.
 It sets the price performance charges you.

We put all our learnings into one growth playbook.
👇 
 DM me if you would like a copy. 
👇 
You can also subscribe to our One Move to Grow newsletter and private Slack channel. 😊

You should not treat a sponsored incrementality test as truth. “We offer a free lift test with our partner.” We have all heard it. Be careful.

Let’s be real: when the channel pays for the measurement, the incentives are not neutral. I have seen channels drop partners after “no lift” outcomes.

That does not mean the numbers are fake. It means you should not accept the headline at face value.

There are many ways a lift test can look better, without anyone lying:

👉 Biased holdout construction: control group is systematically lower intent (worse customers, weaker geos, cheaper inventory) so the gap looks like lift
👉 Non-parallel trends: test and control were already diverging before the experiment, then the difference gets attributed to ads
👉 Poor randomisation / non-random assignment: platform delivery or targeting differences create confounding that boosts the measured effect
👉 Post-treatment exclusions: removing “messy” users, placements, or geos after seeing results (often drops negatives and raises the average)
👉 Mismatch in exposure opportunity: test group has more chances to convert (availability, coverage, frequency), so you measure access, not incrementality
👉 Attribution window and conversion definition: expanding windows, counting view-through, counting weaker proxy conversions
👉 Segment mining: reporting the best looking cohort/geo cut rather than the pre-registered primary result

So the “free” test can be expensive, because you end up optimising to an inflated number.

My rule: when you hear “free lift test”, think twice. If you have already done it, treat the result as directional, not truth, until the vendor shares the full methodology, limitations, and the raw breakdowns you need to sanity check it.

Good gut check: if the vendor is not proactively sharing caveats, you are not seeing the full picture. Every test has limitations.
________________
👇 
We put all our learnings into one growth playbook.
👇 
 DM me if you'd like a copy. Also reach out if you want to subscribe to our One Move to Grow newsletter and join our private Slack channel. 😊

Episode 3: The Link Between Brand and Lower CPC is live

I explain how brand equity shows up inside performance marketing:
 • Why being in the initial consideration set lowers CPC
 • How brand familiarity feeds into Quality Score
 • Why expected CTR and relevance aren’t purely tactical
 • And how the same mechanics apply on Meta

CPC is downstream of demand, familiarity, and preference.
Optimize only the bottom of the funnel and you hit diminishing returns.

In many cases, the fix isn’t better performance tactics.
It’s rebuilding brand demand so performance can work again.

Link in the comment section

Bot traffic. Ashburn. A town of ~50k people, showing up among the top US cities for website activity.

Some of you may have spotted “AWS us-east-1” on the map.

Home to a lot of virtual machines.

This is just the top of the iceberg. It spots the lazy automation that does not even bother with a residential proxy to hide real location.

Tip: Look at the campaigns driving the most Ashburn traffic, and the campaigns where more than 5% of traffic comes from Ashburn. When a town of around 50k people drives that much traffic, you almost certainly have an issue. I have seen some US display campaigns with 50% of traffic coming from there.

If you’re wondering why you’re seeing so much traffic from Des Moines, it’s for the same reason: Microsoft Azure has a data centre there.

A lot of the rest can be surfaced through time series anomaly detection. It is a simple tool, but it is already an excellent exercise I run with clients to show that bot traffic is not a small issue, and that some channels are far more prone to it than others.

𝐈𝐟 𝐲𝐨𝐮’𝐫𝐞 𝐢𝐧𝐭𝐞𝐫𝐞𝐬𝐭𝐞𝐝, 𝐰𝐞 𝐜𝐚𝐧 𝐫𝐮𝐧 𝐚 𝐪𝐮𝐢𝐜𝐤 𝐬𝐜𝐫𝐞𝐞𝐧𝐢𝐧𝐠 𝐟𝐨𝐫 𝐲𝐨𝐮.

Also,

We put all our learnings into one growth playbook.
👇
DM me if you’d like a copy.
👇
And reach out if you want to subscribe to our One Move to Grow newsletter or join our private Slack channel.

Why 80% of your buyers already knew the winner before they started looking.

In audits, we see the same pattern repeatedly. 

Around 80% of buyers purchase a brand they already knew before starting their journey.
 If you are not in the initial consideration set, your odds drop sharply. 

Your CPC rises because the platform prices in lower expected performance.

Here is what a lot of teams miss. 
69% of brands bought by switchers were already in the consideration set. 

The switching did not happen because of better targeting at point of purchase.
 It happened because the name was already familiar when the decision started.

A lot of performance budgets concentrate on the narrow slice of people buying right now. 

When we look at the data, this creates a problem. You end up competing in a small, expensive pool while leaving future demand to competitors. That shows up as rising CAC even when dashboards say spend is efficient.

We saw this with one client. 

Share of search increased over three and a half years. 
Revenue followed with roughly 125% growth. 
The awareness moved first. The revenue came after.

The reallocation we often recommend: cut 10 to 15% from bottom-performing acquisition spend where marginal returns are flattening. 

Move it into brand activity that builds share of search. Protect the campaigns where payback is already strong. 

Track share of search monthly alongside CAC. 

This is a capital allocation decision, not a channel decision.
👇 
We put all our learnings into one growth playbook.
👇 
 DM me if you would like a copy. You can also subscribe to our One Move to Grow newsletter and private Slack channel

#Odoo stopped Google Ads as an experiment: 5m€ saved per month...

- 120k leads per month, 19% of all our leads
- conversion rate 3-7x less than other leads (depending on region)
- value was 25% less than regular leads

The 5m€/month question: is it better to get more leads, or less leads to spend more time on valuable leads? (don't have the answer yet) Is this worth 5m€ / month?

Here is our last 30 days graph (we stopped a week ago)

Fortunately, over the past years, we diverted our marketing efforts to build up other sources of leads. We went from 50% of our leads from Google Ads & Bing (CPC) to 19%.

The large majority of our leads are organic now: word of mouth. The power of a strong community...

Most marketing allocation falls into one of two traps.

Trap one: optimise to revenue only
You over-invest in high-intent markets, short-term capture looks great, and the top of the funnel quietly starves. Spend becomes less efficient over time because you never built the demand to feed it. That is where most attribution, MMM, and incrementality tools will lead you.

The problem is structural. A buyer journey is awareness, consideration, purchase. And your product is twice as likely to be chosen if you are in the initial consideration set. Optimising towards revenue only is missing that important part.

Trap two: optimise to upper funnel metrics.
You chase cheap reach. Spend flows to the wrong geographies because the CPM is lower there. Not because demand is there.

Both ignore spatial demand structure. Nobody connects funnel stage with DMA-level economics.

That's the gap we solve.

If you're a senior marketing leader at a high-consideration B2C brand and your budget conversation keeps circling back to brand vs. performance, it's probably the wrong question.

The right question is which markets need awareness and consideration now to unlock conversion efficiency later.

Happy to show you what that looks like in practice.

This bag sells for €2,500. It costs €55 to make.

You cannot achieve a 5,000% markup with a Google Search ad. 

That margin exists because the brand created preference that overrides price comparison.
Yet a lot of companies we audit bet everything on performance marketing. They treat brand as fluff because attribution cannot measure it.
Performance marketing rents attention. 

Brand building owns it.

If you rely on performance channels, you are constantly pressured to discount. A strong brand is the only asset that protects margin, lowers CAC (the total cost to acquire one customer), and raises LTV (the total profit a customer generates over time).

Here is the maths a lot of teams miss:

Only 5% of buyers are in-market at any time. Of those, 80% already have a preferred brand before they start searching.

That leaves 1% who are ready to buy and open to a brand they do not know.
If you are dumping 90% of budget into performance marketing, you are fighting over that 1% sliver.
 Meanwhile, you are starving the brand investment that would get you into that 80% consideration set before they ever search.

We put all our learnings into one growth playbook. 
👇 
DM me if you would like a copy. 
👇 
Also reach out if you want to subscribe to our One Move to Grow newsletter or join our private Slack channel.

New Episode: Does Paid Brand Search Drive Real Growth? 

I break down how brand bidding actually shows up in your numbers:

• Why paid brand search campaigns often look efficient
• Where incrementality gets overstated
• How to think about brand search as a capital allocation decision
• And when defending brand does make sense

Paid brand search often captures demand. It does not always create it.

Video Link in the comments.

Do you know your ROI dispersion across geos? 

Teams look at blended ROI. Maybe channel-level ROI. Almost never geo-level dispersion. 

But when you actually plot it, the picture is very clear. 

In many accounts, ROI varies more than 2x across markets. 

Same brand, same creatives, same budget logic. Wildly different outcomes. 

The reason is structural. 

Google does not see Meta. Meta does not see TikTok. None of them see offline revenue cleanly. 

They all optimise from incomplete feedback and the quality of that feedback is not evenly distributed. 

The platforms learn well where the data is clean and complete. They learn badly where it is not.

If your geo ROI differs by 100%+ it is an optimisation lever. 

Before you run another test, debate lift, or blame creative look at the distribution. 

If dispersion is extreme, the biggest opportunity might not be in channels. It might be in geo execution.

A lot of geo experiment vendors are selling results they cannot fully defend.
Meta GeoLift, Google GeoX,.... They all use some flavor of synthetic control. They all make the same implicit assumption: 

If pre-trends match, the counterfactual is valid. Nobody questions it because it is mathematically elegant, it produces confident-looking outputs, and clients do not push back because they do not know what is inside the box.

Here is what that assumption actually says:
"Whatever would have happened to your treated markets absent the intervention, the control markets would have experienced the same thing."

Have you ever seen the same test setup run with a different geo split give completely different results? That is not a coincidence. It is these assumptions breaking. Two DMAs can have similar revenue trends in the pre-period for completely different underlying reasons. Different price elasticity. Different competitive pressure. Different brand penetration. The model does not know that. It just sees two lines going in the same direction and calls it a valid counterfactual.

This is why it is extremely risky to run a test on a few DMAs and assume it reflects national impact. You are not measuring a universal truth. You are measuring what happened in those specific markets, under those specific conditions, with those specific structural assumptions baked in.

Always ask your vendor for a DMA-level split analysis. Not just the aggregate number. Every single market. If the story only holds when you squint at the top line, it is not a story worth acting on.
👇 👇 👇 
And if your vendor is making these assumptions without telling you, reach out. We can show you a better alternative.

A €2,500 bag that costs €55 to make isn’t explained by performance marketing.

In audits, we see this margin comes from preference that overrides price comparison.

Performance rents attention. Brand owns it. And ownership is what protects margin, lowers CAC, and raises LTV.

 
We put all our learnings into one growth playbook. DM me if you'd like a copy. Also reach out if you want to subscribe to our One Move to Grow newsletter and join our private Slack channel.

You're Fighting Over 1% of the Market

80% of buyers pick a brand they already knew before starting their journey.

Only 5% of your market is in-market at any given time.
Do the maths. 

If 80% already have a favourite and only 5% are buying, you're fighting over roughly 1% of the market with your performance spend.

That's why returns decay. 
That's why CAC keeps climbing. You're not bad at performance marketing. You've just exhausted the tiny pool of people who were ever going to buy from you.

A lot of teams respond by spending more on capture. Bidding harder. Optimizing harder.

That's like fishing the same pond faster after you've already caught everything in it.

The fix isn't more efficient. It's making more people willing to buy from you in the first place.
👇 
We put all our learnings into one growth playbook.
👇 
 DM me if you'd like a copy. Also reach out if you want to subscribe to our One Move to Grow newsletter and join our private Slack channel. 😊

Your MTA Is Pointing Your Budget in the Wrong Direction
One client’s MTA platform suggested social media deserved 2.5% of the budget.
Incrementality testing put the real contribution closer to 30%.

MTA only sees touchpoints that can be directly linked to a conversion. 

Social often works earlier. 
It builds preference weeks before purchase. 

Someone sees an ad, remembers the brand, searches later, and converts. 
MTA credits the search.
Budgets follow the credit.
Social gets cut.

More money moves to bottom-funnel channels.
CAC rises. 
Payback stretches. 
Margin tightens.

Nothing looks broken on the dashboard.
 Efficiency appears to improve.
 The risk shows up later.

The system is doing what it was designed to do. The problem is what it cannot see.

If your measurement keeps pointing budget toward the same conversion channels, the right question is not whether they convert. 
It’s what demand-creating activity is being mispriced or ignored, and what that blind spot is costing you over time.
👇 
We put all our learnings into one growth playbook. 
👇 
DM me if you'd like a copy. Also reach out if you want to subscribe to our One Move to Grow newsletter and join our private Slack channel. 👏

To help companies move money from performance marketing to brand, I use the exact strategy I used in trading. Minimise downside risk. Maximise upside.

The honest answer to "will it work?" is: we don't know.

And that is ok. As long as you know you don't know and don't mistake confidence for certainty.

Once you acknowledge it, you approach things differently. 

Just like you approach things differently when you focus on growth instead of measurement.

Your objective becomes finding out how to minimise risk and maximise outcomes.

This is why the first thing we do is look for diminishing returns. 

When you find budget that does not drive incremental revenue, the downside is very limited.

We find it in about 20% of conversion spend across the brands we work with.

If you move money that isn't driving any revenue to brand, the worst that can happen is nothing. Your ROI remains the same. 

And in the better alternative, the one we want to achieve, you increase your ROI.

Most companies think they don't have budget for brand. It's right there in front of them.
👇 👇 👇 
If you're part of the majority of CMOs and growth leaders who want to increase brand investment but struggle to make it happen, let's talk. This is what I do.

I've tried journaling more times than I can count. Never stuck. Then I set up a Claude Project for it. Now it actually works.

Here's how it runs:
Every  evening type "start journal". It walks me through 8 questions each time.

What's your win today? 
What's your biggest challenge? 
How did you take care of yourself? 
How overwhelmed do you feel (1-10)? 
Did you have a meaningful connection? 
Did you do something just for you? 
One thing you'd do differently? What are you grateful for?

A few things that make it genuinely useful:
It corrects your grammar and typos as you go, so you can write fast and messy without it mattering.

It picks up on themes. Partway through it'll say something like "a theme is emerging here" and reflect back what you're actually saying. It's not just recording, it's listening.

At the end it spits out a structured spreadsheet row with all your answers. I copy it into a Google Sheet. By the end of the month you have a full log you can actually analyze. Patterns in mood, energy, what you're avoiding, what keeps coming up.

Simple to set up. Full instructions in the comments if you want to copy it.

taplio

The Battle

Finn Thormeier

Charles-Antoine D.

Followers

Posts

Engagements

Likes

Performance ScoreCharles-Antoine D.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Finn Thormeier

Charles-Antoine D.

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);}Charles-Antoine D.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreCharles-Antoine D.'s LinkedIn Performance