Anton Larsson · Growth strategy & marketing economics

Selected work

Context, decision,
mechanism, result.

Four things I have built or run, each one an argument that measurement changes decisions.

01 · Marketing intelligence

Ess Group

One of Scandinavia's largest hotel groups

Their marketing was reported channel by channel, campaign by campaign. I convinced the entire executive team that it all comes down to one number: how many people book a room. That reframing changed their whole marketing setup.

I built the measurement and intelligence platform behind it myself, connecting spend, bookings and revenue into one view the leadership actually uses. Complex data, translated into decisions that move the whole business.

Executive reportingBooking attributionBuilt end to end

02 · Performance function

Stellar

Hybrid SaaS marketing group, Skeppsbron, Stockholm

Stellar acquires niched marketing agencies and packages their expertise into productized services for large companies in Scandinavia and beyond. I built the performance team from zero: hired the people, designed the offering, own the numbers.

The stack is mine too: server-side tracking, CAC and payback models per channel, and AI agents that automate the reporting so time goes to analysis. Client revenue scaled to 75M SEK in under two years.

Budget allocationCAC & payback modelsAI-automated reporting

03 · Founder

.stack & TeamStack

My own products, built solo. .stack replaces a small team's tool sprawl on a gainshare model, it only earns when the client saves. TeamStack renders ad creative programmatically from templates and a data feed, turning creative testing into a software problem.

dotstack.io ↗teamstack ↗

04 · Live lab

STHLMA

A music project run as a measurement experiment with my own money: Spotify Ads, discovery mechanics and detailed spend tracking, grown to 2M+ streams. Nothing teaches acquisition economics like your own budget on the line.

Listen on Spotify ↗

Approach

Measure. Explain. Decide.

Measure everything, in detail

Server-side tracking, clean event pipelines and channel-level unit economics, because blended numbers hide the truth and platform-reported numbers flatter themselves.

Explain it so decisions get made

A model no one understands changes nothing. I translate the data into the one number leadership cares about, and get marketing and finance operating from the same truth.

Automate the repetitive with AI

I write JavaScript and TypeScript, work daily in Claude Code and Cursor, and build agents that collapse the time between question and answer. Assembly is automated; thinking is not.

What I use AI for

The tool I never had.

I use it every day. Not as a productivity trick, more as the thing that finally closed a gap I have carried my whole career: I have always had too many ideas and a clear picture of how they should look, but never the tool. Now I can build more or less whatever I want and have it come out the way I saw it.

I built myself a server that holds the context for everything I do: every project, its learnings, its data and design files, in one place. Something I work out on a private build on a Sunday is available to me at work on Monday. That compounding is the most useful thing I have.

The other half is data. My instinct has always been to challenge a result rather than report it, and that instinct used to hit a wall of hours. Now I can look at all of it at once. The interesting things are almost always hiding in the parts nobody had time to open.

From buildathon to product

ads by dotstack

In June I was picked as one of fifteen teams for Pinecone Ventures' buildathon in Almedalen, with a one million prize and Lovable among the sponsors. I was outclassed, comfortably, and I am still proud to have been picked for the idea. It moved my sense of what is reasonable to attempt.

The idea became a product: an AI media buyer that never clocks out. It audits connected ad accounts around the clock and returns every finding as a proposal with evidence attached, plus the finished replacement ad, copy and landing page. Nothing publishes until a human approves. It thinks across platforms instead of inside one, and it is used by my team today and by one paying customer outside it.

ads.dotstack.io ↗

The story

I chose my own path, and it always led back to the numbers.

I am 34 and have been drawn to technology my whole life, alongside a fascination with sales and the psychology behind marketing. I chose my own path deliberately: starting companies, teaching myself as I went, always following the pull toward numbers, math and learning new things.

As one of eight founders at N1CE I measured every initiative myself, from festivals to online sales, while the company grew from zero to 100M SEK. At SALT, a startup training software developers, I learned to code alongside the students. At Stellar I got to build my own team, and we are trusted to challenge traditional marketing at some of the largest companies in Sweden and sometimes Europe.

With your own money on the line, the actual payoff is the only thing that matters. That perspective is the thread through everything on this page.

Résumé

Where the thread has run.

Since 2023

Head of Performance · Stellar

Built and lead the performance function. Own budget allocation, channel ROI and forecasting across all paid media, and the measurement stack from server-side tracking to CAC and payback models.

2019 – 2022

Technical Marketing · SALT, salt.dev

Ran marketing and acquisition for a Stockholm startup training software developers, and learned to code alongside the students.

Until 2018

COO & Co-founder · N1CE Company

One of eight founders of a consumer brand tied to Swedish House Mafia. Owned operations and digital marketing across European markets as revenue grew from zero to 100M SEK. Exited in 2018.

Context, decision,mechanism, result.