Influencer Marketing Engineer

Hakon Junge

Influencer marketing operator who builds the systems he runs on.

Twelve years in GTM — Pleo, Capdesk, Figures, Timeleft — now applying agent engineering to creator partnerships.

The influencer workflow, run as systems —
six of them cover this loop end to end

6systems built & shipped in five months — the whole team runs them today
14×monthly subscriptions from the creator book, month 1 → month 5
89%of the team’s Q3 subscriptions from net-new creators — from my book, in 3 months
−95%cost per subscription over five months — after the channel review I initiated
0 → 541creators sourced & vetted per month, automated
4 h → 30 mindaily inbox time on a live negotiating book — the freed hours went back into creators & strategy
§01

The systems — what I build

Six systems that run together as one operating loop for creator partnerships — doctrine and versioned code the whole team shares, not tribal knowledge. Kittl is where they shipped first. Pick a module; each has a plain overview and the full schematic as engineered, and every node is readable.

Shipped as a library

32 versioned skills in a self-updating, CI-enforced library — every improvement lands in the team's own terminals automatically, instead of being trained on it.

NOTE — Schematics are simplified for confidentiality: internal identifiers, data-model names, benchmarks and teammate details are removed. The mechanics are as shipped.

TOOLING — BigQuery / SQL · HubSpot (pipelines, custom objects, automation) · Metabase · Meta Ads · GA4 & UTM attribution · Stripe promo-code attribution · Tipalti · Python · Gmail API · Claude Code (agent & skill development, MCP) · Git / CI · Clay · Notion

§02

One case study — Kittl

§01 is what I build; this is one case study of it working. The systems were switched on inside an already-mature creator program — one that had worked with nearly everyone in its space and was struggling to find net-new, high-value creators, with no fresh pipeline to inherit. Sourcing and activating that new bench was the very challenge they were built to solve. Run end to end from one seat, with the results reviewed and improved week over week. These are the production numbers.

CompanyKittl — AI design platform
WindowApr → Sept 2026
Starting bookZero — net-new, self-sourced
OperatorOne human + the systems
14×

more monthly subscriptions by month five than month one — from a book built entirely from zero, every subscription attributed to its creator.

86%

of subscriptions come from creators I sourced myself — I built the book, I didn’t inherit it.

49%

of tested creators turned into repeat, mid-to-long-term partners rather than one-off posts — so the book compounds every month instead of resetting to zero.

1.54×

cash-positive in month five: the book's revenue exceeded its cost. Reached while rates fell and relationships held.

What this looks like inside a team

None of this is about replacing people — it's about what an influencer team stops spending its days on. Every role in this space carries hours of manual work that can be automated, measured and improved week over week. Building that leverage, and handing it to the team, is the job.

One seat, several functions

Sourcing, research, pricing, ops and measurement ran from one seat here — operator-grade output from a single hire, before any headcount conversation.

The team gets faster, not smaller

Every workflow ships as a versioned skill teammates run from their own terminal, in their own voice. The systems lift the whole team's ceiling — human time goes to creators and strategy.

Compounding, week over week

Drafts are scored, gates tuned, rates recalibrated — the systems are reviewed and improved weekly, so the same team gets better results every month.

Kittl is the worked example. The doctrine transfers — to a brand, and to the team already running it.

Start the conversation — junge.hakon@gmail.com →
§03

Track record

Twelve years across GTM and partnerships — from first sales hire to CCO to building the influencer-marketing-engineer role itself.

MSc, Drug Discovery & Pharma ManagementUniversity College London · 2012 – 2013
BSc, PharmacologyThe University of Edinburgh · 2008 – 2012
§04

Writing & talks

Featured · about the role "From Influencer Marketing Manager to Influencer Marketing Engineer" Yuliia Maryniak, Substack — a profile of the discipline this six-month project defined. ↗
§05

Off duty

Danish & Dutch, raised in the Caribbean, based in Berlin — still working on the German, fully committed to the French wine.

🏝️Caribbean start

Grew up on Sint Maarten — home to Princess Juliana Airport, one of the world's most dramatic runways.

🤸Standing backflip

Joined a university hip-hop crew and competed across the UK — badly, happily. The backflip survived.

🧙"Spencer" of Edinburgh

Freelance tour-guide alter ego, heavy on the Harry Potter lore. 17 five-star reviews on Tripadvisor.

🎙️Unsung Heroes

Hosted a podcast curated by employees, with content dedicated to employees.

🏃Community builder

Ran Founders Running Club Berlin; now building an influencer-marketing community and a HYROX crew.

⚡ENTP-A

Debater profile: starts the systems, recruits the believers, argues both sides for sport.

§06

Building a creator program that should run itself?

Open to conversations about influencer marketing engineering — building creator operations, agent systems and the measurement that keeps them honest. The offer is the §01 → §02 loop: build the systems, prove them in production, and leave your team owning them — a team that gets faster, not smaller.

junge.hakon@gmail.com →