Measurement & Data Analyst
There is no pitch team and delivery team. If you build it, you present it, and you answer the hard question about it.
The most junior person with a clean test result wins the argument. This is enforced, not aspirational.
We invest in the boring craft
Measurement design, data hygiene, rights management. The work that makes the interesting work possible.
Honest about the trade
A small team means narrower specialisation options and a lighter support structure. If you want a defined career ladder, a larger agency will probably serve you better, and we would rather say so early.
Open roles
Performance Media Manager
Media
Italy · hybrid · Full time
Own the planning and buying on a small number of accounts, end to end, with a principal alongside rather than above you.
The work Plan and buy across search, social, video and commerce media
Design creative tests that produce readable results
Run weekly pacing and quarterly reallocation
Present performance, including when it is bad What a fit looks like You have run meaningful budgets and can explain a decision you got wrong
You are suspicious of platform-reported ROAS
You would rather consolidate five tools than add a sixth Build the evidence layer: experiments, mix models, and the unglamorous data work that makes both trustworthy.
The work Design and run geo holdouts and matched-market tests
Maintain and calibrate marketing mix models
Audit attribution and conversion tracking setups
Turn results into a recommendation a CFO will accept What a fit looks like Comfortable in SQL and either Python or R
You can explain a confidence interval to a non-technical stakeholder
You have said 'this test will not reach significance' out loud before Build the generative production pipeline: stills, motion, versioning and localisation, inside brand rules and rights constraints.
The work Design and operate generative image and video workflows
Encode brand rulebooks as generation constraints
Build versioning and localisation automation
Own provenance, rights and disclosure hygiene What a fit looks like You have shipped generative work to a real brand, not just to a portfolio
You have opinions about where these tools are genuinely bad
You can talk to a designer and to an engineer without translating twice Own the pipelines and the modelling layer that everything else reads from, so the numbers in every dashboard agree.
The work Build and maintain ingestion from media and commerce platforms
Model a single measurement frame used by every downstream view
Server-side tagging and conversion API implementations
Data quality monitoring that fails loudly What a fit looks like Strong SQL, comfortable with a modern warehouse and transformation stack
You treat consent state as a first-class field, not an afterthought
You have cleaned up somebody else's tracking implementation and lived Hold a small portfolio of relationships, with enough craft depth to challenge the work rather than relay it.
The work Own the commercial relationship and the quarterly plan
Translate between a client's business problem and the practice teams
Keep engagements scoped honestly, including scoping them down
Be the person who delivers unwelcome findings What a fit looks like Agency or client-side background with real performance marketing depth
You can read a mix model output and ask a good question about it
You have talked a client out of something Italy · hybrid · Tempo pieno · entry level
A genuine entry route: structured mentoring, real account responsibility early, and no spreadsheet purgatory year.
The work Day-to-day campaign operation under a senior owner
Build reporting and spot what changed
Run test setups and document them properly
Learn the measurement stack from the inside What a fit looks like Numerate and curious; formal background matters less than evidence of both
You have built something, anywhere, and can talk about why it worked
01 A short call to work out whether the problem set interests you.
02 A conversation with the practice lead about work you have actually done.
03 A paid exercise drawn from a real (anonymised) problem, discussed rather than graded.
04 A final conversation about how you would want the first ninety days to go. Compensation is discussed openly from the second conversation onward. We do not publish bands here because we do not want to advertise a number we would then treat as a ceiling.
Nothing here fits?
We hire when we meet someone we should hire, not only when a role is open. If you work in performance media, measurement or applied AI and you like the way we write about this stuff, write to us.
The most junior person with a clean test result wins the argument. This is enforced, not aspirational.
We invest in the boring craft
Measurement design, data hygiene, rights management. The work that makes the interesting work possible.
Honest about the trade
A small team means narrower specialisation options and a lighter support structure. If you want a defined career ladder, a larger agency will probably serve you better, and we would rather say so early.
Open roles
Performance Media Manager
Media
Italy · hybrid · Full time
Own the planning and buying on a small number of accounts, end to end, with a principal alongside rather than above you.
The work Plan and buy across search, social, video and commerce media
Design creative tests that produce readable results
Run weekly pacing and quarterly reallocation
Present performance, including when it is bad What a fit looks like You have run meaningful budgets and can explain a decision you got wrong
You are suspicious of platform-reported ROAS
You would rather consolidate five tools than add a sixth Build the evidence layer: experiments, mix models, and the unglamorous data work that makes both trustworthy.
The work Design and run geo holdouts and matched-market tests
Maintain and calibrate marketing mix models
Audit attribution and conversion tracking setups
Turn results into a recommendation a CFO will accept What a fit looks like Comfortable in SQL and either Python or R
You can explain a confidence interval to a non-technical stakeholder
You have said 'this test will not reach significance' out loud before Build the generative production pipeline: stills, motion, versioning and localisation, inside brand rules and rights constraints.
The work Design and operate generative image and video workflows
Encode brand rulebooks as generation constraints
Build versioning and localisation automation
Own provenance, rights and disclosure hygiene What a fit looks like You have shipped generative work to a real brand, not just to a portfolio
You have opinions about where these tools are genuinely bad
You can talk to a designer and to an engineer without translating twice Own the pipelines and the modelling layer that everything else reads from, so the numbers in every dashboard agree.
The work Build and maintain ingestion from media and commerce platforms
Model a single measurement frame used by every downstream view
Server-side tagging and conversion API implementations
Data quality monitoring that fails loudly What a fit looks like Strong SQL, comfortable with a modern warehouse and transformation stack
You treat consent state as a first-class field, not an afterthought
You have cleaned up somebody else's tracking implementation and lived Hold a small portfolio of relationships, with enough craft depth to challenge the work rather than relay it.
The work Own the commercial relationship and the quarterly plan
Translate between a client's business problem and the practice teams
Keep engagements scoped honestly, including scoping them down
Be the person who delivers unwelcome findings What a fit looks like Agency or client-side background with real performance marketing depth
You can read a mix model output and ask a good question about it
You have talked a client out of something Italy · hybrid · Tempo pieno · entry level
A genuine entry route: structured mentoring, real account responsibility early, and no spreadsheet purgatory year.
The work Day-to-day campaign operation under a senior owner
Build reporting and spot what changed
Run test setups and document them properly
Learn the measurement stack from the inside What a fit looks like Numerate and curious; formal background matters less than evidence of both
You have built something, anywhere, and can talk about why it worked
01 A short call to work out whether the problem set interests you.
02 A conversation with the practice lead about work you have actually done.
03 A paid exercise drawn from a real (anonymised) problem, discussed rather than graded.
04 A final conversation about how you would want the first ninety days to go. Compensation is discussed openly from the second conversation onward. We do not publish bands here because we do not want to advertise a number we would then treat as a ceiling.
Nothing here fits?
We hire when we meet someone we should hire, not only when a role is open. If you work in performance media, measurement or applied AI and you like the way we write about this stuff, write to us.