Work and portfolio
We read what they claim, then ask what they personally built, changed, measured or repaired.
Screening notes
We interview the work, not the vocabulary.
A tidy CV can be written in an afternoon. The desk is looking for range: what the person has actually owned, where their knowledge thins out, and what they do when the familiar answer runs out.
The point of the conversation
Confidence is useful. Operating range is the hire.
We want to know what they can do without a script: diagnose, build, decide, and keep moving. Terminology is cheap. Ownership is not.
The useful question is not “Do you know this?” It is “What do you do when you don’t?”
Search, paid media, analytics, social platforms, tracking and AI tools change constantly. Employers need people who can investigate, not wait for a complete instruction manual.
Signals we keep
People who can think, build and report.
- Hands-on examples, not channel slogans.
- A way of breaking an unfamiliar problem into next steps.
- The habit of researching what they do not yet know.
- Enough depth for the seniority on the brief.
- Judgement about what they can decide and what must be escalated.
- Clear communication when something is blocked or going wrong.
Sequence
Not every vacancy needs every stage. The aim is the same.
We adapt to craft and seniority. We do not stop at the first person who sounds plausible.
Probe the claim
Broad answers get narrower until we can tell experience from memorised talking points.
Bounded practical
Where it helps: a time-boxed audit, campaign scenario, brief or controlled site. Not unpaid client work.
The edge of the craft
We may walk past familiar ground on purpose, to see how they behave when the map runs out.
Figure-it-out
Questions they ask, assumptions they challenge, what they would look up, and how they form a plan.
Ownership
Can they take a seat, report progress, and escalate the decisions that actually matter?
Craft-specific checks
A generic marketing quiz tells us almost nothing.
The test should look like Monday morning in that seat.
Search
Review a site, explain a traffic drop, rank technical issues, or walk through an indexing problem.
Paid & performance
Read structure, find wasted spend, diagnose a conversion fall, or sketch a test plan from thin data.
Content & social
Turn a commercial goal into a brief, critique a calendar, or explain intent without “post more” as the plan.
Analytics & CRO
Read a messy picture, name what is missing, and say what evidence is required before a recommendation.
Automation & AI
Map a workflow, find failure points, pick tools, and investigate an integration they have not seen before.
Web & technical
Trace a front-end issue, reason about performance, or explain how they would approach an unfamiliar stack.
Independence
A self-starter still keeps people informed.
We want progress without hand-holding. That is not the same as guessing for three days in silence.
Five habits we actually look for
- Clarify the outcome and the constraints before charging ahead.
- Start with what can be solved without a meeting.
- Research the unfamiliar instead of stopping at “I don’t know”.
- Surface risks, blockers and assumptions early.
- Escalate budget, client, legal, security or strategy decisions.
A self-starter is not someone who never asks. It is someone who knows when asking is the work.
Employer questions
How the desk actually runs a search.
Does every candidate sit a practical?
No. We use a bounded test when the CV and conversation still leave a gap. Seniority and craft decide.
Do they do unpaid live client work?
No. Assessments can be realistic without becoming free production. Time-boxed, controlled, discarded after.
What does “self-starter” mean here?
Someone who can form a plan, research the gap, and keep moving — while still knowing when to escalate.
Why push past the rehearsed answer?
Because the stack changes. You are hiring the next six months, not a story about the last role.
Shape the screen to the seat
Tell us the outcome. We will test for it.
Role, independence, must-haves and the work they will touch in week one. That is enough to start.