Creative Strategist Interview Questions: 25 Prompts + Portfolio Review + Case Study Test

Why portfolio decks fail as a hiring filter

Every senior creative strategist arrives with a beautiful deck. Five winners. Three case studies. A funnel diagram. The deck is a sales document for past wins. It tells you almost nothing about how the candidate decides what to ship next, when to kill a concept, and how to read ad-level data without flinching.

Three structural problems with deck-led interviews. The pattern echoes the broader research summarized in Google's re:Work hiring guide, which found unstructured portfolio reviews are among the lowest-validity hiring signals available.

1. Survivorship bias by construction. Decks show the ads that worked. They omit the 40 that died, the 12 weeks of flat CPA, the brief that got walked back twice. You hire the storyteller, not the operator.
2. No iteration trace. A static slide cannot show the version history. You see V7 of the hook. You do not see what V1 through V6 taught the strategist about the audience.
3. Measurement is asserted, not defended. "We hit 3.2x ROAS" is a number on a slide. Ask which attribution window, which platform, which audience cohort, and the deck stops answering.

The fix is process, not better decks. Replace half the slide-walking time with structured prompts that force the candidate to reason in the open. The 25 creative strategist interview questions below are organized around the five capabilities that actually predict on-the-job performance: research, hypothesis, iteration, measurement, and communication. Each capability gets five prompts and at least one red-flag answer to listen for.

If you want broader context on the role itself before designing the loop, the creative strategist career path and creative strategist job overview breakdowns map seniority bands, expected workflow, and salary ranges. The Creative Strategist Workflow use-case shows what a competent week looks like end-to-end. Read both before you finalize your scorecard. The structured-interview validity literature is summarized cleanly in the SHRM hiring assessment guidance, if you want academic backing for the loop design.

This is the take from inside paid-media practice: portfolios pass everyone above a threshold of taste, and taste is necessary but not sufficient. Iteration cadence and ad-level data fluency are the load-bearing skills. Test those directly.

How to run the loop in 4 stages

Run a four-stage loop. Stage 1 is a 30-minute intro call. Stage 2 is a structured 60-minute interview built from the prompts in this guide. Stage 3 is a 90-minute paid case study using a live competitor audit on adlibrary. Stage 4 is a culture and team-fit conversation. Anyone progressing past Stage 2 should be paid for Stage 3, no exceptions.

Time budget per candidate, end to end:

Five reasons this loop outperforms the standard portfolio-and-vibes interview.

You see the candidate think in real time during Stage 2. You see them produce a deliverable on data they have never touched in Stage 3. You score with a rubric, not with feel. You compare candidates side by side on the same prompts. And you respect the candidate's time by paying for the only stage that asks them to do real work.

The cost of running this loop on three candidates is roughly $900–$1,500 in case-study fees plus 9–11 internal hours. The cost of a wrong creative strategist hire, measured in dead concepts and walked-back briefs across one quarter, is closer to $40,000 in misallocated paid spend. Spend the $1,500.

Twenty-five creative strategist interview questions

Five prompts per capability. Use 12–16 in a 60-minute Stage 2. Pick the prompts that map to the seniority band you are hiring for. Do not read the prompt verbatim. Read it, then probe.

For every prompt below, three signals matter more than the answer itself. Specificity (does the candidate name a real brand, a real number, a real platform feature?). Structure (does the answer have a method, or is it improvised?). Self-awareness (does the candidate name what they do not know?).

Research — 5 questions

These probe how the candidate finds an angle before funding it.

1. "Walk me through how you discover a new creative angle for a category you've never touched." Listen for: a defined research order (in-market ads first, customer research second, brand archive third), a time-box, and a stop rule. Red flag: "I just brainstorm with the team."
2. "What's your method for a competitive ad audit? Take me through the last one you ran." Listen for: a tool (adlibrary, Meta Ad Library, TikTok Creative Center), a sample size, a tagging schema, an output format. Red flag: scrolling Meta Ad Library with no notes.
3. "How do you map audience pain points to creative hooks?" Listen for: a primary source (review mining, support tickets, sales-call transcripts, Reddit), a translation layer (pain → promise → hook), and a falsification check. Red flag: persona documents with no quotes.
4. "You have 48 hours to produce 10 angle hypotheses for a new client in a category you don't know. What do you do hour by hour?" Listen for: a concrete schedule, named tools, named deliverables, and an explicit prioritization rule. Red flag: "I'd block out the morning to think."
5. "What sources do you trust for in-market evidence vs which sources do you ignore?" Listen for: a hierarchy (live ad inventory > case studies > vendor blog posts), explicit skepticism of agency-published wins, and a method for distinguishing live spend from impression-buying. Red flag: ranking LinkedIn posts as a primary source.

Hypothesis — 5 questions

These probe whether the candidate can write a brief that survives contact with reality.

1. "Show me a brief you wrote that turned out to be wrong. What was wrong and how did you find out?" Listen for: the original hypothesis stated cleanly, the falsifying signal, the lag between launch and discovery. Red flag: no example or "the brief was right; the execution was off."
2. "Define success criteria for a cold-traffic test launching tomorrow. Be specific." Listen for: primary KPI, secondary KPIs, sample size, decision window, kill threshold, scale threshold. Red flag: "We'd see how it performs."
3. "How do you structure a creative test so the result is interpretable?" Listen for: one variable per cell, a control, a minimum spend, an attribution window decision made before launch, a writeup template. Red flag: testing four things at once and "looking at trends."
4. "What's the difference between a hook test and an angle test? When do you run each?" Listen for: angle = the underlying value proposition. Hook = the opening 3 seconds inside that angle. Angle tests come first. Hook tests are downstream optimization. Red flag: using the words interchangeably.
5. "Walk me through writing a brief for a $30K monthly DTC account vs a $300K monthly account. What changes?" Listen for: cell count, statistical confidence required, kill speed, format mix. The smaller account moves faster on lower-confidence cuts. The bigger account can afford patience. Red flag: same brief, different budget.

Iteration — 5 questions

These probe cadence and the candidate's relationship with killing their own work.

1. "What's your kill rule? When does a concept die?" Listen for: a numeric threshold (e.g., CPA >2x target after 7 days at floor budget), an emotional rule (no defending), and a separate rule for cold vs warm. Red flag: "It depends on the situation" with no follow-through.
2. "Describe your weekly iteration loop. What do you do every Monday?" Listen for: a fixed cadence, a written report, an action list, a clear separation between daily monitoring and weekly decisions. Red flag: ad-hoc check-ins.
3. "How many concepts should be in flight at once for a $50K/month account?" Listen for: a defended range (typically 3–7 prospecting concepts), an awareness of learning phase fragmentation below floor budget, and a rotation cadence. Red flag: "as many as the team can ship."
4. "What's the worst iteration mistake you've made on a winning concept?" Listen for: a real story, the mechanism (over-iterating until the original hook was lost, scaling 100% in a single jump, etc.), and the lesson encoded into a rule. Red flag: a non-mistake ("I cared too much").
5. "How do you decide between testing a new variant of a winner versus launching a fresh angle?" Listen for: a saturation read (frequency, creative refresh cadence, CPA drift), an explicit ratio of refresh to net-new, and a budget guardrail. Red flag: gut alone.

Twenty-five questions, continued

Measurement — 5 questions

These are the most diagnostic prompts in the loop. A creative strategist who cannot read ad-level data will quietly burn $20K–$60K per quarter and explain it away.

1. "Which KPIs do you actually look at during weekly review, in priority order?" Listen for: 4–6 named KPIs, a primary that ladders to gross margin (CPA against target, break-even ROAS), and at least one diagnostic (frequency, hook rate, CTR-to-CVR ratio). Red flag: "ROAS" as the only answer.
2. "Explain how attribution windows changed your thinking after iOS 14." Listen for: a real grasp of post-iOS signal loss, the trade-off between 1-day click and 7-day click, and a mention of CAPI or modeled conversions. The candidate should be able to reference Apple's App Tracking Transparency policy without prompting. Red flag: a generic "iOS broke tracking" with no specifics.
3. "Walk me through reading a thumbnail-stop rate against hook rate against 75% video completion. What's the diagnosis if hook rate is high but completion is low?" Listen for: hook works, body of the video does not match the hook's promise, and a specific fix (re-cut the middle, shorten, change demo order). Red flag: confusing the metrics.
4. "How do you set a kill threshold that survives statistical noise on a small account?" Listen for: a minimum sample (often 50–80 conversions per cell), a confidence window, and an awareness that a CPA read on $300 of spend is not real. The 50-conversion floor traces back to Meta's own delivery documentation. Red flag: cutting at $80 spend.
5. "What's the difference between attribution and incrementality? When do you care about each?" Listen for: attribution assigns credit. Incrementality measures lift versus a holdout. Strategists default to attribution. Incrementality matters at scale and when channels overlap. The Meta Lift study documentation is the closest thing to a vendor-neutral primer, despite the source. Red flag: treating them as synonyms.

Communication — 5 questions

The last category. These probe whether the candidate can defend a brief, hand off cleanly to design, and tell a CMO bad news without flinching.

1. "How do you defend a brief when the founder pushes back at the eleventh hour?" Listen for: a prepared one-page defense (hypothesis, evidence, kill criteria), a willingness to update on new information, and an unwillingness to fold on emotional pressure. Red flag: "I'd refine to match what they want."
2. "Walk me through your handoff to design or video. What does the creative team see when you're done?" Listen for: a brief template, named references, hook variants pre-written, a definition-of-done. Red flag: "I throw a Slack with bullet points."
3. "You have to tell a founder that the launch concept is not working at week 3 of 4. How does that conversation go?" Listen for: leading with the data, naming the kill rule that triggered, presenting the next two options with cost. Red flag: hedging or burying the decision.
4. "How often do you update stakeholders, and what does the update look like?" Listen for: a fixed weekly cadence, a written async format, a separation between dashboard and narrative. Red flag: "as needed" updates.
5. "What's the most useful piece of negative feedback you've received from a founder or CMO?" Listen for: a real story, a behavior change, and a current self-aware blind spot. Red flag: humble-brags or generalities.

The structured prompt design above mirrors what Google reported in its Project Oxygen / re:Work hiring research, which found that structured behavioral prompts have roughly twice the predictive validity of unstructured interviews. The prompts are the lever. The rubric is what enforces them.

Portfolio review checklist: what to look for, what to ignore

The portfolio review still happens. It is Stage 1 plus a 10-minute spot-check during Stage 2. The goal is not to fall in love with the work. It is to extract the operating signature underneath the work. Two columns.

What to look for

• Iteration trails, not just winners. Ask to see V1 next to V7. The candidate who can show the trail has been paying attention. The candidate who only shows V7 was probably handed it.
• Kill receipts. A concept the candidate killed early, with the specific signal that triggered the kill. This is rarer than it should be.
• Brief documents, not slide decks. A real brief shows hypothesis, success criteria, and kill rules in writing. A deck shows results.
• Measurement specifics. Named KPIs, named attribution windows, named audience cohorts. "3.2x ROAS, 7-day click, prospecting cold cohort, $80K spend over 30 days" is real. "Strong ROAS" is a vibe.
• Format range. Did the candidate ship static, video, UGC, motion, and carousel? Or are 90 percent of the wins a single format that may not survive the next platform shift?
• Loss tolerance. How does the candidate talk about the work that did not work? With the same specificity as the wins, or only in vague terms?

What to ignore

• Brand logos. A candidate worked on Nike. They may have produced one ad set inside an agency of 80 people. The logo is not a signal.
• Vanity metrics divorced from cost. "10M views" without spend is meaningless. Views per dollar matter. Raw views do not.
• Awards. The Cannes shortlist correlates with budget more than with iteration cadence. Useful as a sanity check, never as a hiring signal.
• The slide design itself. Beautiful decks are correlated with self-promotion talent, which is a mixed signal at best.
• Time in seat. Three years at one shop can be the same actual experience repeated three times, or it can be three years of compounding learning. The interview prompts above tell you which.
• Total ad spend managed. A bigger budget number on a slide says nothing about the candidate's per-dollar discipline. Ask for CPA delta, not gross spend.

A 5-minute portfolio spot-check

During Stage 2, ask the candidate to pick one ad in their portfolio and answer four questions in under 5 minutes:

1. What was the original hypothesis behind this ad?
2. What did V1 look like, and what changed by the version I am seeing?
3. What was the kill threshold, and how close did the concept get to it?
4. What attribution window are you reporting against, and would the answer change under a different window?

A candidate who can answer all four in 5 minutes has been paying attention. A candidate who needs to reach for old slides is selling a story, not running an iteration loop. The point of these creative strategist interview questions is to make the difference visible in real time.

The 90-minute paid case study test

This is Stage 3. It replaces the unpaid take-home that no one finishes well and that produces work the candidate did not actually do alone. The case study is paid, time-boxed, and uses live competitor data the candidate has never seen before.

Why use adlibrary live data

The case study should ask the candidate to use /features/unified-ad-search to audit a real competitor. The deliverable is the audit. Three reasons this is the right test surface.

First, the data is live. The candidate cannot Google their way to an answer because the answer changes weekly. Second, the workflow maps to real on-the-job behavior. The first thing a competent creative strategist does in week 1 is run a competitor audit. Stage 3 just compresses week 1 into 90 minutes. Third, the deliverable is portable across candidates. You can compare three audits side by side because the inputs were identical.

The Competitor Ad Research use-case documents the production version of this workflow. You are asking the candidate to do an abbreviated version of it. Macro context on category ad spend baselines is published annually in the Statista Digital Advertising Market report, which the candidate may want to reference when sanity-checking competitor spend density.

Problem statement template

Send this to the candidate 24 hours before the case study window opens. Customize the brand, category, and goal.

You are a fractional creative strategist. {Brand} sells {product/service} into {ICP}. They are spending roughly {$X}/month on Meta and {$Y}/month on TikTok with a target CPA of ${Z} and a blended ROAS target of {N.N}x. Their three primary competitors are {A}, {B}, and {C}.

Your job: in 90 minutes, produce an angle audit of those three competitors using adlibrary's unified ad search and ad timeline analysis. Identify three angle whitespaces. Recommend three concept tests for {Brand} with hypothesis, primary KPI, kill threshold, and one-sentence creative direction. Return your work as a single document. We will pay {$300–500} for the 90 minutes regardless of outcome.

That is the entire prompt. No additional research instructions, no template handed over. The candidate's research order, tagging method, and output structure are all part of what you are scoring.

Deliverable format

Specify the deliverable shape, but not the content. A 1–2 page document containing:

The candidate may use /features/saved-ads to bookmark evidence as they work. They may use /features/ai-ad-enrichment to pull structured tags. Senior candidates may use /features/api-access to run a programmatic pull. Allow but do not require any of these.

Scoring rubric for the case study

Score on five dimensions, 1–5 each, max 25.

Hire bar on the case study: 18 / 25 minimum, with no dimension scored below 3.

Pay for the work

Pay $300–500 for the 90 minutes regardless of outcome. The cost is trivial against the wrong-hire risk. It also weeds out candidates who treat the test as a formality. The ones who care about doing it well show up disproportionately when the work is paid.

Red-flag answers per category

Patterns to listen for that should drop a candidate's score regardless of how the answer sounds. Five categories, three flags each.

Research red flags

• "I'd just look at Meta Ad Library." A senior strategist names multiple sources, including paid intelligence, customer research, and review mining. Single-source research is a junior signal.
• No tagging schema. A research method that does not produce a structured artifact (spreadsheet, database, tagged swipe file) is gut-feel research, not method.
• Speculation about competitor strategy without evidence. "I think they're doubling down on UGC" with no run-time data behind it is opinion. Strategists deal in observed behavior.

Hypothesis red flags

• No success criteria stated up front. A brief that does not commit to a primary KPI, a sample size, and a decision window is not a brief. It is a wish.
• One variable changed but multiple outcomes claimed. "We changed the hook and the offer" can never produce a clean read. Strategists who do not know this will mis-attribute every win.
• No kill rule. Concepts without kill rules drift indefinitely and consume budget. The strategist who cannot name a kill rule will not enforce one.

Iteration red flags

• "It depends" with no follow-through. "It depends" is the right opening. The wrong ending is no second sentence. Senior candidates will say it depends, then specify the conditions.
• Refusal to talk about killed work. A candidate who only discusses winners has a survivorship-bias problem and will likely under-prune in the role.
• Iterating winners forever. Refresh-only strategists run accounts into the ground because they avoid the cold start of a new angle. A 70/30 refresh-to-new ratio is healthy. A 100/0 ratio is a flag.

Measurement red flags

• ROAS as the only KPI. ROAS without break-even ROAS is decoration. It needs a margin anchor. Candidates who only cite blended ROAS have not run a margin-aware account.
• No grasp of post-iOS signal loss. A candidate who cannot describe how Meta's Conversions API documentation and modeled conversions interact with the in-platform numbers is reading the dashboard at face value. The IAB measurement guidance is a useful neutral source if you want to test cross-platform fluency.
• Attribution window confusion. Reporting numbers without specifying the window. Strategists who do not state the window are either careless or hiding noise.

Communication red flags

• Hedging on bad news. "We'd want to monitor that" is hedging. The right move is to say the concept missed the kill threshold and recommend the next two options.
• No written artifacts in the workflow. Strategists who run on Slack-and-meetings cannot scale and cannot survive a vacation. Written briefs and async updates are a load-bearing skill.
• Founder-pleasing over founder-serving. "I'd refine to match what they want" sounds collaborative. It is the answer of a strategist who will walk back briefs under pressure and produce mush.

Interview rubric: 1–5 scoring per category and the hire bar

The rubric is the document the panel fills out independently after Stage 2 and Stage 3. Independent scoring first, then calibration. Never calibrate before scoring. The calibration discussion will overwrite the dissent.

The 5-dimension rubric

Hire bar

• Total minimum: 18 / 25 across the five categories.
• No category scored below 3.
• Measurement and Iteration scored at 4 or above. These are the load-bearing dimensions, and weakness here predicts spend leakage in the first 90 days.
• Case-study score (separate 25-point rubric, see prior section) at 18 / 25 minimum.

A candidate at 22 / 25 with a 3 in Measurement is a no-hire for a senior creative strategist role. They will produce beautiful concepts you cannot defend in a CMO review. A candidate at 19 / 25 with 4s in Measurement and Iteration is a hire even if Communication is the weakest score, because communication is more coachable than measurement literacy.

Calibration mechanics

Three rules for the panel debrief.

Each panelist submits scores in writing before the meeting. The hiring manager reads scores aloud one category at a time without naming the panelist. Disagreements above 1.5 points trigger a 5-minute defense from each panelist. Then the rubric average decides, not the loudest voice in the room.

This calibration mechanic is the single highest-yield change you can make to a hiring loop. Most loops collapse to whoever talks fastest in the debrief. The independent-then-calibrate sequence preserves dissent long enough to matter.

For deeper context on what the role looks like in production, see the Creative Strategist Workflow breakdown of a competent week and the Competitor Ad Research workflow that the case study replicates in compressed form. The Ad Timeline Analysis feature is what a strategist uses on day one to read run-time as a proxy for performance. Hire someone who already thinks this way.