How Employers Assess AI Skills in Marketing Job Candidates

Across the 28 AI-required jobs whose custom application fields were captured, 5 ask about AI explicitly (17.9%); 0 ask for potentially verifiable evidence (0.0%). These are documented questions, not confirmed interview assessments.

A role may appear in more than one row if it asks separate questions in different categories. The AI-question and verifiable-evidence totals count distinct roles. With no captured forms, shares are withheld rather than displayed as 0%.

Stated demand is not the same as tested competence. An employer may require workflow-building skills without describing an assessment. A question asking “Have you used ChatGPT?” is different from a work sample or a demonstrated business result. Our categories preserve that distinction instead of assigning employers a subjective AI hiring score.

For the employer-side requirement, see what AI-required marketers are expected to build. For the wider candidate-facing process, see marketing application screening questions. Browse AI-relevant marketing vacancies.

This report is generated from active marketing manager-level job listings collected by Marketing Manager Jobs from public employer job feeds and APIs. Listings are filtered to include manager-level marketing roles such as growth marketing manager, product marketing manager, SEO manager, lifecycle marketing manager, demand generation manager, brand manager, marketing director, head of marketing, VP marketing, and CMO.

We exclude non-marketing or non-manager roles such as engineering, product management, sales, support, design, writing-only, analyst, and administrative roles.

The primary cohort comprises 289 currently active, English-language, semantically classified jobs where the AI-required score is at least 0.85. We captured custom application fields for 28 of these roles, containing 219 fields in total. Only that captured-form subset is the denominator for percentages in the assessment table. The forms currently come from supported Greenhouse and Ashby feeds, rather than every employer or ATS.

We classify the wording of each publicly visible custom screening question using reproducible, conservative phrase rules. Questions must explicitly reference AI, LLMs or a named AI tool. Each role is counted once per primary evidence type, and a question is assigned only its strongest matching category. A past project is not considered a live practical interview test. The export contains counts, not private candidate responses.

Limitations: Missing published questions are unknown, not evidence that the employer never evaluates AI. Job descriptions, private interviews, hiring-manager opinions, invitation-only take-home tasks and informal portfolio reviews are not fully observable here. A certificate request does not establish the certificate is credible; a request for a demonstration does not prove it takes place. These are descriptive indicators within a selected ATS sample, not population-wide hiring rates.

The dataset updates as new jobs are ingested. Historical trend data uses daily snapshots, aggregated by calendar month.