Marketing take-home assignment examples that don't ask for free strategy

The classic bad marketing take-home is "Tell us how you'd grow our brand over the next 12 months." It's vague enough that a good answer takes a week, specific enough that you could act on it the next day, and usually unpaid. Senior candidates have learned to spot it. Many decline, and the ones who don't are often the ones with the fewest options.

The fix is to test a skill, not to ask for a plan. Give the candidate made-up or anonymized data, ask for one specific decision or deliverable, and pay them for the time.

Why test at all?

Marketing resumes are full of results without context. "Grew organic traffic 300%" from what starting point, with what budget, and how much of it was their doing? Interviews don't settle it either, because talking about a funnel is easier than diagnosing one.

A short task shows whether someone can look at numbers and say why they moved, not just that they moved.

Examples by specialty

Paid media (2 to 3 hours). Send a CSV of 30 days of search-campaign data: keywords, spend, clicks, conversions and revenue, with a few problems planted in it. "Find the three biggest sources of wasted spend and the two best opportunities to scale. Write two ad variations for the group you'd scale. Keep it to one page." Strong answers tie each decision to a number and say what they'd test next.

SEO (2 hours). Use a page that has nothing to do with your company, or a deliberately broken staging page. "Audit this page. List the five fixes that matter most, in priority order, and rewrite the title tag, meta description and H1 for the query 'payroll software for small business.'" Strong answers put indexability and search intent ahead of keyword density.

Email and lifecycle (2 hours). "Here's a fictional SaaS product with a 14-day trial and its activation data. Design a three-email onboarding sequence: the trigger for each email, the subject line, the one action it asks for, and the metric you'd use to judge it."

Social (1.5 to 2 hours). "Here's a launch brief for a fictional product. Draft a one-week posting plan for LinkedIn and one short-video platform, and fully write two of the posts, including the hook and a description of the visual."

Marketing analytics (2 to 3 hours). "Here's an export of 5,000 customer orders with acquisition channel and first-order date. Calculate CAC and 90-day revenue by channel using the spend table provided. Tell us where you'd move budget and what you'd want to confirm first."

Where the line is between a test and free consulting

Ask yourself one question about your brief: could you act on what comes back? If a candidate's answer could go straight into next quarter's plan, you're asking for consulting and should pay consulting rates, or change the brief.

Tests that stay on the right side of the line:

• use sample, historical or anonymized data instead of access to your live accounts
• ask about a fictional company, or a scenario that's clearly not your current one
• have a scope narrow enough to finish in the time you said

How long, and what to pay

Two to three hours for specialists and up to four for managers. Ask for more judgment, not more pages. If you want to see a senior marketer's strategic thinking, run a short paid task first and talk through a strategy question live afterward.

Pay the time you're asking for at a rate in line with the role, whether or not you hire them, and without making them invoice you. A few hundred dollars for a three-hour task is in a fair range for experienced marketers. For why this pays off for the employer, see You're already paying for candidate screening.

Tools: don't make candidates pay

Don't ask candidates to use their own Semrush, Ahrefs or HubSpot subscription. Provide exports, or stick to free tools like Google Sheets, Google Trends, the Meta Ad Library and PageSpeed Insights. If the job really depends on one tool, give the candidate a temporary login.

How to grade it

Decide on the weights before anything comes in. A reasonable split for most roles:

• Reading the data (about 40%). Did they spot what you planted? Is the math right?
• The decision (about 30%). Is the recommendation realistic for the budget and constraints given?
• Communication (about 20%). Could a busy executive act on page one?
• Execution (about 10%). Copy quality, formula hygiene and clean charts.

A habit worth rewarding: candidates who state their assumptions ("I'm assuming a 60% gross margin" or "I'm assuming the conversion tag fires correctly"). In real marketing work you rarely have complete data, so this matters.

Follow up with a walkthrough

The quickest way to tell real skill from a well-prompted chatbot is to have the candidate share their screen, open their own spreadsheet, and walk you through how they got from the data to the decision. You'll find out within ten minutes whether they built the pivot table or pasted it. Pudding records that walkthrough as part of the assessment. Here's how it works.

Related reading

• Work sample automation
• Writing take-home examples
• Paid case studies for ops, PM and strategy roles
• Our tracker of real non-engineering take-homes

Josh Priollaud

Founder @ Pudding

Josh Priollaud is the founder and CEO of Pudding, a Siligo Ventures portfolio company. Josh has worked in business development roles at Cloud9 IDE, Webgility, MuleSoft and Adobe. Before his Silicon Valley career, Josh served as an Economic Development Adviser in the Peace Corps. Josh holds an MBA from the Leavey School of Business at Santa Clara University and a BS in Business Administration from Portland State University. Josh has been a full stack developer (Ruby on Rails, React) since 2020.

LinkedInsiligo.tech