What Uplift is

Who it is for

Data analysts, product analysts, marketing analysts, data scientists, experimentation specialists and product managers preparing for interviews.

What it covers

Each branch is a subject area with its own subtopics and question set:

• SQL & Data Modeling: Window functions, joins, funnels and event models under interview time pressure.
• Experimentation & A/B Testing: Design, power, guardrails, peeking and readouts for online experiments.
• Product Analytics & Metrics: Metric trees, retention shapes, activation and diagnosing a drop.
• Statistics & Inference: Distributions, intervals, regression intuition and error tradeoffs.
• Analytics Case & Communication: Structuring ambiguous cases and writing readouts a stakeholder trusts.
• Product Management: Product sense, prioritization and the metrics PM loops probe.
• Digital & Marketing Analytics: Incrementality, campaign economics, Adobe Analytics, GA4, experimentation tools and site measurement that survive a finance review.
• AI Agents & Automation: Evaluating agents and LLM features with product-grade metrics.

How questions are graded

Grading is deterministic. Every question shows its evaluation criteria before you answer: what a strong answer has to do, without revealing the answer. After you submit you get detailed grading feedback: multiple choice is compared to the answer key, numeric answers are accepted within a stated tolerance, SQL is checked against required query patterns, and written cases are checked criterion by criterion. The full model write-up appears only when you ask for it. Only scored attempts and submitted cases move your progress, so the progress tree reflects work you actually completed.

How questions are sourced

Uplift combines publicly reported interview questions, company-context scenarios and original questions based on common interview patterns. Reported questions link to their public source whenever one is available. Company-context scenarios use real product or marketplace constraints but are not represented as verbatim interview questions. Original questions are labeled clearly and carry no company name. Every numeric answer and SQL fixture is reviewed for internal consistency before publication.

How the leaderboard stays honest

Leaderboard scopes stay hidden until at least 30 qualifying participants exist. Suppression happens on the server, so a thin sample is never presented as a ranking. See the current leaderboard.

Common questions

What is Uplift?

Uplift is an interview-practice platform for analytics, experimentation, product and AI roles. You answer real interview-style questions in SQL, multiple choice, numeric and written case formats. Each submission is graded automatically against an answer key.

How does Uplift work?

Pick a branch in the catalog, open a question, and answer it in the format the interview would use. SQL questions run against sample tables in your browser so you can inspect the rows your query returns. Choice and numeric questions are checked instantly. Written case questions are scored against a rubric. Attempts are unlimited, hints unlock one at a time, and the model answer stays hidden until you ask to see it.

Who is Uplift for?

Data analysts, product analysts, marketing analysts, data scientists, experimentation specialists and product managers preparing for interviews.

What problem does it solve?

Most interview practice is either pure SQL drills or unstructured case advice. Analytics, product and marketing interviews test judgment: diagnosing a funnel, designing an incrementality test, choosing a north-star metric, ranking a roadmap, or proving an AI agent works. Uplift lets you practice those decisions with feedback on each attempt.

How much does Uplift cost?

The free plan covers all free practice questions, attempt history, the progress tree and the leaderboard, with no credit card required. Pro is $19 per month and unlocks every premium question, full rubrics and model write-ups. Founder is $39 per month and adds early access to new branches and question drops, the ability to request questions for specific companies and topics, a locked-in price and a Founder badge on the leaderboard.

How are answers graded?

Grading is deterministic, not an opinion. Before you answer, a question shows its evaluation criteria: what a good answer has to do without giving the answer away. After you submit, you get detailed grading feedback. Multiple choice is compared to the answer key, numeric answers are accepted within a stated tolerance, SQL is checked against required query patterns, and written cases are checked against each criterion. The full model write-up stays hidden until you ask for it. Scored attempts and submitted cases are what move your progress.

Why is the leaderboard sometimes empty?

Leaderboard tables stay hidden until at least 30 people qualify for that scope. Ranking a handful of attempts would be noise, so small samples are suppressed on the server rather than shown with a disclaimer.

Is Uplift affiliated with the companies named in questions?

No. Uplift is independent and is not a certification. Each question carries one of three labels. "Reported at [Company]" means the question was publicly reported by a candidate and links to that public source. "Company-context scenario: [Company]" means it uses that company's real product or marketplace constraints but is not represented as a verbatim interview question. "Original Uplift question" means we wrote it around a common interview pattern and it carries no company name.