Interview Glossary
Having trouble with some concepts? Here's a list of common vocabulary used by product managers, organized by the question types you'll face in interviews: product strategy, product sense, analytical/execution, behavioral/leadership and drive, and technical.
Product strategy
Strategy questions test long-term product thinking: competitive dynamics, go-to-market approaches, market entry, and pricing. These terms cover how PMs frame markets, segments, and launches.
Beta Test: A trial phase of a product launch scoped to a select group of early users, typically before full public release.
Breakeven: The point at which total revenue equals total costs. In product management, breakeven analysis helps determine when a product or feature will become profitable.
Customer Segment: A group of users divided by shared characteristics so a company can design and market to each group more effectively.
Early Adopters: Users who try a product as soon as it becomes available, often before bugs are fully resolved. Commonly used as beta testers.
Go-to-Market Strategy: The plan for launching a product to users, including decisions about pricing, distribution, positioning, and rollout sequence.
Low-Hanging Fruit: A product or feature change that is easy to implement but delivers meaningful metric improvement.
OKR (Objectives and Key Results): A goal-setting framework where an Objective is a qualitative goal, and Key Results are the specific, measurable milestones that define success. PMs use OKRs to align team work with company priorities.
Product-Market Fit (PMF): The degree to which a product satisfies real, strong market demand, evidenced by users actively seeking it out, retaining, and referring others.
Product Roadmap: A high-level view of a product's direction and priorities over time, showing what will be built and in what sequence.
Product sense
Product sense questions test how you identify user needs, scope solutions, and communicate product thinking. These terms come up when framing users, problems, and what a solution actually looks like.
MVP: Minimum viable product. The simplest version of a product that delivers core value and can be tested with real users to gather feedback.
Pain Point: A problem that users are experiencing. Identifying a specific, underserved pain point is a core step in product sense interviews.
UI (User Interface): The visual layer through which users interact with a product, including buttons, layouts, screens, and controls.
Use Case: A scenario describing how a specific user would interact with a product to accomplish a goal.
UX (User Experience): The full experience a user has with a product, encompassing UI design, onboarding, reliability, and any friction encountered along the way.
Value Proposition: The answer to what specific problem does this product solve, and why is it the right solution for this user.
Vibe Coding: Using AI tools to rapidly build working software prototypes through conversational prompting, with little to no manual coding. Meta introduced a vibe coding component into its AI Product Sense round, where candidates build a prototype live using a Llama interface.
Analytical/execution
These questions test whether you can define the right metrics, root-cause a problem with structured reasoning, and navigate ambiguity when data is messy. Metric fluency is table stakes here.
A/B Test: A method of comparing two versions of a product experience to determine which performs better, typically by splitting users 50/50 between variant A and variant B.
Bounce Rate: The percentage of visitors to a website who leave after viewing only one page.
Churn Rate: The percentage of users who stop using a product over a given time period. The inverse of retention.
Cohort Analysis: A method of tracking the behavior of a specific group of users over time, typically grouped by when they signed up. Useful for understanding whether retention is improving across different user groups.
Conversion Rate: The percentage of users who complete a desired action, such as signing up or making a purchase.
CTR: Stands for clickthrough rate. The percentage of users who click on a specific link or element after seeing it.
DAU/MAU: Stands for daily active users and monthly active users. A high DAU/MAU ratio indicates users return frequently, not just occasionally.
Funnel: The series of steps a user takes toward a desired action, such as signing up or making a purchase. Users drop off at each step, creating a narrowing shape that PMs analyze to find where to improve.
Impression: An instance of an ad or piece of content being seen on a user's screen.
KPI: Stands for Key Performance Indicator. A quantifiable metric that tracks progress toward a specific business objective.
Lagging Indicators: Outcome-based metrics (like revenue or retention) that reflect past performance rather than predicting future results.
Leading Indicators: Predictive metrics (like sign-ups or feature adoption) that provide early signals about whether a product is headed in the right direction.
North Star Metric: The single metric that best captures the core value a product delivers to users, and the top-level signal a PM would defend as the best proxy for long-term product health.
Retention: The percentage of users who continue using a product over a given time period. Often tracked by cohort to understand whether engagement is improving over time.
ROI: Stands for return on investment. Used to evaluate whether the expected value of a product initiative justifies its cost.
Behavioral/leadership and drive
These questions assess how you've handled real situations: conflict, ambiguity, failure, and leadership. Interviewers are scoring the scope of what you owned and how you tell the story, not just what happened.
Bottom Line Up Front (BLUF): Opening a story with one or two sentences that state the project and its impact before any context-setting. Signals confidence and gives the interviewer a reason to lean in, instead of burying the outcome at the end.
Downleveling: Being offered a role at a lower seniority level than the one you interviewed for. Common in employer-driven markets, and structurally hard to reverse once it happens, since not every team has open headcount at both levels.
STAR (Situation, Task, Action, Result): A structured framework for answering behavioral interview questions. Still the enforced standard at Amazon; replaced by narrative-driven storytelling at Meta, Netflix, and OpenAI.
Stakeholder: Any person or group with an interest in or influence over a product, including engineering, design, legal, finance, marketing, and leadership. PMs drive alignment across stakeholders without direct authority over them.
Technical
Technical questions now mean understanding AI at a working level, not as an engineer, but as an informed product owner. This is the largest category, covering both classic engineering process terms and the AI vocabulary interviewers increasingly expect.
Agentic AI/AI Agents: AI systems that can take sequences of actions and complete multi-step tasks autonomously, rather than just responding to a single prompt. Common in product sense prompts and take-home assignments at top companies today.
API: Stands for Application Programming Interface. A product's API allows other products or developers to access its functionality (for example, using Facebook's API to authenticate users in another app).
Code Complete (vs. Ready to Launch): Code complete means development is finished but testing is still needed. Ready to launch means the product has passed all QA and is prepared for release.
Context Window: The maximum amount of text an LLM can process in a single interaction, including both the input and output. Relevant for designing AI features that involve long documents or multi-turn conversations.
Dogfood: To use a product internally before releasing it publicly, in order to catch issues and gather early feedback.
Evals (Model Evaluation): The process of measuring how well an AI model performs on a task, typically using metrics like accuracy, precision, or recall. PMs use evals to define what "good" looks like for an AI feature and track whether model changes help or hurt.
Feature Flag: A technique that lets engineers turn a feature on or off without deploying new code, enabling staged rollouts, A/B tests, and quick rollbacks.
Fine-tuning: The process of taking a pre-trained foundation model and training it further on a smaller, task-specific dataset to improve its performance for a particular use case.
Foundation Model: A large AI model trained on broad, general-purpose data that can be adapted for many tasks, such as GPT-4, Claude, or Gemini. Most product teams build on top of foundation models rather than training from scratch.
Hallucination: When an AI model generates output that is factually incorrect or fabricated, but presents it confidently as true. A core risk category in AI product design that PMs are expected to know how to mitigate.
Inference/Inference Compute: The process of running a trained AI model to generate a real-time response. Inference compute refers to the processing cost of doing this at scale, which affects pricing, latency, and product architecture decisions.
LLM (Large Language Model): An AI model trained on large amounts of text data that can generate, summarize, and reason about language. ChatGPT, Claude, and Gemini are examples, and LLMs are the foundation of most AI-native product features at top companies today.
Mobile Web/Native: Native refers to apps downloaded to a device. Mobile web refers to experiences accessed through a mobile browser. The Facebook app is native; Facebook.com in a mobile browser is mobile web.
Precision and Recall: Two metrics for evaluating AI classification models. Precision measures how often the model is right when it flags something. Recall measures how often it catches what it's supposed to catch. Increasing one often reduces the other, and the right balance depends on the use case.
Prompt Engineering: The practice of designing and refining instructions given to an AI model to improve the quality or format of its outputs. Useful both as a prep tool and as a product design consideration.
QA (Quality Assurance): The process of testing a product before release to identify bugs and confirm it works as intended.
RAG (Retrieval-Augmented Generation): A technique that combines an LLM with a retrieval system, where relevant documents are fetched from a knowledge base and passed to the model along with the user's query. This allows the model to answer based on current or proprietary information, reducing hallucinations.
Refactor: A restructuring of existing code that improves its quality or maintainability without changing what it does externally.
Sprint: A fixed development cycle, typically one to four weeks, during which a team plans, builds, and delivers a defined set of work.
Synthetic Data: Artificially generated data used to train or test AI models when real data is scarce, sensitive, or biased.
Tech Requirements (or Acceptance Criteria): The specific conditions a feature must meet to be considered complete, defined before development begins.
Technical Approach: The high-level strategy for solving a technical problem, including choices about technologies, frameworks, and architecture.
Technical Debt: The accumulated cost of shortcuts taken during development that slow future work and increase maintenance burden over time.
Token/Token Optimization: A token is the basic unit of text that an LLM processes, roughly a word or word fragment, and also the unit of cost in most AI APIs. Token optimization means designing prompts and outputs to use fewer tokens without sacrificing quality.