Mixpanel Overview

Overview of Mixpanel

• Mixpanel is a product analytics tool favored by product teams in software companies or those with digital products like websites, web apps, or mobile apps.
• It helps to understand user behavior within a product, including feature adoption, retention, and onboarding effectiveness.
• It can identify where users get stuck and improve overall engagement.
• Although primarily for apps, it is adaptable to websites but may be less effective than Google Analytics for simple use cases.
• It serves marketing teams interested in understanding the impact of marketing efforts on metrics beyond page views, such as sign-ups, onboarding completion, and retention.
• It is an instrumental tool for product-led growth, enabling product and marketing teams to understand trends, campaigns, A/B tests, and feature releases, informing decision-making processes.
• Mixpanel reports are designed to be user-friendly for non-technical individuals. The point-and-click interface simplifies data analysis, allowing product and marketing teams to independently build reports, funnels, dashboards, and insights.
• Engineers implement data, and product or marketing teams can use the tool to build their own reports.
• Self-sufficiency and the ability to build reports without waiting for technical assistance are major advantages of using Mixpanel.
• Mixpanel has been a well-established product since 2008, and is a leading company in the product analytics space, along with Amplitude, Pendle, and Heap.
• The product offers essential features and design tweaks that facilitate report and dashboard creation for non-technical users.
• The product's design focuses on creating an enjoyable user experience with little friction once users understand how to use the tool and build reports.

Mixpanel Pricing

• Mixpanel's pricing focuses on events, distinguishing it from tools that charge by monthly tracked users (MTUs).
• Event-driven models like Mixpanel's are beneficial for consumer products with many users and low value per user.
• B2B (business-to-business) products with fewer users, but higher value per user, often benefit from a monthly track user model.
• The initial Mixpanel plan offers 20 million events, offering a startup a starting point to get comfortable with the tool.
• Paid plans are available that can accommodate different event volumes.
• Data export to data warehouses requires an add-on.
• Group analytics, useful for B2B companies to group individual users under organizations or companies, is another add-on.
• The way to determine the appropriate event volume is by setting up Mixpanel, letting it run for a month, and then seeing what the event volume is.
• It's possible to reduce the number of events tracked to stay within a specific price tier.
• Enterprise-level plans include single sign-on (SSO) and data governance features.

Free vs Growth Plan

• Spark AI is a Mixpanel add-on that uses natural language models (LLMs) to analyze data, potentially speeding up the familiarization process for new users.
• Formulas and saved metrics are limited in the free plan.
• Creating custom formulas can help with marketing attribution by grouping them into categories like social, paid ads, or organic.
• The number of cohorts are also limited in the free plan.
• Cohorts can categorize users based on various criteria such as new users, paid users, or canceled users who are still active.
• Machine learning features like impact analysis reports are unavailable in the free plan.
• Dashboard functionality is also limited in the free plan; one can save only 5-10 reports.
• One can start with the free plan, and then run into the limits, and then upgrade.

Tracking Plans

• A tracking plan should be developed before writing any code, and can be created using tools like Google Sheets, Google Docs, or specialized software. The goal is to document every possible action a user can take within the product.
• Tracking plans should include notes on every possible action users can take within the product.
• An ideal tracking plan will note when an action can be fired, media to help developers and images for developers to work with.
• A tracking plan should have 20-40 events.
• Each event includes "event properties" providing more information about the action.
• Event properties give lots of detail about the event itself
• Event properties allow the condensation of events into more global events.
• A classic example includes that of an "order completed" event. This one event can provide information about checkout ID, order ID, tax information, shipping information, and products that were ordered.
• "User traits" and "user properties" can also be included in a tracking plan.

User Properties

• User properties contain specific information about individual users.
• Name, email, location (city, region, country), and custom attributes are common examples of user properties.
• Custom attributes can include data points such as number of games played, subscription status, or subscription plan.
• Mixpanel is built around user profiles, which sets it apart from tools like Google Analytics that primarily focus on aggregated data.

Tracking Plan Importance

• A tracking plan helps maintain consistency in event names and properties.
• Having two events that seem similar suggests they should be consolidated into a single event.
• Before writing code, it's essential to refine the tracking plan by condensing, editing, adding properties, and ensuring overall cleanliness.
• Prioritizing events allows for a phased implementation, starting with high-priority events.

Naming Conventions

• Consistency in naming conventions is more important than the specific convention used.
• Lowercase, uppercase, or camel case can be used as long as the same convention is applied across all events, event properties, and user properties.
• Whether to use present or past tense in event names is a matter of choice, but consistency is key.
• Event names, properties, and user properties should be obvious and easily understood.
• Clear labeling improves data adoption by making it easy for users to understand the data.
• Abstract or unclear names should be avoided in favor of obvious and intuitive labels.

Tracking Plan Benefits

• It's easier to make changes to the tracking plan before involving engineers and writing code.
• Changes to the tracking plan are common
• Clean, well-thought-out data is essential to avoid polluting Mixpanel data with incorrect event names.

Implementation Options

• The choice of SDK and whether to use client-side or backend implementation are important considerations.
• Available SDKs include JavaScript for web, PHP, Ruby, Java for backend, React Native, and native mobile SDKs for iOS and Android.
• There is also a basic HTTP API available.

Client vs. Backend Tracking

• Backend tracking is generally recommended as the primary approach due to ad blockers and data restrictions on browsers and iOS.
• 80%-90% of tracking should ideally be done on the backend for reliability.
• Client-side tracking may be necessary for certain events that can only be tracked on the client.
• Teams with fewer backend engineers may start with client-side tracking and gradually transition to backend tracking.

Debugging Data

• The goal of implementation is to create a single profile containing all the data that's needed
• A successful implementation results in a single user profile with all activity, regardless of the device on which the activity occurs.
• User profiles contain events with properties, as well as user properties.
• Events should be tracked and have the right values.
• User properties that are passed on manually should be stored and have the right value
• Data from web, iOS, and Android devices should be consolidated into a single profile.
• Multiple profiles for a single user indicate a problem with the user ID or other settings.
• The Users report allows you to find information on users
• The Events report shows events coming in in real time.
• Mixpanel processes data in nearly real time, which is useful for debugging.

Mixpanel Reports

• The main area for report generation is the Reports section.
• Key reports: Insights, Funnels, Flows, and Retention.
• Insights is the most important report, where building blocks are used to construct reports.
• Events are used as building blocks and can be combined to create reports.
• Reports can be customized by changing date ranges, intervals, and chart types.
• Mixpanel is valuable for its ability to break down and segment data.

Event Properties Breakdown

• Event properties tracked include default browser, city, and custom properties like payment method.
• The total number of orders/events over the last three months can be broken down by payment method (credit card, PayPal, Apple Pay, Afterpay).
• Data can be displayed in various chart formats like pie charts or bar graphs.
• Segments can be combined; a segment on a segment can be created to see the breakdown of shipping method for every user who uses a credit card.
• Filters can narrow down data: example: only showing users who chose standard shipping.
• Breakdowns can include user properties like country to see the data segmented by payment method, shipping method, and country.

Saving and Updating Reports

• Reports can be saved with a title and added to dashboards (boards) for future reference.
• Data in Mixpanel updates in real-time, usually within the last few minutes.
• Data can be manually refreshed if needed.
• Comparisons can be made to previous periods (e.g., previous three months, previous day) or segments.
• Reports are easy to build and filter, allowing for granular understanding of product usage.

Implementation Prerequisites

• A tracking plan and correct implementation are crucial for easy report building in Mixpanel.
• Issues arise when the events are not logical, obvious, or correct, leading to tracking errors.

Funnel Analysis

• Funnels can be quickly built and destroyed to analyze various flows like checkout or onboarding.
• Conversion rates at each step are visible (e.g., product added to cart -> checkout started -> purchase completed).
• Conversion windows can be adjusted from seconds/minutes to days to analyze user actions within specific timeframes.

Funnel Breakdown

• Breakdowns can segment the funnel by properties like payment method or country.
• Payment method values were unavailable when segmenting on step one, because the payment method is captured on step three.
• Filters are available to narrow down the funnel to specific segments like users from the US or specific products.
• Funnels are useful for companies accustomed to high-level numbers that want to segment data by product lines, countries, or marketing sources.
• A-B tests can be analyzed by setting up specific events for each variation and tracking conversion rates through the funnel.

Conversion Metrics

• Conversion rate can be measured by total conversions, not just unique users.
• Windows as broad as months mean conversions can be tracked long-term (e.g. purchases by long-term subscribers over 6-12 months).

Retention Analysis

• Retention reports measure product stickiness and user return rate.
• Cohort analysis shows how many users who signed up make a purchase within a specific time frame (e.g., one day, two days, three days).
• Retention is helpful for mobile games, consumer apps, B2B apps, and subscription data to understand retention rates at day 7, 30, 90, 180, etc.

Retention Implementation

• Data is needed over a period to establish data, 30, 60, 90 days or more
• Over time can look at how the data changes
• Filters do breakdowns as analysed previously

Data Export

• Charts can be exported as CSV or PNG files.

Dashboards (Boards)

• Dashboards allow for combining multiple reports and adding text for context.
• Dashboards enables high-level overviews can be reviewed, and the user can dive deeper as needed.
• Text provides context to reports and dashboards.
• Layout and chart styles can be customized.
• Filters can be applied to entire dashboards (e.g., filter by specific payment method).
• Dashboards can be shared via email and have a TV mode for display on TVs.

Data Integrations

• Data can be brought in and out: export data to data warehouses like BigQuery or Amazon S3.
• Data can also be exported to integrations like Facebook Ads for remarketing campaigns.
• Data warehouses with existing product data can be connected to Mixpanel for visualization.

Additional Features

• Cohorts: Groups of users with shared characteristics
• Lexicon: Organizes data and add descriptions to event names.
• Custom Events: Combine events together eg purchase event and subscription started event.
• Lookup Tables: Allows for lookup based on values which provide an output on the right
• Formulas: Formulas for marketing channels the organic
• Behaviors: Specific filters that can be saved and reused.