Overview of the basic functions: Tracking, User Profiles, Events, User Properties, Opt-In, Custom Reports, Metrics and Funnels
This article explains what MarvelMetrics is and how it works. It explains what kind of data you can import or track and which reports you can create. The document is an overview - you find details on each topic in other articles.
MarvelMetrix is an analytics platform: It gives you the tools to create reports from data collected from different sources. The platform itself is data agnostic — it does not care where the data comes from nor what data you send.
You can adapt it precisely to your needs:
The basic idea of MarvelMetrics is to collect data about the users of your web page, app or game to build user profiles. These profiles consist of events — actions a user performed — and user properties.
MarvelMetrics calculates the analytics reports based on these user profiles. They tell you how well a specific group of users converts, how much time and money they spend on your web page, which products they download, and what they put into their cart.
MarvelMetrix always calculates the reports on the full set of data. It does not interpolate or sample data. You can create reports retroactively from all data that is available in the tracking project.
The calculations are swift - it usually takes less than 1 second to get the reports.
MarvelMetrix collects data in user profiles. A user profile consists of two types of data:
You collect these events and properties from your web page, app or game using an SDK that we provide. The SDK sends the data to our tracking server where it’s collected and processed.
We also offer an API that you can use in case we can’t provide an SDK for your specific needs.
Data that you send to us is processed and collected on your behalf within the scope of the GDPR. Please note that you either need the consent of the users (opt-in) or collect the data based on legitimate interest.
Please remember Art. 9 GDPR:
Processing of personal data revealing racial or ethnic origin, political opinions, religious or philosophical beliefs, or trade union membership, and the processing of genetic data, biometric data for the purpose of uniquely identifying a natural person, data concerning health or data concerning a natural person’s sex life or sexual orientation shall be prohibited.
MarvelMetrix also does not permit you to track any security-related or sensitive data such as credit card data, bank accounts, social security number, identity card numbers, passwords.
MarvelMetrix assigns a unique ID to each tracked user. The user ID is the link that ties the events and properties together into a user profile.
The ID can be anonymous (a random number/letter combination) — or an ID that you provide from another source like your customer database (e.g. customer number, email address).
User properties are values that describe a user, but there is no temporal interest in the values.
Examples for user properties are:
All these values are possible user properties you can track with MarvelMetrix.
Of course, not all of them are relevant for your specific setup. E.g. if you sell software, the operating system might be relevant to you. If you have an online store that sells shoes, the shoe size is what matters.
It is tempting to collect all the data you can get, but please restrict the collection to the minimum of information that is relevant to your business.
It’s up to you which values you want to track to create a valuable user profile that helps you understand your customers.
You use these values to segment the users in the reports to answer questions like:
Events — in contrast to user properties — happen at a specific point in time. They can even occur multiple times - e.g. when the user adds multiple items to the shopping cart or downloads several products.
Examples for events are:
You can add properties to events to describe them in detail:
Use the events in your reports to answer questions like:
The MarvelMetrix SDKs use parallel server-side and client-side opt-in / opt-out. The client side controls the sending of the data, the server side controls the recording and processing.
With the client-side opt-in there is no transfer of tracking data to the MarvelMetix server until the user has agreed to the processing of the data. In case of a revocation of the consent no further data will be sent to the server in the future. The consent or rejection is stored in the form of a cookie in the user’s browser. Opt-Out cookies do not contain any identifiable user data.
In the case of tracking with a legitimate interest, an opt-out cookie will be set if the user objects.
MarvelMetrix stores events and properties only for known users — the server discards all other tracking data. Deleting a user from the database not only erases his data but also disables future data collection. There’s no data left in the database for an opted-out user - not even the information that the user does not want to be tracked.
To track users, you have 2 options:
This server-side opt-in/opt-out has the advantage that tracking is globally disabled for a specific user ID — even if the user logs in from multiple devices using the same account. The user can also withdraw his consent from email tracking (tracking pixels).
A user requires 2 pieces of information to withdraw his consent:
MarvelMetrix provides an easy way to opt-out for web projects using a snippet that you can add to the privacy section on your web page.
If you track the user from an app or game, provide an easy to find button in the user interface where the user can withdraw his consent.
MarvelMetrix, by default, creates unique IDs for each user per web page. Imagine that you have 2 domains: One is your marketing page, and the other is the checkout page which on a separate domain. Cross-site tracking ensures that the user shares the same ID on both domains — adding events to the same user profile.
Tracking users across devices (e.g. tablet and mobile phone or desktop) is only possible if there is a common way to identify the user. E.g. the user is using the same account for your service or if he uses a personalized link that contains a user ID.
MarvelMetrix allows you to combine multiple tracking profiles into one. This process is called aliasing. Start tracking a user independently from multiple sources — e.g. email, web page and app. Create an alias if you find a way to tie the profiles together later — e.g. when the user logs into his account.
You can currently create 2 types of reports in MarvelMetrix. We’ll add more types in the future such as A/B testing and cohort analysis.
MarvelMetrix calculates the reports always on the full data. No sampling — no interpolation. You can apply filters and segments at any time, and you can even create reports retroactively. The reports are displayed instantly (the average response time is less than 1 second).
Metrics show you the change of a value over time.
This metric counts all occurrences of a particular event. Events can happen multiple times per user.
Examples:
You can filter the events by event properties — e.g. to only see page views on a specific page.
This metric is similar to the number of events metric but in contrast, only counts each user once.
Examples:
You can apply event filters — e.g. to only count users who have visited a specific landing page.
This metric calculates the average number of events a user performed.
Examples:
You can, of course, apply a filter — e.g. see the number of times a user plays a specific song.
You can calculate this metric on events with number properties.
Examples:
Use this metric to calculate the total sum of a property value.
Example:
This metric calculates the average sum of an event property value per user. The user must not necessarily have performed the event.
Example:
This metric calculates the average sum of an event property value per user who has performed the specific event.
Example:
The conversion rate is the relative number of users who have performed 2 events in sequence.
Examples:
This metric is similar to the conversion rate but calculates the average time between two events.
Examples:
Funnel reports show you how users propagate through a series of events. They are similar to a conversion rate but with multiple steps.
For example an online store might have the following sequence of events that lead to a purchase:
Each of these steps takes the user closer to the goal of completing the purchase - but not all webpage visitors go through all steps. Some leave the funnel earlier.
The funnel chart shows you where users drop off and where you can make the most significant improvements in the purchase process.
The same applies if you use funnels to optimize a game. The players complete levels 1-6 without problems, and there’s a significant drop-off between level 7 to 8? Level 7 is what you should check.
Funnels also help you to analyze issues that occur suddenly — seeing a drop-off in your revenue? Create funnels retroactively from the events in your data set, including filtering and segmentation, to find the root of the problem.
Create filters that only show a distinct group of users or exclude specific users from the reports. You can build filters from any user property with value comparisons:
User property filters can be combined using and and or logic.
Examples:
Use segmentation to build groups of users and compare their values for a metric or funnel. You can build segments from any event or user property.
Examples: