Amplitude
A product analytics service for examining how people use a product, built around events and the users who produce them.
Amplitude is a product analytics service. Applications send events describing what users do, and the service provides analysis over them: how many people complete a sequence of steps, whether groups return over time, and how behaviour differs between segments. It is aimed at understanding product usage rather than website traffic alone.
Product analytics against website analytics
The distinction matters, because the tools look similar and answer different questions.
Website analytics is oriented towards traffic. How many visitors, from where, viewing which pages. The unit of interest is the visit.
Product analytics is oriented towards people over time. Did this person come back? How far did they get through the sequence? Do users on one plan behave differently from another?
Amplitude is built for the second kind of question, and that shapes what it records and what it can analyse.
Events tied to users
Events carry who performed them.
That connection is what enables everything else. Without it, you can count how often something happened. With it, you can ask whether the same people are doing it repeatedly, whether people who did one thing later did another, and whether a group is still active a month later.
Sequence analysis
Products have paths people are meant to follow. Register, verify, configure, complete a first action.
Knowing that only a small proportion finish is not useful on its own. Knowing which step loses them is.
Sequence analysis shows the count reaching each step. A large drop between two adjacent steps is a specific problem in a specific place, which is something a team can actually fix.
Retention
This is the analysis that most distinguishes product analytics, and it answers a question totals conceal.
A product gaining users every month looks healthy. If the same product is losing nearly as many, the total hides the truth: people arrive, try it and leave.
Retention analysis groups people by when they started and follows each group over subsequent periods. That separates arrival from continuation, and it is frequently the most important measurement a product team has.
Segmentation
Averages across all users hide differences.
Splitting analysis by user or event properties shows where behaviour varies: by plan, by region, by how somebody arrived, by which features they use.
The pattern that matters is often visible only after splitting, because two groups behaving very differently average into something that resembles neither.
The work that comes first
None of this is available without instrumentation.
Events have to be defined and sent from the application, carrying the properties later analysis will need. A property nobody thought to include cannot be analysed retrospectively, because the data was never collected.
This is the substantive work in adopting product analytics, and it is where most of the effort goes. Deciding what to record is worth doing deliberately rather than instrumenting everything and hoping.
Who uses it
Amplitude is used by product managers, product analysts and engineering teams building software products, particularly where understanding continued use matters more than counting visits.
Points to consider
Data about user behaviour is sent to the service, and what may be sent is a question to settle before implementation.
Analysis quality depends entirely on instrumentation quality. Poorly defined events produce analysis that looks authoritative and is not.
Pricing is generally related to volume, and the official site is the reference for current terms, which are worth checking against realistic event volumes.
Getting started
The documentation covers planning which events to send, installing the collection libraries, and using each analysis type. Instrumenting one important sequence and examining where people stop is the usual first exercise, and it typically produces a finding immediately.
Key features of Amplitude
Capabilities described in the official documentation.
Events tied to users
Each event records who performed it, so behaviour is analysed per person rather than per session alone.
Step sequence analysis
A defined sequence of steps shows how many people reach each one and where they stop.
Return behaviour over time
Groups defined by when they started are followed to see how many continue using the product.
Segmentation by property
Analysis can be split by user or event properties so differences between groups are visible.
Advantages of Amplitude
Factual advantages that follow from the features above.
Where people stop becomes visible
Analysing a sequence step by step shows which step loses people rather than only that the total is low.
Retention is measurable
Following groups over time answers whether people keep using a product, which a single total cannot show.
Questions can be asked without engineering
Once events are collected, new analysis is configured in the interface rather than requiring new work.
Differences between groups surface
Splitting by properties shows where behaviour varies, which averages across all users conceal.
Common use cases for Amplitude
Situations the official documentation describes this tool as being used for.
Finding where sign up fails
A step sequence shows how many people reach each stage of registration and where they abandon it.
Measuring whether users return
Groups are followed over subsequent weeks to see what proportion continue using the product.
Comparing behaviour between segments
Analysis split by plan or region shows where usage patterns differ between groups.
Assessing a released feature
Events for a new feature show adoption and whether use continues after the first attempt.
Official website
Everything on this page is based on the official documentation for Amplitude. You can read the source here.
Amplitude official documentation