Event tracking basics: event vs user properties

Maya

The developers sent me last week's raw export: 147 rows I can't read. What's actually in there?

You

Every row starts with a time, a user id and a word like app_opened. So each row is one thing a person did?

Alex Tutor

Close. Each row is an event: a record that something happened. It answers who (user_id), what (event) and when (event_time, in UTC). The other columns are details.

Alex Tutor

Dataset · 13 rows

Thirteen rows from Halves's event log, 1–4 June

A few rows picked from last week's export, in time order. Times are UTC.

Download CSV

Columns
event_time: When it happened, in UTC
user_id: Who: the Halves account
event: What happened
platform: Where the event was logged: ios, android or server
entry_method: expense_added only: manual, receipt_scan or recurring
amount_usd: Amount in US dollars, for expenses and settle-ups
settlement_id: settle_up_recorded only: which settlement the row belongs to

Data

event_time	user_id	event	platform	entry_method	amount_usd
2026-06-01T06:00:00Z	u001	expense_added	server	recurring	420
2026-06-01T06:41:00Z	u008	app_opened	ios
2026-06-01T06:44:00Z	u007	app_opened	ios
2026-06-01T06:46:00Z	u007	expense_added	ios	receipt_scan	116
2026-06-01T16:36:00Z	u020	app_opened	android

Show 8 more rows

2026-06-01T16:38:00Z	u020	expense_added	android	manual	35
2026-06-04T12:22:00Z	u013	app_opened	ios
2026-06-04T12:27:00Z	u013	settle_up_recorded	ios		185	s004
2026-06-04T18:10:00Z	u020	app_opened	android
2026-06-04T18:12:00Z	u020	expense_added	android	manual	91
2026-06-04T18:15:00Z	u020	settle_up_recorded	android		67	s008
2026-06-04T21:21:00Z	u014	app_opened	ios
2026-06-04T21:22:00Z	u014	settle_up_recorded	ios		185	s004

Row 4: two minutes after opening the app, u007 added a $116 expense by scanning a receipt.

You

And entry_method and amount_usd are the details?

Alex Tutor

Yes, event properties. They describe that one moment. The same person's next expense can have a different amount and method.

Alex Tutor

Question 1

Read row 10 of the excerpt: 2026-06-04T18:12:00Z, u020, expense_added, android, manual, 91. What does it record?

Choose one answeru020 opened the Android app at 18:12 UTC on 4 Juneu020 typed in a $91 expense by hand in the Android app at 18:12 UTC on 4 JuneThe server added a $91 recurring bill to u020's accountu020 paid someone in the group $91

You

The users file has columns too: signup_date, acquisition_channel, timezone. Event properties as well?

Alex Tutor

No, user properties. They describe the person, not a moment, so they're the same for all of that person's events.

Alex Tutor

Dataset · 6 rows

The same six people in the users file

One row per person. These values don't change from event to event.

Download CSV

Columns
user_id: The Halves account
signup_date: Date the account was created
platform: The user's phone platform: ios or android
acquisition_channel: How the user came to Halves
timezone: The user's home time zone
is_internal: true for a team account, false for a customer

Data

user_id	signup_date	platform	acquisition_channel	timezone	is_internal
u001	2026-04-23	ios	invite	Europe/Kyiv	false
u007	2026-05-08	ios	referral_link	Asia/Bangkok	false
u008	2026-05-08	ios	app_store_search	Asia/Bangkok	false
u013	2026-05-11	ios	paid_social	Europe/Lisbon	false
u014	2026-05-30	ios	referral_link	Europe/Kyiv	false

Show 1 more row

u020

2026-05-06

android

referral_link

Europe/Kyiv

false

Quick test: could the value differ between two events a minute apart? An expense amount can. The channel someone signed up from can't.

Theo

I want to know if people from paid social add more expenses than invited people. Where does that answer live?

Alex Tutor

In both files. Take expense_added events, look up each user_id in the users file and group by acquisition_channel. Event properties say what happened; user properties split it by who.

Alex Tutor

Question 2

Which of these columns are user properties rather than event properties?

You

Could Halves Plus be an event too? Something like user_is_premium?

Alex Tutor

Being on Plus is a state, so it's a user property. Starting a trial happens at a moment, so it's an event.

Name events as object plus past-tense verb: expense_added, group_created, one name on every platform. If Android logs addExpense instead, every report shows two halves of one number.

You

And screens? Should “balances screen shown” be an event?

Alex Tutor

Only if a question needs it. Log the actions a decision depends on, not every tap: each event has to stay correct through every release.

Alex Tutor

Question 3

Which of these should Halves log as events?

Choose all that applyA member accepts an invite to a groupThe user's time zoneA paywall is shown after someone taps receipt scanningThe phone model the user hasThe user's current balance in a group is $40Someone creates a group for a trip

You

Row 1 looks odd. u001 added a $420 expense at 06:00 UTC, and the platform is server.

Alex Tutor

See entry_method = recurring. Plus subscribers can make a bill repeat, like rent, and the server adds it on schedule. u001 may have been asleep.

That's a server-side event: the backend logs it, not the app. It's never lost when a phone is offline, but it can happen with nobody touching Halves.

App events are the opposite: real actions, but a phone with no signal may send them late.

Maya

So if I count “people who added an expense”, u001 is in there?

Alex Tutor

If you count rows, yes. We'll fix that next lesson. For now, ask of every row: who or what caused it?

Alex Tutor

Question 4

True or false: a row with event = expense_added, platform = server and entry_method = recurring shows that the user opened Halves and added an expense.

TrueFalse

You

Also, s004 shows up twice, from u013 and u014, both for $185. Did they pay each other?

Alex Tutor

No, it's one payment. The payer records a settle-up, and when the other person confirms, their phone logs it again. Same settlement_id, two rows.

The log is right. But “how many settle-ups?” is answered by the ids, not the rows.

You

So to read a log I need to know how the product works, not just the column names.

Alex Tutor

Exactly. For each event, note who triggers it, where it's logged and which properties it carries. That list starts a tracking plan, which you'll write at the end of this module.

Alex Tutor

Question 5

The excerpt has three settle_up_recorded rows. How many settle-ups (payments) do they describe?

Choose one answer132Can't tell without the users file

Maya

Next month receipt scanning will split the bill between people automatically. I'll want to know how often dinners get split this way. What do we log?

You

expense_added with entry_method = receipt_scan, plus something about the split?

Alex Tutor

Good start. Keep the existing event rather than inventing dinner_scanned, and add properties for Maya's next questions: how many people, split how, in what kind of group.

For each detail, decide: does it describe this moment or the person? The person's details already live in the users file.

Alex Tutor

Question 6Short answer · AI-checked task

Someone in a trip group scans a $120 dinner receipt in Halves and splits it equally between 4 people. Name the event that should record this and list its properties. For each property, say whether it's an event property or a user property, and why.

Write your answer and get a score with feedback from our AI reviewer.

Log in to get AI feedback

Alex Tutor

Next time we take all 147 rows and count how many people really used Halves last week.

Alex Tutor

That’s the lesson. You answered every task — nicely done.

Next lesson: Totals, unique users and counting traps