Quick answer
User properties are attributes that describe the person behind the events, such as plan, country or signup date. They are stored once on the user profile and carry across all of that person's events, unlike event properties, which describe a single action as it happened.
What are user properties?
User properties are attributes that describe the person behind the events, or in B2B the account they belong to: plan, country, signup date, company size. They are stored once on the user's profile and travel with every event that person sends. That is what separates them from event properties, which describe one action as it happened, such as the price of one purchase or the path of one page view.
Tools use the words a little differently. Google Analytics 4 defines user properties as attributes that describe groups of your user base, such as language or location, and uses them to build audiences. Amplitude defines them as attributes of the person who triggers an event that carry across all of that person's events. This page uses the profile-level meaning, which is the common one in product analytics. Some enterprise software uses the same two words for configuration fields; that is a different topic.
A quick test: would this value be the same on the person's next event? If yes, it is probably a user property. If it changes with every action, it is an event property.
User properties vs event properties
| User property | Event property | |
|---|---|---|
| Describes | The person (or account) | One action |
| Question it answers | Who is this? | What happened this time? |
| Stored on | The profile, one row per person | Every event row |
| Value over time | One current value; a new write replaces the old one | Frozen at the moment the event happened |
| Example | plan = pro, country = DE, signup_source = blog | item_sku = A-17, price = 40, page_path = /pricing |
| Best for | Segments, cohorts, filtering people | Breaking down actions, sums and averages per event |
| Needs a known user | Yes: anonymous visitors have no profile | No: anonymous events carry them too |
The last row matters in practice. If you only identify people at login, everything an anonymous visitor does before that carries event properties but no user properties.
Which layer does an attribute belong to?
| Attribute | Layer | Why |
|---|---|---|
| Plan (free, pro) | User | Changes rarely and you segment people by it |
| Signup date | User, written once | Never changes after it is set |
| Company size | User or account | Describes who they are, not what they did |
| Order total | Event | A different value on every purchase |
| Items in cart | Event | Describes this moment only |
| Campaign of the first visit | User, written once | A first-touch fact about the person |
| Campaign of this visit | Event | A different campaign can bring the same person back |
| Lifetime orders | User, incremented | A running count that belongs to the person |
When you need both views, store both: the current plan on the user, and a separate plan_at_purchase on the purchase event.
Worked example: the same data in the wrong layer
Sample data: 500 users and 1,900 report exports in one month. 100 users were on Pro all month, 100 upgraded to Pro on day 15, and 300 stayed on Free.
| Group | Users | Exports while on Free | Exports while on Pro |
|---|---|---|---|
| On Pro all month | 100 | 0 | 600 (6 each) |
| Upgraded on day 15 | 100 | 200 (2 each) | 800 (8 each) |
| On Free all month | 300 | 300 (1 each) | 0 |
Question: what share of exports came from Pro?
- Plan recorded on each event (state at the time): (600 + 800) ÷ 1,900 = 73.7%.
- Plan read from the user profile (current state, so the 100 upgraders count as Pro): (600 + 200 + 800) ÷ 1,900 = 84.2%.
The gap is 10.5 points. It comes from the 200 exports the upgraders made while still on Free, which the profile view credits to Pro. Neither number is wrong; they answer different questions. The profile answers "what do today's Pro users do". The event-level value answers "what did Pro usage look like at the time", which is what you need to ask what made people upgrade.
The reverse mistake is just as common. Sample data: one user places three orders worth 40, 25 and 95. If order_total is written to the profile, each order replaces the last and the profile ends at 95. Summing the event property gives 160, so the profile view understates spend by 65 ÷ 160 = 40.6%.
How user properties get written
| Write mode | What it does | Typical use |
|---|---|---|
| Set (overwrite) | The latest value replaces the old one | plan, display name, company_size |
| Set once | Writes only if the field is empty | signup_source, first campaign |
| Increment | Adds to a number | lifetime_orders, credits_used |
| Append | Adds an item to a list | features_tried |
| Unset | Removes the field | A value you no longer collect |
Choosing the mode is choosing what the field means. Written with "set" instead of "set once", a first-touch field quietly becomes a last-touch field.
Common mistakes with user properties
- Putting per-action values on the profile. Cart value, last page and current search term are overwritten on every event and describe nothing stable.
- Storing contact details as ordinary properties. An email or phone number is a direct identifier. Keep it in the identity fields and let masking apply, instead of copying it into free-form keys.
- Inconsistent keys and types.
plan,Planandtierare three fields, and the number 10 and the string "10" sum and filter differently. Agree on one name and type per attribute in your tracking plan. - Letting values go stale. If a downgrade never writes back to the profile, every segment built on
planis wrong from that day. - Expecting anonymous visitors to have a profile. No known user, no user properties. Identify first; see identify.
- Turning every attribute into a dimension. Free-text values with thousands of distinct entries make breakdowns unreadable; see custom dimensions for what is worth promoting.
User properties in TapCub
Take the Pro-share example. Here is how the same setup looks in TapCub.
- Identify the people. User properties sit on an identified person, so identified mode has to be on (this depends on the setting in the compliance center). Call
tapcub.login()when someone signs in and pass a display name. Visitors who never log in stay in aggregates and have no profile. Steps: identify users. - Write plan from your server. Fields only your backend knows, such as plan or membership, go through the server profile calls (set, set once, increment, append, unset). In the browser,
setPropsoverwrites a field andsetPropsOncefills it only if empty. To load fields from a CRM, use the CSV user-properties import under All-platform analytics › Data › Data import, mapping at least a user id, email or phone. - Check type and privacy. Open All-platform analytics › Data › Property dictionary. The tabs are User, Event and Account. Click Scan now, confirm each key's type, and check its sensitivity class; direct identifiers are masked in lists and exports by default.
- Use the values. In All-platform analytics › Users › Segments & tags, rule segments can filter on properties, and a tag can turn something like "top category in 30 days" into one reusable field. Open a person in All-platform analytics › Users › People to see their current values.
- Send "at the time" values on the event. For the 73.7% question, send the plan as a property of the export event itself; it appears under the Event tab of the same dictionary.
Sources
- Amplitude Docs, User properties and event properties.
- Google Analytics Help, User properties.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Data Engineer
Writes about event design, data quality, and collection workflows, with verifiable implementation methods for reliable analytics.