ARPU, ARPDAU, and ARPPU are average-revenue ratios. They compress a period of revenue and a count of people into one number. The abbreviations are easy to swap, and the swap changes the story, because each one divides by a different group.
They are descriptive. They do not say which campaign caused the revenue.
How do you calculate each one?
Write the numerator and the denominator before you publish the ratio.
- ARPU (average revenue per user). Revenue in the period divided by users in the period. "Users" might mean installs, accounts, or monthly active users. Pick one and keep it.
- ARPDAU (average revenue per daily active user). That day's revenue divided by that day's daily active users. The activity rule for DAU (an app open, a session, a level start) has to be the same one used everywhere else you quote DAU.
- ARPPU (average revenue per paying user). Revenue divided by users who paid in the period. People who spent nothing are not in the denominator. ARPPU is therefore higher than ARPU whenever some users do not pay, which is the usual case.
The revenue sum has to name its parts: validated store purchases, subscription renewals, impression-level ad revenue, refunds. A purchase counted before receipt validation can include transactions the store did not charge. Ad revenue counted from both ESTIMATED and PRECISE rows needs a label. See impression-level ad revenue.
A concrete example
Use field names, not a target.
For one calendar day in one currency:
- Sum
purchase.valuewherevalue_typeisobservedand the store state is purchased. Subtractrefund.valuewith the sametransaction_idset. - Optionally add
ad_impression.valuewhereprecisionisPRECISEor where you have explicitly chosen to includeESTIMATED. - Count distinct
user_idwith asession_startthat day. That count is DAU for this report. - Count distinct
user_idwith apurchasethat day. That count is paying users.
ARPDAU is the revenue sum divided by the DAU count. ARPPU is the purchase sum (usually without ad revenue, unless you document otherwise) divided by the paying-user count. ARPU for the month uses the month's revenue and the month's user definition, which is not the sum of daily DAU.
If purchase.value is still in micros, or ad revenue is in micros while purchases are in currency units, the ratio is wrong by orders of magnitude. Convert units before you divide.
No sample dollar figure belongs in the definition. Two products with the same ARPU can have completely different paying rates.
What are the limits?
Averages hide the skew. A few high spenders raise ARPU and ARPPU while most users contribute nothing. Pair the ratio with the share of users who paid, or with a median, when the decision depends on how typical the average is.
Time zones and late events move DAU and revenue across dates. A refund recorded next month lowers this month only if your rule says so. Say which rule you use.
Currency mix needs a conversion or a split by currency. An unweighted sum of USD and EUR values is not a revenue total.
These ratios do not attribute credit to ads. Cost per install and predicted value answer different questions. See cost per install and predicted lifetime value.
How Hardal handles it
Hardal does not define a single ARPU for you. It stores the revenue and user events you send, including purchases, refunds, and ad impressions, so a report can divide a documented numerator by a documented denominator. Data Hub is the workspace for building that dataset from the events you already collect.
If the events are missing or duplicated, the ratio will be too. Validate store purchases first, deduplicate transaction_id, and keep estimated ad revenue out of a settled-revenue ARPU unless the label says it is included. Hardal for Gaming is the mobile context where these ratios usually sit next to install cost.
Related terms: Session, Events, Predicted lifetime value, Cost per install
Frequently asked questions
What is the difference between ARPU, ARPDAU, and ARPPU?
ARPU divides revenue by users. ARPDAU divides one day's revenue by daily active users. ARPPU divides revenue by paying users only. The same revenue total produces three different ratios because the denominator changes.
Do these metrics include ad revenue?
Only if you say so. Some reports use in-app purchases alone. Others add impression-level ad revenue. A comparison is valid only when both sides list the same components and the same refund rule.
Can I add daily ARPDAU to get a monthly number?
No. ARPDAU is a daily ratio. Adding thirty of them does not produce monthly ARPU, because the user counts overlap and the denominators are not the same people.
Does a higher ARPPU mean the product got better?
Not by itself. ARPPU rises when paying users spend more, and it also rises when low spenders drop out of the paying count. Read it next to the number of paying users.