A protocol claims 100,000 new users, but you might not know—more than half could be "passersby" wallets created last night and used only once, while many others are returning users who came back after a break. Equating "new wallets" directly with "new users" is one of the most common mistakes in on-chain analysis. To tell if a protocol is genuinely growing, you must separate one-time visitors, returning users, and true first-time users.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
Here are 3 steps to distinguish these three user types and avoid being misled by data.
Step 1: Use "entity-adjusted" metrics to filter out fake addresses
Many so-called "new users" are actually multiple wallets controlled by the same person. Filter out this noise first.
What to do: Look at "entity" data provided by tools like Dune Analytics or Glassnode, not raw "address" counts.
How to do it:
Raw address counting treats 10 wallets controlled by one user as 10 "users." Entity adjustment uses clustering algorithms to merge multiple wallets belonging to the same entity, giving a more realistic user estimate.
For a quick trend check, visit a Dune Analytics dashboard and search for "New Addresses," which shows the number of unique addresses appearing on-chain for the first time in a given period. But you'll need to filter this indicator by subsequent activity to make it meaningful.
When you've completed this step: You can find a protocol's "New Entities" data on Dune or Glassnode, not just "New Addresses."
Common pitfall: Habitually treating "new addresses" as "new users." A single user farming an airdrop can create hundreds or thousands of addresses, all counted as new addresses but representing the same person.
Step 2: Separate three user types—new users, returning users, and one-time users
Even after filtering wallets, if an address appeared 30 days ago and returns today, it shouldn't be counted as "new growth." In Web3, user lifecycles typically break down into new users, retained users, returning users, and dormant users.
What to do: Calculate the protocol's "new user rate" and "returning user rate" to see the true composition of growth.
How to do it:
New Users: Addresses that first interacted with the protocol during the measurement period (e.g., the past 7 days). This signals the protocol is attracting genuinely new people.
Returning Users / Resurrected Users: Addresses that had no interaction for more than 30 days but reappeared in the measurement period. This tells you retention is problematic; returning old users do not represent new incremental growth.
One-time Users: Addresses that interacted only once and never came back. A high proportion of one-time users means the product lacks stickiness and appeal.
Data shows that platforms with low activity have 10 times fewer new and returning users compared to high-activity platforms, yet the number of one-time users differs by only about 1x.
When you've completed this step: You have broken down the protocol's new addresses into three buckets: "first-time users," "returning users," and "one-time passersby."
Risk flag: If a protocol's total address count is rising but over 70% of each wave of new addresses are "appear-once-and-disappear" types, growth may be driven by airdrop farming or one-off speculation, not real user adoption.
Step 3: Track user lifecycles with Dune queries
Dune lets you distinguish "New" and "Active" addresses using SQL queries, so you can assess growth quality.
What to do: Write or fork a query in Dune to compute user lifecycle metrics.
How to do it:
Fork an existing Dune dashboard and inspect its user segmentation logic. Typically you need to group the
transactionstable bywallet_address, find each address'smin(block_time)as its "first appearance time," and then determine whether the address appeared within specific time windows later on.Identify new addresses: If an address's first interaction time falls within the analysis period, label it "New."
Identify returning addresses: If an address's first interaction time is earlier than the analysis period but it still had interactions during the period, label it "Returning."
When you've completed this step: You've generated a simple user cohort table that shows the ratio of new users to returning users for the protocol each day.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
How to confirm you've done it right?
After filtering and segmenting, answer these three questions:
After entity filtering, how much of the raw "new address" data remains? If it shrinks by more than 50%, the protocol has a serious problem with one person controlling many addresses.
What share of "new" activity comes from returning users? If over 40% of "new activity" is old users returning, the protocol is relying on reactivating its existing base rather than attracting fresh users.
Is the proportion of one-time users consistently rising? If yes, the protocol's user retention is in trouble.
If your answers are: entity-adjusted data is still growing, the new user rate consistently outpaces returning users, and the one-time user share is stable or declining—this signals relatively healthy growth. New users are truly increasing, not just airdrop hunters or old users churning.
