The median ClawPump founder owns 0.01% of their own token
We went looking for why so few builders stick around. The answer was in our own defaults: 7,215 launches, a median founder allocation of 0.004 SOL, and a hero graduation that bought thirty cents of itself before doing $35.7M in volume.
The question
Our thesis says the first creator-fee check flips a builder from tinkering to building for real. How often does that check actually land?
Every launch in claim_service.tokens from 2 Feb to 27 Jul 2026 (n=7,215), joined to fee_collections for realised builder earnings. Fee rates are measured, not quoted: across the 220 tokens with >$10k lifetime volume ($88.4M combined), creator fees ran 0.277% of volume, of which the builder's 65% share is 0.187%. SOL valued at $75.28.
The result
“0.0040”
— $SQUIRE's dev buy, in SOL. Our first graduation bought thirty cents of its own token, then did $35.7M in volume.
We were measuring the wrong graduation
ClawPump's north star is graduations per quarter — agents that become real companies. The database has an `is_graduated` column, so for a long time that felt tracked. It isn't. `is_graduated` means a token filled its pump.fun bonding curve and migrated to a pool. Forty tokens have done that. Three have become companies.
Once we separated the two, the funnel got much less flattering. The curve metric is genuinely good — 0.55% against pump.fun's 0.198–0.26% is a two-to-three-times edge, and it is the strongest single claim we have. The company metric is three, and nothing in the product tracks it or shows it.
The cause was in our constants, not the market
The obvious story is that most tokens fail everywhere, so ours do too. That is true and not the interesting part. The interesting part is that our builders finish the launch flow owning none of the thing they just launched.
Gasless launch performs no dev buy — correctly, because that is what lets an agent tokenise without onramping gas, and it is the whole cold-start story. But any positive buy forces the self-funded path, so a builder faces a cliff: launch free owning 0%, or fund a wallet before they know whether the thing is real. Almost everyone takes the free path. And then there is no route back — nothing in the product has ever let them buy into their own token later.
So the fix is not to force capital up front. It is to add the path that was missing: launch free, wait for signal, and when you decide it is real, take a disclosed and vested position. Commitment after evidence, which is how it actually happened for the builders who stuck.
What a real allocation would have been worth
$SQUIRE is the cleanest illustration because it worked. A 7% allocation — the standard Virtuals makes default — would have cost roughly 2.1 SOL, about $158 at today's price, and would be worth around $83,000 at $SQUIRE's current $1.19M market cap.
The second finding surprised us more. Creator fees are themselves a sizeable position. At our realised rate of 0.187% of volume to the builder, and a median lifetime volume of about 25× market cap, the fee stream is worth roughly a 4.7% allocation — earned in SOL as people trade, without ever selling a token. A builder taking 7% and holding is looking at close to 12% of market cap in total economics, half of it arriving as cash.
That is a much better thing to put on a launch screen than a blank field labelled "Dev buy".
Why we are not recommending 25%
The instinct when you see a 0.01% median is to swing hard the other way. We tested that against how traders actually read supply concentration, and it does not hold. Analysts treat a single cluster under 15% as healthy, 15–30% as elevated rug risk, and above 30% as high risk.
A 25% founder position reads as an insider position and suppresses exactly the trader participation the builder needs. The version that gets you the capital and the credibility is 5–10%, held openly, with a vesting contract anyone can verify. Disclosure and a cliff are what separate a founder from an exit — and no memecoin launchpad can comfortably copy that, because for most of them it would be an indictment.
The first check lands for 0.4% of builders. Not because the market rejected them — because we never gave them anything to own.
We spent a long time treating momentum as luck. It isn't luck if there was no capital and no ownership to create it in the first place. Two thirds of our launches take no allocation, the median founder holds 0.01% of supply, and the product has never once offered to fix that after the fact.
So we are building the allocation planner: share of supply rather than a raw SOL figure, what it is worth at target market caps, a recommended 5–10% band with a warning above 15%, and a Streamflow vesting contract by default. Plus the path that matters more — letting a builder who launched gasless take a real position later, once there is signal.
One hard constraint we found while scoping it: this cannot ship without anti-sniper protection. Telling a builder to buy 7% of supply while snipers can front-run the launch makes the advice actively harmful. We currently ship a sniper skill and no defence, which is a strange thing to notice about yourself.
We will publish the follow-up whether or not it works. The number to watch is median share of supply per launch cohort, and today's baseline is 0.01%.
What we’re doing about it
Post-launch allocation, not a bigger dev buy
Gasless-at-zero stays exactly as it is — it solves cold start and we are not breaking it. What was missing is the route back: buy into your own token after launch, disclosed and vested. Commitment arrives after the evidence, which is how it happened for every builder who actually stuck.
Disclose the allocation on the token page
An undisclosed 7% is indistinguishable from a rug; a disclosed and vested 7% is a founder with skin in the game. Publishing allocation, cliff and unlock progress turns the scariest part of the mechanism into the most trustworthy one.
Anti-sniper in the same release, or not at all
A founder allocation recommendation is only safe if the launch is protected. 60 seconds buy-side by default, matching what the rest of the category already ships.
A 25% recommended allocation
Our first instinct, and wrong. Above 15% in a single cluster reads as elevated rug risk to the people whose buying the builder depends on. 5–10% held openly and vested beats 25% held quietly.
Scoring launch ideas with a model
We considered rating ideas before launch. It would be noise and builders would correctly ignore it. Ticker collisions and outcome base rates from our own 7,215 launches are facts, and facts are the only thing worth showing at that moment.
Limits & what we got wrong
- Fee rates are approximate: historical SOL collections are valued at today's price of $75.28, and SOL moved a lot across the window. The order of magnitude survives — $SQUIRE's implied $66.7k against a largest-actual-payout of $64,957 is a close check — but do not read the third decimal.
- The 25× volume-to-market-cap multiple comes from only 11 tokens with enough volume and market cap to compute it. It is directional, which is why the planner exposes it as an adjustable input rather than a constant.
- Bucketing outcomes by allocation size shows a steep gradient, but the top buckets contain 4 to 7 tokens each. We are not claiming causation from that, and the relationship is confounded — builders willing to spend 1+ SOL are more committed before they touch the field. The planner default should be run as a real experiment against the current blank field, not justified by this.
- Half of our headline volume is two tokens, and one of them is $CLAW, our own. Any platform-wide rate computed here inherits that concentration.