Anyone Can Rent a Number. Keeping It Is the Whole Job

Every yield product launches with a screenshot. A big total-value-locked figure, a green line, a pre-deposit campaign that filled up in hours. The founder posts it, the community cheers, and the number becomes the story for about a week.

Then the emissions taper, and the number walks out the door.

We have watched this cycle enough times to stop being surprised by it. When a protocol we work with asks us to "grow TVL," the first thing we do is pull apart two questions that usually get collapsed into one: how do you get capital in, and how do you make it stay. The first is mostly a spending problem. Anyone with a large enough incentive budget can rent liquidity for a launch. The second is a product and marketing problem, and it happens to be the only one that compounds.

Rented liquidity always leaves

You do not have to theorize about this. Berachain is the cleanest recent case. Pre-deposit campaigns pulled in $3.1 billion before mainnet even went live, TVL peaked around $3.35 billion in February 2025, and within a year it had fallen roughly 88% as the incentives that attracted that capital dried up. The chain was explicitly designed to make liquidity "sticky" by tying it to governance. Capital left anyway the moment yields dropped.

This is not a 2025 discovery. Go back to the first liquidity-mining wave: when Uniswap wound down its UNI rewards program, its liquidity fell about 40% in a single day. Same mechanism, five years apart. There is a name for this capital now, mercenary capital, and it is not meant as an insult. It is a rational response to an offer, and when the only reason to be somewhere is the payout, the exit is more or less priced in from the start.

A launch high tells you what you were willing to spend. Whether any of it stays is a separate question, and it is the only one worth reporting.

The metric that survives the incentive

Here is the split we actually care about. Real traction holds a meaningful share of its peak after the incentive tapers. One practitioner benchmark draws the line at 40% or more of peak TVL retained after taper for organic growth, versus a fall below 20% within two weeks for the artificial kind. Those two curves look identical on launch day and nothing alike ninety days later, which is inconvenient, because ninety days later is where the real business lives.

That same analysis notes that even under normal conditions, only about 25% of first-time DeFi users become regular users. During an incentive campaign the figure gets worse and, more dangerously, harder to see, because the reward distorts why anyone showed up in the first place. You cannot read intent off a number you paid to manufacture.

This is why the sharper teams we sit across from have started asking for something different. Instead of one more trading competition to spike volume for a week, they want sticky TVL and holders who are still there next quarter. It is a harder brief to deliver against, and it is the one worth taking.

How we build for the number that stays

Give capital a reason to stay that is not the yield. Incentives get the first deposit. Utility gets the second month. For a yield-bearing asset that means integrations where the token is genuinely useful, as collateral, in payments, or as a building block inside someone else's product, so that leaving carries a real cost beyond forgoing a reward. We treat those integrations as launch infrastructure, not as a phase-two nice-to-have.

Reward the actions that correlate with staying. Paying people to park capital rewards parking. Paying people to do the things that predict retention, providing durable liquidity, using the asset in-protocol, referring users who also stick around, buys behavior you actually want to keep. The emission costs you the same either way; the difference is what it leaves behind once it stops.

Separate the cohort you rented from the cohort you earned. From day one we tag inflows by source and track each cohort's decay curve on its own. Incentive-sourced capital and organically-sourced capital behave nothing alike, and blending them into a single TVL line hides the one signal that predicts what happens after the campaign.

Put retention in front of the founder, not the peak. A launch-day high is a vanity screenshot. Week-four and week-twelve holder retention, net of incentive outflows, is what we report, because it is the figure an investor should be underwriting and the one that actually forecasts the next raise.

If your growth chart only looks good on the days the campaign is live, you did not buy growth. You leased a graph.

The part founders skip

None of this means incentives are bad or that launch campaigns do not work. They work, and our team runs them: pre-deposit pushes, points programs, competitions. These are legitimate tools. The error is treating the number they produce as the finish line rather than the opening move.

The current wave makes the stakes higher. Yield-bearing stablecoins alone grew roughly 300% year over year heading into 2026, which means an enormous amount of capital is about to go shopping for the best available return, and it will keep shopping indefinitely. If your entire go-to-market is a bigger offer than the last protocol's, you are one epoch away from being someone else's outflow.

So before you post the launch screenshot, answer the question that decides whether there is a business underneath it: on the morning the rewards stop, who has a reason to still be here. Get that answer built first, and the number tends to hold on its own.