Markets price what they can already see coming.
Keyrock studied more than 16,000 unlock events across 40 major tokens and found that 90 percent generated negative price pressure, with the impact typically beginning around 30 days before the unlock actually happened. Not on the day. Thirty days early, driven purely by anticipation.
That behavior tells you something uncomfortable about launch day. It is a settlement date, not a starting line. By the time your token trades, most of what determines the outcome has already been decided and is visible to anyone paying attention.
The evidence backs this up. A separate Keyrock analysis of 62 airdrops across six chains found that 88 percent of airdropped tokens declined in price, with the steepest drops concentrated in the first 15 days. Around six in ten tokens sit under their listing price a month later. Research covering two million airdrop wallets put same-day selling at roughly 64 percent of recipients.
Those failures did not begin at launch. They were built into the launch months earlier.
What the Numbers Say About the Environment
Across 18.67 million Pump.fun deployments between January 2024 and June 2026, CoinGecko counted 68.67 percent that never traded past their launch date, and just 4.55 percent that were still active at the 90 day mark. Against that, the total market was worth about 2.19 trillion dollars in August 2026, roughly half the 4.27 trillion peak reached the previous October.
Thinner capital, denser competition, and buyers whose default expectation is that this one fails too. A launch showing up with no prior record has no way to counter that. Nobody is obliged to give your project the benefit of the doubt, and fewer people do each cycle.
Lead Time Is the Real Constraint
The reason pre-TGE work decides outcomes is not philosophical. It is mechanical. The assets that matter take longer to build than a launch campaign lasts.
Search presence takes months, not weeks
New websites generally need three to six months to reach page one for reasonable terms, and competitive spaces can take twelve to eighteen. Roughly 95 percent of newly published pages never reach the top 10 within their first year. If you start publishing at TGE, you are invisible in search during the exact window when people are searching your name most.
Press relationships cannot be bought quickly
Coverage in outlets that carry weight comes from relationships built over months. Pitching CoinDesk or The Block a week before launch produces nothing, and the alternative, paid placement on sites nobody reads, is visible to anyone doing diligence.
Audits and legal opinions have queues
Reputable audit firms book out, and remediation after findings takes additional time. Legal opinions classifying your token take weeks. Both are prerequisites for serious listings, and both fail if started late.
Community depth is time, not money
You can buy members in a week. You cannot buy the kind of community that answers questions for you during a drawdown. That takes months of consistent presence, and it is immediately distinguishable from the purchased version.
The compounding assets are the ones with lag
This is the pattern. Everything that genuinely helps you at launch has a build time measured in months. Everything you can buy in the final week is rented and disappears when payment stops.
What Compounds Versus What Is Rented
Splitting your pre-launch plan along this line clarifies almost every budget decision.
Compounding assets
Documentation, search presence, original research, audit history, named counterparties, community relationships, and shipping track record. These accumulate, survive the launch, and keep working afterward.
Rented attention
Paid influencer posts, sponsored placements, quest campaign participation, and incentive-driven liquidity. All useful, none permanent, and all of it stops the moment the budget does.
The correct ratio shifts over time
Early in the 90-day window, spend almost entirely on compounding assets. Rented attention has no value if there is nothing credible for it to point at. In the final two weeks, rented attention amplifies what you built. Reversed, which is the common pattern, you pay to direct scrutiny at an empty site.
Rented attention is not optional either
This is not an argument against paid promotion. A project with excellent documentation that nobody sees fails differently but still fails. The point is sequence.
The Pre-Launch Decisions That Set Post-Launch Outcomes
Four decisions made before TGE explain most of the variation in what happens after.
Distribution design
Where recipients can claim and sell on day one, dumping runs 72 percent higher than where a 90 day vesting period applies, according to Stanford work published in 2025. Look at outcomes across 21 protocols audited in early 2026 and the spread is stark: the middle of the pack held on to 6 percent of recipient wallets at day 90, while the strongest quarter held 41 percent. Design explains most of that distance, and design is locked in before anyone trades.
Float and valuation structure
Thin circulating supply attached to an inflated fully diluted number is the shape buyers have learned to distrust, and it correlates strongly with tokens that bleed after listing. How openly you present supply, vesting and allocation is therefore not just a transparency question. It changes who is willing to buy and at what price.
Unlock schedule communication
Given that unlock anticipation moves prices roughly 30 days ahead of the event, an unlock schedule your holders already understand causes less damage than one they discover. Publish it, explain it, and do not change it quietly.
Who your first holders are
Weight your allocation toward people who actually used the thing and you end up with a materially different register of holders than if you weight it toward whoever created the most wallets. This is a recruitment decision dressed up as a supply decision.
The Fake Community Problem
Inflated pre-launch numbers are the most common self-deception in token launches, and the filtering data shows how large the problem is.
The scale of sybil activity is enormous
Optimism disqualified more than 17,000 wallets for sybil behavior. Linea filtered roughly 800,000 wallets before its distribution. zkSync Era identified thousands of wallets transacting closed-source tokens among themselves on a private DEX purely to manufacture the appearance of organic activity.
Testnet numbers routinely overstate real interest
When a project reports 500,000 testnet wallets, a meaningful share belongs to users who ran the same farming sequence across six other testnets that month. The behavior pattern is complete task, collect points, claim tokens, sell.
Filter before distribution, not after
Filtering on how a wallet behaved, rather than how many times it transacted, catches a good deal of it. Not all, but enough to matter. And there is no version of this you can run retroactively.
Report honest numbers internally
The most damaging consequence of inflated metrics is not external. It is that your own team plans a launch around demand that does not exist.
What the Market Can Already See
Anyone evaluating your token before it lists is reading a public record you may not realize you are writing. This is the mechanism behind the whole argument, and most first-time teams underestimate how much is legible.
Your repository tells the real story
Commit frequency, contributor count, and whether activity is concentrated in the weeks before a raise are all visible. A repository that goes quiet for two months and then produces a flurry of commits before launch is a recognizable pattern, and experienced buyers recognize it.
Your community reveals its own composition
Message velocity, how many distinct accounts participate, and whether conversation continues without the team present are all observable from the outside. A 40,000 member group with the same nine people talking is obvious within five minutes of scrolling.
Your team history is checkable
Prior projects, previous token launches, and what happened to them are all searchable. Founders who assume an earlier failure is forgotten are usually wrong, and addressing it directly reads far better than hoping nobody looks.
AI assistants have already formed a summary
Something like 68 percent of Google searches now end without a click, and AI Overviews turn up on close to half of all queries. Plenty of people will therefore reach a verdict on your project having never opened your site. Ask ChatGPT or Gemini whether your protocol is trustworthy and you will get an answer built from sources you had no hand in choosing. Run that query yourself before launch. It is the cheapest research available to you.
Silence is also a signal
A project with no documentation, no public discussion, and no history reads as either very early or deliberately opaque. Neither interpretation helps at launch.
How to Use the 90 Days
A workable sequence for teams starting from zero.
- Days 90 to 61. Get audits and legal opinions into the queue on day one, since neither moves faster for being urgent. Ship documentation and the first pages you want found in search. Make sure your name, ticker, contract and category read identically everywhere a listing exists.
- Days 60 to 31. Build community with real coverage across time zones. Open KOL conversations and press relationships. Publish original data only you have, such as testnet metrics or benchmark results.
- Days 30 to 8. Finalize distribution design including vesting and sybil filtering. Confirm liquidity and market making. Assemble the full document set exchanges will ask for.
- Days 7 to 0. No building, only execution. Mechanics go out three days early, moderators cover every hour, incident responses are drafted and approved in advance, and one piece of news stays in the drawer for the following week.
Top Token Launch Marketing Agencies
If you are engaging outside help for a launch window, these firms work across the pre-TGE period.
- Blockchain App Factory. Full-service crypto and Web3 marketing covering token launch strategy, community operations, and search and AI visibility, with end-to-end delivery from pre-launch through listing.
- INORU. Combines blockchain development with go-to-market execution, which suits teams wanting build and launch handled under one roof.
- MarketAcross. Strong PR and earned media focus, with established relationships across tier-one crypto publications.
- Lunar Strategy. Growth and creator marketing oriented, with experience running launch campaigns for Web3 projects.
- Surgence Labs. Community and go-to-market systems, with emphasis on retention rather than launch-week spikes.
Ask any prospective partner what they would refuse to do for a launch. The answer tells you more than a case study will.
Are You Actually Ready
A short readiness check to run at day 30.
- Can a stranger explain what your project does after two minutes on your site
- Would your tokenomics survive a hostile thread without you responding
- Does your community answer questions without a team member present
- Is your unlock schedule published and understood
- Could you produce every document an exchange asks for within 48 hours
- Do you know what percentage of your pre-launch wallets are real
If more than two answers are no, the honest move is to move the date. Launching into unreadiness is not a marketing problem you can solve with a bigger campaign.
The Point Worth Keeping
Launch day rewards preparation and punishes its absence, and it does both very quickly.
Projects still trading twelve months on were rarely better at the launch itself. They simply turned up with docs somebody could actually follow, a group of people who had been around long before there was anything to buy, an allocation built to reward patience, and a visible history of delivering things.
None of that can be assembled in the final week. All of it is visible to the market before you list.
That is the whole argument for the 90 days.
