Founders talk about compounding channels as if the hard part is choosing one. Often the harder part is waiting until the channel has enough substance to compound at all.
Before a loop becomes a loop, it is usually just a person doing awkward work by hand. The good news is that this manual phase is not wasted motion. It is where the team learns what kind of density the product actually needs.
The first density is usually social
Glasp's early push is useful because it was narrow on purpose. The team used persona-curated community outreach in product-management circles instead of trying to appeal to everyone at once. A tightly chosen user type gave the product a place to start making sense.
Then it added a secret-signup waitlist illusion and ran a founder screen-share onboarding sprint. That sounds scrappy because it was. But the point was not theater. The point was to find out who cared enough to lean in and what they actually did once they arrived.
Buffer Analyze did something similar with manual empty-state concierge onboarding. It filled the product with proof before asking the user to imagine the value. Different product, same instinct. Do the human work first so the team can see what the system eventually needs to preserve.
The second density is signal
The Glasp story gets more interesting when AI starts heating up. The team did not jump straight to a grand product thesis. It used a playful side-project wave radar move first. DALL-E-dle was small, odd, and easy to ship. That made it perfect for learning whether the new wave had real appetite.
This is a useful discipline because side projects are less expensive than category bets. If the strange little experiment gets traction, you earn more conviction. If it does not, you have only lost a small amount of time instead of organizing the whole roadmap around a hunch.
The third density is content
Only after Glasp had built enough user activity did it ship critical-mass UGC SEO release pages. This is the part I wish more teams understood. Programmatic pages are not valuable because they are numerous. They are valuable when each page contains enough real material to deserve being found.
Glasp waited until it had millions of highlights spread across roughly a million articles. Then the pages were not empty templates. They were actual reading artifacts. That is why the indexing spike was meaningful. The density came first. The crawlability came second.
Where this applies
For SaaS, this means watching real setups before you pretend the onboarding is solved. For AI products, it means testing emerging waves with small experiments before committing the whole brand to them. For SEO-heavy products, it means holding back the auto-generated surface until the underlying data is rich enough to make the pages useful.
The trap is automating a thin surface. A waitlist with no real demand behind it, a content system with no substance under it, a funnel that gets faster at moving people into the wrong first experience. Density is what makes compounding worth having. Until you have that, the manual work is not the embarrassing part. It is the part that tells you what deserves to scale.