Is the AI startup gold rush over—or just getting honest?
The last couple of years felt like a feeding frenzy: spectacular demos, viral sign-ups, headlines that made startups look like overnight winners. Now founders, operators and investors are sobering up. The central question has shifted from “Can this tech impress?” to “Can this product keep customers coming back while making money?”
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.
2. The numbers that actually matter
Drop the vanity metrics. The metrics that separate hype from a business are gritty and specific:
- – Churn (cohort or monthly): does the product solve a repeatable problem?
- LTV (lifetime value): how much revenue does a retained customer generate?
- CAC (customer acquisition cost): how much do you spend to win each customer?
- Gross margin and ARPU: are you making meaningful profit per user?
- Burn and runway: how long can you iterate toward product-market fit?
A viral loop can temporarily depress CAC. But if retention is weak, you’ll need paid channels to sustain growth—and CAC will climb. That arithmetic catches startups off guard more often than you’d think.
3. Stories from the field: what works and what doesn’t
Failure from the inside
At my second startup we built a demo that looked magical—press coverage, pilot customers, and a spike in weekly active users. But that spike masked a problem: monthly churn sat around 28%. The product solved a niche workflow that didn’t reoccur, so signups didn’t become steady users. We chased paid signups, CAC rose, and LTV never covered acquisition costs. We burned runway iterating on features that didn’t move retention. Demos got attention; recurring utility earns renewals.
Public success examples
Look at companies like Stripe and Slack. They laser-focused on a single, repeatable job-to-be-done and shortened time-to-value. They tracked feature-driven cohort retention and ruthlessly optimized onboarding for immediate usefulness. Users who returned because the product continuously solved a problem made acquisition economics manageable—LTV rose, CAC stayed reasonable, and scaling became sustainable.
4. Concrete lessons and how to act on them
Map one recurring job the product must own
Define the single repeatable outcome that makes a customer come back. Don’t try to be everything to everyone at once.
Measure cohort retention by feature
Instrument your funnel so you can see which features cause users to return. Activation and 30/60/90-day retention by cohort are far more revealing than raw downloads.
Shorten time-to-value
Remove steps between discovery and the moment a user experiences real benefit. The faster you deliver measurable value, the lower the churn and the healthier the CAC.
Price for outcomes, not just usage
When possible, tie pricing to measurable business outcomes or clear tiers that reflect value. Outcome-linked pricing reduces pilot friction and shortens procurement conversations.
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.0
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.1
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.2
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.3
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.4
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.5
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.6
1. Ask the uncomfortable question first
Before you celebrate a growth spike, answer this: how many of those users will still be active six months from now, and at what cost did you acquire them? Downloads and demo views are noise. Real businesses are built on repeat use, defensibility and sustainable unit economics—not on momentary attention.7