This article is written by a VC (venture capital investor) who has analyzed retention (revisit/reuse) data from a wide range of startups over 15 years, offering a clear explanation of the essence and laws of product retention based on real-world experience. The core conclusion is that bad retention is nearly impossible to fix, while products with great retention possess a kind of "magic." It covers the keys to product success, practical insights, and common mistakes frequently made across the industry.

1. An Investor's Perspective from 15 Years of Observing Retention Curves

The author has worked as a founder, product manager, and now a VC at Andreessen Horowitz (a16z), personally analyzing data from hundreds of startups each year. They emphasize that the very first thing they look at when reviewing startup data is the retention curve. Through thousands of data rooms, countless A/B tests, and numerous onboarding and notification experiments across various products, the key realization is that retention has consistent patterns (properties) and laws.

"Just as there are laws of physics, retention curves have consistent patterns that appear repeatedly over time."

2. Nine Fundamental Laws Found in Retention Curves

The author identifies nine fascinating rules in the retention curves observed:

• Bad retention can't be fixed: Additional notifications (AI-powered messaging, feature additions, etc.) alone cannot fundamentally improve retention.
• Retention only goes down, never up: With rare exceptions, retention curves naturally decline, and if the early numbers are bad, they only get worse.
• Revenue retention grows while user retention shrinks: A smaller remaining group of users can generate even more revenue.
• Retention is relative to product category: It's impossible to make a hotel booking app something people use daily.
• As users grow, retention declines: The best users are the early core fans.
• Churn is asymmetric: Bringing back a lost user is far harder than acquiring a new one.
• Measuring retention is surprisingly tricky: Seasonality, bugs, and new experiments muddle the data.
• Even viral growth collapses without good retention: Products with chronically low retention ultimately fail.
• Products with amazing retention are true magic: These products, which appear only rarely, are on a completely different level.

3. The Nature of Retention: Why Is It So Hard to Fix?

Bad retention is nearly impossible to fix after the fact. If initial metrics are extremely poor, the author emphasizes from experience that meaningful improvement is virtually impossible without a fundamental change in direction (a major pivot).

"If your D1 retention is 10%, you've probably built a product nobody wants. At that point, no amount of A/B testing is enough to 'bend' the curve."

To truly improve retention, the author points out that what's needed is not small incremental improvements but a major pivot that completely redesigns the product's meaning, core functionality, and target market.

4. Retention Curves Fundamentally Decline

For most products, the retention curve drops exponentially. For example, if D1 is 40%, D7 might be 20%, D30 might be 10%, halving each time until only a tiny fraction remains.

"If early retention isn't strong, later retention will never be good either. The beginning must always be strong for the end to be strong."

The author also explains that recovery curves can appear in some exceptional cases, such as hardcore products (e.g., online poker) or products with strong network effects, but these are extremely rare.

5. Revenue Retention vs. User Retention: The Power of B2B

While user-based retention keeps declining, it is notably pointed out that revenue-based retention (Revenue Retention) can actually grow as remaining users/customers generate increasingly more revenue over time.

"One of the greatest strengths of B2B SaaS products is that the revenue retention curve grows over time. Products like Slack are a prime example."

In contrast, most consumer products cannot enjoy this blessing of revenue retention, making them a much harder business, the author advises.

6. The Relativity of Retention: Category-Specific Limits and Nature

Retention has inherent limits (Nature vs. Nurture) that vary by product category. For example, team collaboration apps, messengers, and social networks are "daily use," but travel/hotel booking apps and medical information apps are used infrequently, so their retention is naturally lower.

"It's hard to turn a travel app into a social-purpose product used every day. There are natural limits to product usage frequency."

Therefore, the author advises that retention potential is already determined differently at the category selection stage, and business models need to be designed according to these inherent limits.

7. Retention Deterioration During User Expansion

Even if the initial core user group (the golden cohort) has excellent retention, retention drops dramatically among new user segments when mass adoption occurs through advertising and other channels. This is due to differences in initial product-market fit, needs, and referral-based acquisition.

"The best users are the ones who came in first, organically. The later they arrive, the lower the fit tends to be."

This is why how to retain the core user base remains critically important even after expansion.

8. Asymmetric Churn: Users Who Never Come Back

The most painful reality in retention is that churned users don't come back. If the churn rate is too high, the author explains that it's actually more efficient to focus entirely on acquiring new users rather than trying to win back lost ones.

"It's easier to acquire a new user than to bring back an old one. Natural re-engagement through network effects is almost the only path."

9. Retention Is Hard to Measure in the First Place

Measuring retention metrics is no easy task. The author emphasizes that data interpretation must be done very carefully due to seasonality, bugs, experiments, and new market launches.

"Retention curves should always come with an asterisk (*). Comparisons are frequently impossible because of seasonal effects, bugs, and market launch effects."

10. Even Viral Growth Fails Without Good Retention

Even if a product grows rapidly through massive early influx (virality, influencer promotion, etc.), the author emphasizes through multiple case studies the lesson that products with fundamentally poor retention will inevitably collapse.

"If you go all-in on virality to scoop up users while neglecting retention, the bubble will eventually burst and collapse. The industry has already experienced this countless times."

In fact, the author provides empirical evidence that only products combining virality with high retention, such as Facebook and Substack (blog case studies), survive.

11. Great Retention Is the Product of Magical Discovery

The very few products with truly outstanding retention are born not from rigorous data-driven experimentation or repetitive optimization, but from genuine intuition (insight) and discovery about the market and users, the author emphasizes.

"Sometimes you come across a product with 50% D30 retention, and that's truly 'magic.' These products had extraordinary insight from the very beginning."

12. How Can You Build a 'High Retention' Product?

After reading all of the above, the natural question everyone asks is, "So how exactly do you design for high retention?" The author acknowledges that the answer is not easy, but offers the following hints:

• The idea itself is genuinely very important
• Target categories known for high retention
• Aim for product categories that are already successful and used daily
• Build with the resolve to directly "replace" an existing popular product head-on
• You must be able to intuitively explain the core value to users within 60 seconds

However, the author emphasizes that even when taking on an existing product head-on, timing and differentiation regarding "why now" must follow as essential prerequisites.

"There are already satisfying products in the market. If differentiation is lacking or the timing is off, the problem may be with acquisition rather than retention."

13. The Exceptional Value of Creating an Entirely New Market

Finally, the author adds that in the realm of innovation, it is always possible that a genuinely new market, not just a remix of an existing one, can open up. Yet even in such cases, most innovations evolve (jump) based on previously little-known precedents or clues, and the "last mover" often achieves the greatest success.

"Google was the 10th search engine, and the iPhone wasn't the first smartphone. Truly new paradigms are extremely rare, but those attempts are what created today's innovation industry."

Conclusion

This article repeatedly emphasizes that the laws of retention are the most fundamental factor in product development and business success. From the nature of categories, the power of early users, the difficulty of measurement, the relationship between retention and revenue, to the rare magic of true innovation -- these are lessons that should be etched into the DNA of every startup. In summary, if you cannot design for "good retention," all your efforts can crumble in an instant. The true secret to product success lies in never losing sight of the "essence" that pierces the fundamentals.