Building a product is easy; keeping people from running away is the hard part. If you’ve spent any time in the Silicon Valley ecosystem, you’ve likely heard of Andrew Chen. He’s the guy from Andreessen Horowitz (a16z) who basically wrote the book—or at least the most famous blog posts—on why most apps die a quiet, lonely death.
His Andrew Chen good retention heuristic isn't just a single number you can memorize. It's a sobering reality check for anyone who thinks a spike in downloads means they’ve "made it."
The Brutal Reality of the Retention Curve
Most people look at their total downloads and feel a rush of dopamine. They shouldn't. Chen’s research, particularly his work with data from Quettra, shows that the average Android app loses 77% of its daily active users (DAUs) within the first three days.
Think about that.
You spend thousands on ads, pull every growth lever you have, and 72 hours later, three-quarters of those people are gone. By thirty days out? You’re looking at a 90% loss. This is what Chen calls the "shark fin" curve, and honestly, it’s where most startups go to die.
The Benchmarks That Actually Matter
If you want to know if you're actually building something people give a damn about, you need to look at where your curve flattens. A "good" product doesn't just have a higher starting point; it has a floor.
Here is how the Andrew Chen good retention heuristic breaks down across different categories:
- Social Networks: These are the gold standard. A "good" social app should see D30 retention between 15% and 25%. If you’re Facebook or Instagram in their prime, you’re looking at much higher, but for a new player, hitting 20% after a month is a massive win.
- SaaS and Productivity: These products aren't always used daily, so the heuristic shifts. You’re looking for a "smile" curve. If your weekly retention is stable at 30% or 40%, you’ve likely found product-market fit.
- On-Demand Services: Think Uber or DoorDash. These have a different cadence. A good retention rate here might be lower in terms of daily usage but higher in "lifetime" value.
The real "heuristic" here? If your retention curve doesn't flatten out—if it just keeps sinking toward zero—you don't have a marketing problem. You have a product problem.
Why 20% is the Magic Number for Stickiness
Andrew Chen often points to "stickiness" as the ultimate health metric. This is usually calculated as the ratio of DAU to MAU (Daily Active Users divided by Monthly Active Users).
Basically, it tells you what percentage of your monthly crowd shows up every single day.
For a standard consumer app, 20% stickiness is considered "good." It means your users are building a habit. If you hit 50%, you’re in the "world-class" territory alongside WhatsApp and Slack. If you’re at 5%, you’re essentially a leaky bucket. You can pour as much "acquisition" water into that bucket as you want, but you’ll never fill it.
The First Visit is the Only Visit That Counts
One of Chen’s most famous insights is that you can’t "optimize" your way out of a bad first experience. Most teams spend months tweaking their email re-engagement campaigns or sending push notifications to "bring users back."
Chen argues this is mostly a waste of time.
The leverage is almost entirely in the first 3-7 days. If a user doesn’t find "the magic" in the first session or two, they are gone. Period. No amount of "We miss you!" emails will change the fact that your app didn't solve their problem or entertain them when they first opened it.
The Adjacent User Theory
This is a nuance many people miss. Once you’ve captured your "power users"—the people who naturally love what you built—your retention will start to drop. Why? Because you’re starting to attract the Adjacent User.
🔗 Read more: Eastman Credit Union Video Johnson City: Why Banking by Screen is Changing Everything
These are people who are almost a fit for your product but find it slightly too confusing, too expensive, or too niche. To maintain a good retention heuristic, you have to stop building for your fans and start building for the people who are currently bouncing.
How to Fix Your Numbers
If you’ve run the numbers and realized your retention is trash, don’t panic. Well, maybe panic a little, but then do this:
- Watch the "N-Day" Curve: Stop looking at averages. Look at cohorts. People who joined in January—how many are left in February? If the January group is retaining better than the February group, your product is getting worse (or your marketing is getting "shallower").
- Identify the "Aha" Moment: What is the one thing a user does that correlates with sticking around? For Facebook, it was adding 10 friends in 7 days. For Slack, it was sending 2,000 messages as a team. Find your number.
- Aggressive Onboarding: If the first 3 days are where 77% of users leave, your onboarding shouldn't be a "tour." It should be a beeline to the value. Strip everything else away.
The Andrew Chen good retention heuristic is a reminder that growth is a vanity metric, but retention is sanity. You can't build a business on a foundation of people who don't come back.
Next Steps for Your Product:
Start by pulling your D1, D7, and D30 retention numbers for your last three monthly cohorts. If the line doesn't flatten out by Day 30, stop all paid acquisition immediately. Spend that budget on user research to find out exactly where people are getting confused in their first session. Move your "Aha moment" earlier in the user journey—literally within the first 60 seconds if possible.