How to Make Better Predictions
Today’s newsletter is powered by Rise.
Rise is a modern job search platform that makes finding a job... gasp... joyful. We’re the rebels, misfits, and underdogs rewriting the rules on how to land jobs you love (yes, even remote jobs). Come join us today!
Innovation depends on accurate predictions. When we built myYearbook (which later became MeetMe and The Meet Group), we didn't stay relevant by accident. We made bets on where social networking was going, what features would matter, and how people would interact online. If we had invested in the wrong trends—ones that fizzled before we even launched—we wouldn't have lasted.
Every decision is a prediction. Whether launching a product, hiring a key employee, or deciding where to allocate resources, you're betting on an outcome. The problem is that too many people make predictions like fortune tellers—based on gut feelings and wishful thinking—rather than like forecasters who systematically assess probabilities.
When predictions go wrong, it's easy to shrug and say, "Well, nobody saw that coming." But some people did.
Here's how you can, too:
1. Make More Predictions (and Track Them)
Every year, experts and companies make bold predictions about the next big trends. Then, the year passes, and no one goes back to check how they did because it's easier to remember the flashy call than the accuracy rate.
If you want to improve your forecasting, you need to record your progress. Make many predictions, track them, and revisit them. How did they go?
In Managerial Decision Making at Wharton, we did an exercise to estimate the probability of many random future events.
What’s the probability that it will snow on February 28 in Philadelphia?
What’s the probability AAPL stock will close over $245 on March 31?
There were maybe twenty predictions in each round covering all sorts of topics: finances, politics, the weather, sports, war, etc. At first, our guesses were all over the place. But we did the exercise twice, and unsurprisingly, in the second round, the class improved. The difference? We were forced to think in calibrated probabilities instead of gut instincts.
2. Be Certain About Your Uncertainty
Most people are overconfident in their predictions. And the more confident they sound, the more convincing they are, especially in business.
But confidence doesn't equal accuracy.
In class, we used a scoring method (Brier score) that penalized overconfidence. If you said something had a 100% chance of happening, and it didn't, you weren't just wrong—you got destroyed.
If you’re forecasting DAU, revenue, or engagement, you should assume your model will be “wrong.” It’s hard to get to an exact number when forecasting something a year or five years in the future. The question is by how much and why. Stress-test your assumptions and build ranges into your estimates.
A good forecaster isn't afraid to hedge. They don't say, This will 100% work. Instead, they say:
There's a 40% chance this works, a 25% chance it flops, and a 35% chance it lands somewhere in between.
Identify the potential outcomes and assign probabilities to each.
We assessed B2B partnerships at TMG using probabilities. Instead of treating deals as binary—they'll sign or won't—we assigned probabilities based on discussions. This allowed us to prioritize deals more effectively.
3. Remove Your Bias (You Have More Than You Think)
Deep industry expertise is excellent until it becomes a blind spot.
Superforecasters (people who consistently excel at predictions) tend to be generalists. They draw from multiple sources rather than relying on their experience alone.
Political scientist Philip Tetlock's research found that people who think like hedgehogs—people who see the world through one big idea—tend to make worse predictions. Foxes tend to be better forecasters. They gather insights from everywhere and update their thinking as new information emerges.
People's biases affect their predictions all the time.
If you're conservative, you likely predict that a conservative leader will improve the economy.
If you're liberal, you likely predict the opposite.
If you've worked in an industry for years, you might assume you "just know" how it will evolve—even when data suggests otherwise.
I fall into this trap, too. Coming from the dating industry, I have strong opinions about dating apps. But if I rely solely on my experience instead of researching industry shifts, I'll get things wrong.
The best way to remove bias? Expose yourself to different perspectives.
Read from sources outside your usual bubble.
Seek out opinions that challenge yours.
Don't ignore what the data says because it doesn't fit your worldview.
And when in doubt, listen to the crowd. Want to win a "Guess how many jelly beans are in a jar" contest? Average all the (independent) guesses. The bad forecasts and predictions will likely cancel out, allowing you to land on the real most likely scenario.
The Goal: Consistently Making Better Bets
We can't see the future, but we can get better at predicting it. It's not about being right every time—it's about making higher-quality decisions over time.
And when we do get something wrong? The best thing we can do is analyze why—so next time, we're just a little bit better.
📖 Grab a copy of my book! Amazon | Barnes & Noble | Signed Copy
💥 Learn more about my marketing crash course.
🔌Need help with performance marketing or GTM strategy? Contact me here or book a power hour