When people ask, “What’s the ROI on a startup investment?” they usually expect a clean percentage. After a few deals, you realize that startup ROI is rarely clean. Cash comes back late (if at all), valuations move before liquidity, and the most important number is often the probability that your investment goes to zero. In public markets, you can sell tomorrow. In startups, you’re often locked in for 5–10+ years, and the outcome distribution is brutally lopsided.
Why simple ROI is not enough
Traditional ROI is straightforward: profit divided by invested capital. But startups add three complications that make the simple ratio misleading:
- Time: a 3x return in 2 years is a very different outcome than 3x in 10 years.
- Illiquidity: you may “have” gains on paper while still holding an unsellable position.
- Power-law outcomes: a few outliers drive most of the portfolio return while many deals fail.
That’s why serious analysis uses multiple lenses at once: MOIC for the multiple, IRR for time, and fund-level measures (TVPI/DPI) for what’s real versus what’s still marked to model.
The core metrics you should know
1) MOIC (Multiple on Invested Capital)
MOIC = Total value returned (or current value) ÷ Total invested capital.
This is the simplest “how many times my money” number. It’s intuitive, but it ignores time. A 5.0x sounds great until you learn it took 12 years and required follow-on checks that weren’t in your original plan.
2) IRR (Internal Rate of Return)
IRR is the annualized return that makes the net present value of all cash flows equal to zero. In practice, IRR answers: how efficiently did this investment compound per year, given when cash went out and came back?
My mental shortcut: if two deals both return 3.0x, the one that returns sooner is usually better. Time is not just a detail; it’s the deal.
3) TVPI and DPI (Fund-level reality checks)
- TVPI = (Distributed cash + Remaining value) ÷ Paid-in capital
- DPI = Distributed cash ÷ Paid-in capital
TVPI tells you the total “reported” value, including unrealized marks. DPI tells you how much cash actually came back. In slow exit environments, DPI can lag for years, even if TVPI looks healthy. Personally, I treat DPI as the honesty meter.
Startup ROI Metrics at a Glance
| Metric | What it answers | Best for | Common trap |
|---|---|---|---|
| MOIC | How many times did it return? | Simple outcome comparison | Ignores time and follow-ons |
| IRR | How fast did it compound per year? | Time-sensitive deal evaluation | Can be distorted by timing quirks |
| TVPI | Total value incl. unrealized marks | Fund performance snapshot | Marks can lag reality in down markets |
| DPI | How much cash has been returned? | Liquidity and realized success | Can look low until exits reopen |
A practical workflow I use to evaluate a startup ROI story
Step 1: Map cash flows, not headlines
Start with a simple timeline: when you invested, when you invested again (if you did), and when/if cash came back. Valuation updates are useful, but cash flows are the real story for ROI.
Step 2: Compute MOIC, then pressure-test it
MOIC is your first filter. But I immediately ask: did this require pro-rata participation? Were there bridge rounds? Did my ownership dilute more than expected? A “paper 10x” can shrink quickly once you model dilution and follow-on capital needs.
Step 3: Convert the story into IRR intuition
Here’s the reality check I lean on: the same multiple can be mediocre if it takes too long. For example, 3.0x over a decade feels very different from 3.0x over two years. That’s why IRR matters so much in venture.
Step 4: Add failure probability and portfolio logic
Single-deal thinking breaks in venture. You have to assume a meaningful chunk of investments return 0–1x. The portfolio is designed so that a small number of outliers dominate returns. This is why diversification across 10–20+ deals is often recommended, and why “one great pick” is more myth than process.
Step 5: Separate realized vs unrealized value (TVPI vs DPI)
If you’re looking at a fund or an angel portfolio, I always split performance into what has been distributed (DPI) and what’s still marked as value (TVPI). In tough exit markets, TVPI can flatter. DPI keeps you grounded.
A concrete mini-example (how the numbers change the conclusion)
Imagine you invest $100k at seed, add $50k two years later, and exit six years after the first check for $1,000k returned to you.
- Total invested: $150k
- Total returned: $1,000k
- MOIC: 1,000 ÷ 150 = 6.67x
That sounds excellent, but the real question becomes: was the cash tied up for too long, and how many similar deals went to zero in the same portfolio? A single 6.67x can be the hero that rescues a portfolio, or it can be “nice but not enough” if too many others failed and capital stayed locked for a decade.
Key takeaways
- Use MOIC to measure magnitude, IRR to measure speed, and TVPI/DPI to measure reality.
- Always model time and follow-on capital; they change the outcome more than people expect.
- Assume power-law results: most deals won’t drive returns, a few will.
- Judge performance at the portfolio level, not by one trophy investment.
If you treat startup ROI as a single percentage, you’ll miss the two things that matter most: when you get paid, and how often you don’t. The best analysis turns the narrative into cash flows, then uses multiple metrics to keep the story honest.
Understanding Risk-Adjusted Returns: A Practical Guide to the Sharpe Ratio
0 Comments