One thing I’ve learned about product metrics is that averages can sometimes hide the most important part of the story.

Imagine you measure how long it takes users to reach their first meaningful value and get an average of 12 minutes.

That sounds useful.

But what if half your users reach value in 3 minutes while a smaller group takes more than an hour?

The average doesn’t tell you that.

This is where percentiles become particularly useful for measuring Time to Value.

What Is a Percentile?

A percentile tells you how a particular percentage of observations compares with a specific value.

Suppose you measure Time to Value for 1,000 users.

If the 50th percentile is 8 minutes, it means 50% of users reached value in 8 minutes or less.

If the 90th percentile is 35 minutes, it means 90% reached value within 35 minutes, while the slowest 10% took longer.

This gives you a much better view of the distribution than a single average.

Why Average Time to Value Can Be Misleading

Time to Value often has a long tail.

Most users might reach value quickly, while a smaller group gets stuck for a very long time.

For example:

Average: 18 minutes
P50: 7 minutes
P90: 42 minutes
P95: 75 minutes

If you only look at the average, you might conclude that users generally need around 18 minutes.

But the percentiles reveal something more interesting.

Most users are getting there relatively quickly, while a meaningful minority is experiencing significant friction.

That is a very different product problem.

P50 Tells You About the Typical User

The 50th percentile, or median, is often a good starting point.

It answers:

“How long does it take for a typical user to reach value?”

Because the median isn’t heavily affected by extreme outliers, it can provide a more realistic picture than the average.

If your P50 Time to Value falls from 10 minutes to 6 minutes after an onboarding improvement, that’s a useful signal.

But don’t stop there.

P90 Shows the Long Tail

The 90th percentile is particularly useful when thinking about friction.

It answers:

“How long does it take before most users reach value?”

Suppose your P50 improves from 10 minutes to 5 minutes, but your P90 remains at 60 minutes.

You’ve made the typical experience faster, but a significant group is still struggling.

That could point toward:

• Complex setup
• Confusing workflows
• Missing information
• Technical issues
• Specific customer segments needing different experiences

The long tail can reveal problems that the median hides.

Compare Percentiles Over Time

Percentiles become even more useful when you track them over time.

Imagine:

Metric	Before	After
P50 TTV	12 min	7 min
P90 TTV	45 min	25 min
P95 TTV	90 min	50 min

Now you can see that the improvement isn’t limited to the typical user.

The experience has improved across the distribution.

This is much more informative than simply saying, “Average Time to Value decreased by 30%.”

Segment Your Percentiles

Overall percentiles can still hide important differences between user groups.

Compare:

Self-serve users: P50 = 6 min
Enterprise users: P50 = 18 min

That tells you something worth investigating.

Enterprise users may have additional configuration, integrations, permissions, or approval steps before reaching value.

Similarly, you migh