"VS Mode": The Power of Comparative Analytics

The human brain is notoriously bad at processing absolute numbers. If I tell you "Your Time To Value is 4 hours," you have no reaction. Is that good? Is that bad? Is that normal?

But if I tell you:

"Your Mobile TTV is 4 hours, but your Desktop TTV is 10 minutes."

Now you are listening. Now you have a problem to solve.

Comparison provides Context. And Context is the oxygen of insight.

The Danger of "Global" Metrics

Most analytics tools show you a "Global" view by default. "Total Sessions," "Average Bounce Rate," "Global Conversion."

Aggregates smooth out the complications. They hide the bodies. You might have:

US Users: 2 minutes (Fantastic).
EU Users: 18 minutes (Terrible - likely due to GDPR cookie banners or latency).
Global Average: 10 minutes (Mediocre).

If you look at the Global Average, you try to "Fix Onboarding" generically. You tweak button colors. You change copy. But if you look at the comparison, you realize: "Oh, the EU consent banner is broken on mobile." You fix that one bug, and TTV drops for everyone.

3 Ways to Use Tivalio's VS Mode

We built VS Mode into the core of Tivalio because we believe comparison is the default state of analysis.

1. Time vs. Time (Trend Analysis)

"This Week vs. Last Week" Did our Tuesday deployment break anything?

Signal: P90 shifted right by exactly 4 minutes.
Cause: The new "Verify Email" step is slowing people down.

2. Segment vs. Segment (Cohort Analysis)

"Marketing Traffic vs. Direct Traffic" Are the users we are paying for actually activating?

Signal: Direct traffic activates in 5 mins. Paid Facebook traffic activates in 3 days.
Cause: Creating a misalignment between the Ad Copy ("Instant results!") and the Product Reality.

3. Outcome vs. Outcome (Success Analysis)

"Churned Users vs. Retained Users" What did the dead users do differently?

Signal: Retained users completed "Step 4" 90% of the time. Churned users skipped "Step 4" 80% of the time.
Cause: Step 4 is the "Aha!" moment. You need to make it mandatory or highlight it more.

The "What's Different?" Game

You don't need to be a data scientist to use this. Just open VS Mode. Pick a "Good Group" and a "Bad Group." Look at the TTV histograms side-by-side.

Are the shapes different?
Is one bi-modal (two humps) and the other uni-modal?
Is the "Bad Group" failing at the same specific timestamp?

Difference is data. Find the difference, and you find the fix.