← Back to JunoDesk
Incrementality Testing

Don't Just Measure Performance.
Prove Incrementality

Geo-holdout experiments that isolate what a channel actually caused, run end-to-end by JunoDesk and delivered under your agency's name.

Book a Demo →
Total Revenue
Incremental Control Lift Test
Control Lift Test
Incremental delta +18%
Why It Matters

Modeling Tells You.
Testing Proves It

A model estimates what worked. A geo-holdout measures it — real markets, a real control group, a real number.

Test an Existing Channel

Confirm a channel already in the mix is actually earning its budget — not just riding on demand that would've converted anyway.

Validate Before You Scale

Prove a new channel works in a handful of markets before your client commits a national budget to it.

Find the Ceiling

See exactly where a channel's returns start diminishing — before the next budget review finds out the hard way.

Alaska Alabama Arkansas Arizona California Colorado Connecticut Delaware Florida Georgia Hawaii Iowa Idaho Illinois Indiana Kansas Kentucky Louisiana Massachusetts Maryland Maine Michigan Minnesota Missouri Mississippi Montana North Carolina North Dakota Nebraska New Hampshire New Jersey New Mexico Nevada New York Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Virginia Vermont Washington Wisconsin West Virginia Wyoming
Control Test
The Process

How Does Incrementality Testing Work?

You tell us what to test. We plan it, run it, and hand back a number.

01 — Design Experiment

We design the test

Send us the channel, the markets in play, and the KPI that matters. We handle the rest — spend-up or spend-down, how long the test needs to run, and the statistical power it needs to hold up.

What comes back is a test plan already built: which markets to hold out, how much budget shifts, and the smallest lift the design can reliably catch — nothing for you to configure.

Minimum Detectable Lift, by Weekly Spend
12% 4% WEEKLY TEST SPEND →
Baseline · $22K/wk High-Confidence · $41K/wk
02 — Analyze Experiment

We read the results

Once the test wraps, we compare what happened in the geos where spend actually changed against the holdout markets where it didn't. The gap between them is the lift the channel caused — not the credit it claimed.

You get a confidence interval, not just a single number, so you know how far to trust the read before it goes into a budget conversation.

Cumulative Revenue, Test vs. Control
+$19.2K Incremental
Test Control Incremental Lift Day 0 Test start Day 35
90% Confidence Interval: $14.8K – $26.3K
03 — Action Results

We turn it into a plan

A result on its own isn't a plan. Here's what we do with it:

  • Move budget into the channel the test proved — not the one the platform dashboard claims credit for.
  • Fold the number into your client's MMM, so the next forecast starts from a measured lift instead of an assumption.
  • Size the next test off what this one just taught us.

It's a number your client's finance team can hold you to, because it came from a market that didn't get the spend — not from a model.

Results Channel ROAS Confidence Interval
Experiment 1 YouTube 1.5x 1.1x – 2.1x, 90% certainty
Experiment 2 Meta 4.9x 4.1x – 6.0x, 90% certainty
Recommended Budget Shift
Before Test After Test
Meta +31% YouTube −38%
Contact

Ready to Go Beyond Reporting?
Let's Talk

Let's talk about your next client, project, or pitch.

Book a Demo →