Contact

Results

Two brands. Donut helped both drive real results.

Case study 1 · Destination launch

$1.54M in room revenue our model credits to a luxury resort launch.

~3,395 room nights credited to the ads3

We don't just run campaigns. We measure what they actually move. Here's a recent launch, by the numbers.

Luxury International Resort & Theme Park

Launching a new destination to families and couples.

Over 2025–2026, one omnichannel program carried it across Social, Google, CTV, DOOH, and OTA.

At a glance

Client
A luxury international resort and theme park, kept anonymous
Goal
Launch a new destination to families and couples
Channels
One omnichannel program across Social, Google, CTV, DOOH, and OTA, 2025–2026
Room revenue credited by our model
~$1.54M, from ~3,395 room nights. A model estimate, not a holdout test, and the range around it is wide3
Awareness lift, ages 55–64
+63% for our lead YouTube spot, at $0.23 per lifted user. Not every age group or device showed lift1
Best cut on TV screens
Played to the last frame on 89% of its TV plays. All 12 cuts averaged 40.7%2
Ad to hotel stay
2–6 weeks, strongest around week 43
Measured with
A YouTube Brand Lift study, the campaign's video report, and our marketing effectiveness model

Side by side

Our number What it's compared with

+63% awareness lift among 55–64s1

People aged 55–64 who saw our lead YouTube spot, against a control group who didn't. Overall awareness rose 7%.

Saw the spot 6.73%
Control group 4.13%

Cost per lifted user1

Lower is better. Same study as the awareness lift.

Ages 55–64 $0.23
Whole study, blended $1.02

Played to the end on TV screens2

Our best cut, a 10-second spot, played to the last frame on 89% of its TV plays. That's more than double the average across all 12 cuts in the campaign.

Best cut, 10 seconds 89.0%
All 12 cuts, average 40.7%

The living-room effect1

From the same brand lift study. On TV screens, people who saw the spot were about 20% more aware than the control group. The story landed in the living room.

Saw the spot on a TV screen 9.01%
Control group 7.54%

2–6 weeks from ad to hotel stay3

The model found the ads show up in room nights 2–6 weeks after people see them, strongest around week 4. Anything past six weeks, like brand building, is outside what this model measures.

Sources and notes

  1. YouTube Brand Lift study, Dec 21–24, 2025. It measured awareness for our lead video spot, comparing people who saw it with a control group who were eligible but didn't. Among 55–64s, awareness climbed from 4.13% to 6.73% at a $0.23 cost per lifted user, versus $1.02 across the whole study. Overall awareness rose 7%. On TV screens, 9.01% of people who saw the spot were aware, versus 7.54% of the control group, about 20% higher. Not every age group or device showed lift.
  2. Video completion. From the campaign's YouTube video report, TV screens only (the YouTube app on connected TVs), 2025–26. Completion is the share of impressions that played to 100%. The 10-second cut played to the end on 89.04% of its 6.1M TV impressions. All 12 cuts together played to the end on 40.7% of 46.1M TV impressions. That 40.7% is the campaign's own average, and it includes the 10-second cut. It isn't an industry benchmark. Across every device, the same two numbers were 80.4% and 27.9%.
  3. Revenue and ROAS. From our marketing effectiveness model, which lines up daily ad impressions by channel against room nights from Nov 2025 to Apr 2026 and separates out day-of-week and holiday demand. It credits ~3,395 room nights and ~$1.54M in room revenue to the ads, showing up in stays 2–6 weeks after people see them. It's a model estimate, not a holdout test, and the range around it is wide.

Case study 2 · Retail media mix

$10M+ a year in incremental revenue driven by advertising – measured for the first time.

Tens of millions over the past three years1

They didn't know if their advertising was working. Our model showed them, for the first time, that it was. Now they use it to steer their media mix, and we used it to make the case for a 20%+ bigger ad budget.

Florida Retailer

Is our advertising actually working?

Most of their sales happen in their stores. Ad platforms can only estimate store visits, and none of them can say which sales would have happened anyway. Ad partners' own reports claimed returns that looked too good to be true. We built a media mix model that measures advertising against all sales, in store and online, and we update it every month.

At a glance

Client
A Florida retailer, kept anonymous
The question
Is our advertising actually working?
Where sales happen
Mostly in their stores
Method
A media mix model built on Google's open-source Meridian, measured against all sales and updated every month1
Incremental revenue credited
$10M+ a year from advertising, the model's estimate. Tens of millions over the past three years1
Budget
The next year's digital ad plan came in more than 20% higher2
Ad cost per 1,000 impressions
Down 54% on a key channel over the last 12 months4

How a media mix model works

A made-up example to show how a model works. Nothing in it is the client's data.

  1. Line up the data

    The model looks at weekly sales next to weekly ad spend in each channel. It also takes in the things that move sales on their own, like seasons, holidays and promotions.

  2. Separate what ads added

    It estimates how much of sales would have happened anyway (the baseline) and how much each channel added on top. Ads keep working for a few weeks after they run, and the model allows for that.

  3. Move the next dollar

    Past a point, each extra dollar in a channel earns less than the one before. The model shows where the next dollar earns more, and that guides where the budget goes.

The split between baseline and ads is the model's estimate from the data, not a controlled test.

What changed

For the first time, an answer1

The model's answer was yes: advertising adds sales on top of what would have happened anyway. When most sales happen in a store, that's a number ad platforms can't give you.

  • More in sales than it costs

    Leadership could see what advertising brings in, in their own sales numbers rather than an ad platform's.

  • A bigger budget2

    We used the model to make the case for spending more. The next year's digital ad plan came in more than 20% higher.

A mix they can steer3

The model estimates the mix as a whole adds more than $10M a year in incremental revenue. It also estimates which channels earn the most on the next dollar, and we refresh it every month. The client plans the mix with it. It informs the calls, and the people who buy each channel still make them.

  • Choosing between partners

    When two options look alike on price and reach, the model's read on the next dollar helps decide.

  • Moving money toward what earns more

    Each refresh shows which channels earn more on the next dollar, and the plan shifts toward them.

Cheaper reach4

Over the last 12 months, the cost per 1,000 impressions on a key channel fell 54% against the 12 months before, so the same budget reaches more people.

Cost per 1,000 impressions on a key channel, last 12 months −54%
The 12 months before

Sources and notes

  1. Media mix model. Built on Google's open-source Meridian MMM architecture. It looks at the effectiveness of all digital advertising against the baseline, controlling for the macro environment. How we use Meridian.
  2. Budget. The client's digital ad plan for the fiscal year after the model's first results, compared with the year before.
  3. Mix decisions. The next-dollar reads behind these calls are model estimates. The client makes the final decisions.
  4. Ad costs. The change in cost per 1,000 impressions is from the ad platform's own reporting, the 12 months to August 2026 against the 12 months before.

Want numbers like these?

Let's build a measurement-first plan for your brand.

Let’s Chat