The Impact Engine · scoring + attribution

Know which media is moving the business.

Most reports stop at CPM, impressions, and clicks. The Impact Engine scores every dollar against the outcome you hired media to move, then returns one decision: invest more, hold, or reallocate.

The problem

Reports prove the ad ran. They do not prove the business moved.

Every channel produces its own report, in its own currency. All of it proves delivery. None of it answers the question the business is asking: which media moved velocity, intent, and share, and what should we do about it tomorrow.

What most reports close on · currency

CPM. Impressions. Clicks.

CPM prices the buy. Impressions count delivery. Clicks are a platform proxy. They prove the spot aired and the ad served. The next budget review still asks: “did it sell?”

What Diray scores · outcomes

Retail velocity. Purchase intent. Category share.

Retail velocity, units off the shelf. Purchase intent, demand moving before the cart. Category share, the only metric that names winners. The metrics that survive the post-buy meeting.

The inputs

What feeds it

The engine does not start from zero. It starts from four decades of compounding signal.

Spend

Every dollar, every channel, every day. Linear TV, CTV, streaming, YouTube, digital video, retail media.

Ad data

Airings, placements, dayparts, networks, audiences, platforms. The context around every impression.

Retail signal

Sell-through at Walmart, Amazon, Kroger, Target. Orders, revenue, leads, site traffic. The outcome side.

Four decades of buying data

First-party performance history since 1987. The training signal other engines do not have.

The method

How it works: the daily loop

Once a day, every client, every channel. The same loop, scored against your outcome.

THE DAILY LOOP SCORE → ATTRIBUTE → RECOMMEND
STEP_01
Ingest the day.
Every channel’s spend and delivery lands in one place: linear TV, CTV, streaming, YouTube, digital video, retail media. Invalid rows are cleaned before anything is scored.
STEP_02
Score the signals.
Each channel is read on the signals that actually predict movement: daypart performance, network and platform strength, audience alignment, segment quality.
STEP_03
Combine into one number.
The signals are normalized and weighted into the Media Impact Score, a daily 0 to 100 read per channel. Weights are learned per client, not copied from a template.
STEP_04
Attribute with lag logic.
TV moves outcomes across multiple days. Digital answers fast. Attribution respects each channel’s response speed instead of crediting whoever ran last.
STEP_05
Return the call.
Invest more, hold, or reallocate. Every score ships with its why: which factors moved it and what to do about it tomorrow.

The Media Impact Score

One number. And its why.

The score is never left as a number. Every read decomposes into the factors that moved it, and ends in a decision.

Media Impact Score · today

84/100

78th percentile vs category

↑ 6 pts vs last cycle

Media efficiency 88 ↑ 3
Audience fit 82 ↑ 2
Retail velocity 91 ↑ 7
Purchase intent 76 ↑ 1
Category share 68 ↑ 2
Channel mix 80 ↓ 1
Invest more

CTV upper-funnel. Scale 1.4×. Velocity headroom before saturation.

Hold

National linear flat. Efficient reach, no fatigue signal.

Reallocate

Move display retargeting budget to Walmart retail media.

Illustrative cycle · national CPG · scores shown for explanation, not a live client

Named workflow · named human decision

The engine scores. A Diray operator makes the call.

The Impact Engine does not spend money. It returns a recommendation. A Diray operator reviews the score, applies four decades of buying judgment, and makes the decision. The technology amplifies the expertise. It does not replace it.

Context

Where it sits in the measurement world

MMM

Tells you what happened last quarter.

Sales lift studies

Validate what a specific campaign flight drove.

The Impact Engine

Tells you what happened yesterday and what to adjust today. The operational layer between strategic measurement and daily buying.

For every product the engine also runs a baseline with zero media features: outcomes predicted from momentum alone. If media is not outperforming that baseline, the system says so. It never assumes media is always the driver.

The honesty check, built into the model

Next step

Tell us the outcome you need to move.

We will show you which screens, schedules, networks, audiences, and retail signals are moving it.