Attribution · Quick-Start Guide
Prescient AI
/ Attribution · Quick-start guide 2–3 min read
Attribution

Understand how your media drives business performance

Attribution shows how Prescient's models distribute performance across your media mix — comparing modeled results with what each channel reports, so you can see the incremental impact behind every dollar.

Example · Performance Summary — Revenue target
Spend
$257,868
Daily avg.$9,210
Channels6
Reported Revenue
$2,825,981
Daily avg.$100,928
Channels5
MMM Revenue
$1,933,742
Base$1,046,865
Halo$886,877
MMM Revenue %
67.4%
Base36.5%
Halo30.9%
MMM ROAS
7.50
Base4.06
Halo3.44

What you'll find here

Five sections, stacked top to bottom — each answers a different question.
1
Filters & Views
Set target, model, channels and date range — applied across the page.
2
Performance Summary
Headline metrics: Spend, MMM Revenue, Revenue % and ROAS.
3
Base & Halo Flows
A flow diagram of direct and cross-channel contribution.
4
Performance Table
Channels, Tactics and Campaigns — modeled vs reported.
5
Attribution Over Time
How metrics trend across the date range — compare two, or break one down.

When to use this page

Reach for Attribution when you need the real story behind reported numbers.
Compare modeled vs reported
See where Prescient's incremental view differs from platform-reported results.
Analyze channel performance
Rank channels, tactics and campaigns by spend, MMM Revenue and ROAS.
Reveal cross-channel Halo
Find media that creates demand but gets no credit from last-click attribution.
Spot trends over time
Check whether shifts in spend were followed by shifts in modeled performance.

How it works

A natural top-to-bottom workflow — from setup to trend.
1

Select your view

Choose a Target, the models to include, your channels, and a date range. These filters shape every section below — and can be saved as a View.

Example · Filters bar
Target: Revenue Model: All Filters Views May 4 – May 31, 2026
2

Read the Performance Summary

Start with the headline cards for the period (shown above). Metrics adapt to your target — the green cards are Prescient's modeled results, each split into Base (direct) and Halo (cross-channel).

See the MMM Revenue, MMM Revenue % and MMM ROAS cards in the example above.
3

Explore Base & Halo Flows

Follow each ribbon from a media channel on the left to the outcome it drives on the right. Thicker flows mean larger contribution; click any node for a detailed breakdown.

Example · Base & Halo flow
Base (direct) Halo (cross-channel)
4

Compare in the Performance Table

Switch between Channels, Tactics and Campaigns to change the level of detail. Each row lines up modeled metrics against channel-reported ones — expand a row to see its trend and saturation.

Example · Performance Table — Channels
Modeled (MMM)Channel Reported
ChannelSpendMMM RevenueMMM ROASChannel Rev.Ch. ROAS
Total$257,868$1,933,7427.50$800,4583.10
MMeta Ads$117,712$748,4976.36$351,6452.99
aAmazon Ads$59,700$588,6649.86$280,7244.70
tTikTok GMV Max$20,363$302,07814.83$00.00
5

Track Attribution Over Time

Use Compare to plot two metrics together (e.g. Spend vs MMM Revenue), or Breakdown to split one metric by channel or tactic and see what's driving it.

Example · Attribution over time
MMM Revenue Spend
CompareBreakdown
May 10May 17May 24May 31

Saturation Plot

Expand any campaign to see how spend has driven incremental revenue — and where returns start to flatten.

The Saturation Plot uses historical model data to estimate how a campaign's incremental revenue changes at different spend levels — making it easier to spot room to scale, or the point where returns begin to diminish.

Historical Model Confidence band Current spend Scenario spend
Seasonal Forecast CompareAttributionSaturation
$0$450$900$1,350$1,800$2,112
Explore Spend Scenario
Current $393Scenario $393
Current incr. ROAS
2.35
Current avg daily spend
$393
Scenario incr. ROAS
2.35
Est. incremental revenue
+$0/day
Model confidence
Strong

How to read the chart

Historical Model
The green curve — the modeled relationship between daily spend and incremental revenue.
Historical data points
Each gray dot is an observed spend level and the modeled incremental revenue at that spend.
Confidence band
The shaded range shows the model's uncertainty — wider bands mean less certainty.
Current & scenario spend
The dashed line marks your current average daily spend; the slider tests a different scenario.
Seasonal Forecast

Toggle Seasonal Forecast to fold expected seasonal patterns into the curve — so you can evaluate scenarios during periods of unusually high or low demand, not just the flat historical model.

Guidance, not a forecast

The plot is based on historical modeled performance. Use it to weigh budget scenarios — not to predict exact results. Treat scenarios far outside your historical spend as directional only.

Compare scenarios close to your historical spend for the most reliable estimates.

Weigh both incremental ROAS and estimated incremental revenue when changing budget.

Use the confidence band to gauge how certain the model is at each spend level.

Enable Seasonal Forecast when judging spend during peak or off-peak periods.

Two ideas worth knowing

These explain most of the differences you'll see on the page.
Base vs Halo

Base is a channel's direct contribution to an outcome. Halo is its indirect lift — where one channel influences another channel, retailer, or model. MMM Revenue includes both.

Modeled vs Reported

Channel reporting asks "which conversions did this platform claim?" Prescient asks "what incremental impact did this investment have across the business?" Differences are expected — and useful.

Tips

Small habits that make attribution easier to trust.

Compare similar date ranges when reviewing performance changes.

Read modeled and reported metrics together — never one in isolation.

Weigh both Base and Halo — a channel's real value is often in the Halo.

Save filter combinations you use often as Views for one-click access.

Confirm model health before acting on any unexpected finding.

Testing a future budget change? Head to Media Forecaster next.

Prescient AI · Attribution quick-start guide · Modeled through May 31, 2026