Ask AI How to Split Your Ad Budget | A Draft Built on ROAS and Per-Channel Revenue

In the meeting that sets next month's ad budget, the proposal that passes most often is "same as this month." Not because it's right — because assembling the numbers to justify a change is heavy. Reallocation isn't hard to judge; it's hard to prepare.

So hand the preparation to the AI.

Ask: "How should I split ¥500K?"#

Everything below runs on the demo store's sample data, refreshed daily. Have an AI connected to your numbers read per-channel ROAS and saturation, and the draft comes out like this:

RevenueScope MCP🔗 RevenueScopedata connected

You

I have ¥500K/month for ads. How should I split it next month?

Here's a draft from what the numbers can actually support. Criteo: ROAS 6.28, saturation 55% — efficient, with room left to grow. Google Ads: ROAS 1.51, saturation 86% — less efficient and close to its ceiling. Read plainly: weight Criteo heavier, hold or trim Google Ads. Three caveats come attached: ① doubling budget won't double revenue (diminishing returns), ② precision drops where cost-per-session is unknown, ③ expect ±20–30% error on this split. This is a draft, not an autopilot — the final call is yours.📊 ROAS & saturation by channel · last 30 days📊 sample store · demo data

Two measurements do the work. ROAS (revenue per ad spend) captures efficiency; saturation captures whether a channel is approaching the point where more budget stops helping. In the sample store, Criteo runs ROAS 6.28 at 55% saturation while Google Ads runs 1.51 at 86% — one channel efficient with room left, the other neither (sample values as of 2026-08-19; they move daily). The asymmetry is visible in numbers, so a concrete alternative to "same as last month" writes itself.

"Draft" means exactly that#

Notice that the AI's answer arrives with its assumptions and limits pre-attached: doubling budget won't double revenue (diminishing returns); precision falls where acquisition costs are invisible; expect ±20–30% error. These caveats are attached by RevenueScope itself — not the AI being humble.

This is not a tool that reallocates automatically. It's a draft with receipts, for a human to decide on — which is why you can carry it straight into the meeting. The answer to "why weight Criteo?" comes bundled with the measurements.

If you haven't run ads yet, start one step earlier with where your first ad money should go. The exit-side decision lives in what happens if you stop your ads.

RevenueScope keeps your allocation evidence current, every month

Once your ad spend is in RevenueScope, per-channel ROAS lines up on the same screen as revenue, and the monthly allocation meeting's prep work becomes one question. Saturation, though, is not something RevenueScope estimates on its own: the only place a value is supplied is the demo store's sample data, so the split returned for your own site is ROAS-weighted with no saturation cap on it.

The numerator of ROAS is RevenueScope's own measured revenue, not the conversion value an ad platform reports — which is why it differs from the ROAS in your ad console by definition. The platform's impressions, clicks and reported conversions sit in the same table, so you can hold the two side by side and carry the draft — ±20–30% band included — into your monthly allocation decision.

In the sample store, decide with the AI which channel the ad budget should go to

FAQ#

Q. Does this require connecting ad spend?

The ROAS-based split turns on once your ad spend is in. Before then, you get an even three-way starter draft with its assumptions attached, and it shifts to ROAS-based allocation as your data builds. A saturation cap only applies where a value is supplied — the demo store.

Q. What is saturation?

An indicator of whether a channel's growth is starting to flatten as you add budget. High ROAS with high saturation reads as "efficient, but hard to grow further."