One channel has the weakest ROAS in the account. Finance wants savings. UA needs to name the first budget to remove.

So the channel moves to the top of the cut list.

That is a reasonable place to investigate. It is not yet a decision. The same weak row can describe a reporting shift, a temporary break, an intentional volume trade-off or a genuinely removable upper spend layer.

Low ROAS nominates the suspect. Triage decides what kind of suspect it is.

First ask what actually became worse

CHANNEL VIEW

Conversions and revenue credited to this channel.

BUSINESS VIEW

Total paid sales, revenue or contribution in the same market and period.

If channel credit fell while total paid sales stayed stable, the immediate problem is not proven business loss. A platform, attribution window or channel interaction may have moved credit between rows.

Route: REPORTING SHIFT. Reconcile where the conversions moved with the credit-versus-business-loss check. If current spend is already unacceptable, use a temporary risk cap while the comparison is repaired.

Then ask when it became worse

Did completed sales break suddenly after a release, campaign change, creative rollout, store issue or market event?

That is an incident question. Route: SUDDEN INCIDENT. Do not hide a sharp break inside a three-month efficiency average. Decompose and localize it with Acquisition Incident Compressor.

If the weakness persisted across normal campaign cycles and no single break explains it, continue.

Name what the channel is paid to do

Below portfolio-average ROAS is not an economic role.

Some channels maximize near-term contribution. Others intentionally buy additional volume, enter a market or create learning. Those roles do not excuse unlimited losses. They require an explicit boundary.

THE MISSING CONTRACT

What result must this channel create?
What is the maximum acquisition cost or minimum contribution—and when does learning or market-entry spend end?

If nobody can answer, route: ECONOMIC BOUNDARY MISSING. The channel cannot be called inefficient until the business says what it is allowed to buy.

Check whether weakness lives at the top of spend

PATTERNWHAT IT MEANSNEXT ROUTE
Weak across the rangeThe channel remains below its economic boundary at lower and higher spend.Bounded reduction or pause test.
Productive core, flat topCompleted sales stop rising when spend enters the same upper range.Saturation candidate.

A channel can have acceptable average ROAS while the last 10–20% adds almost no completed sales. It can also have weak average ROAS without showing any saturation pattern.

If completed sales repeatedly stop rising in the same upper range, the next question is no longer why the whole channel looks weak. It is whether that upper layer can be removed. Continue with Can You Cut 15% From This Channel Without Losing Sales?

ONE SUSPECT CHANNEL / ONE NEXT INVESTIGATION

What kind of problem is this?

Do not choose the percentage yet. Choose the question that must be answered first.

01 Did total paid sales or contribution weaken too — not only this channel's credited result?

One channel should leave triage through one route

ROUTEWHAT WAS LEARNEDFIRST CHECK
Reporting shiftCredit moved without confirmed business loss.Reconcile where conversions moved.
Sudden incidentThe loss begins at a change point.Decompose and localize.
Boundary missingThe team has not defined acceptable economics.Name the result and stop rule.
Saturation candidateThe core works while the upper range flattens.Test the suspected edge.
Persistent weaknessThe channel is weak across the observed range.Run a bounded reduction or pause.

That is the output of Channel Spend Triage: one reason code, one first check and one condition that would change the route. The percentage comes later.

Evidence notes

The routing logic combines the existing attribution-versus-business-loss method, Acquisition Incident Compressor and bounded channel-overinvestment research. Saturation is used only when additional business repeatedly flattens at the upper spend range.