Average ROAS tells you what's working in aggregate — it doesn't tell you where the next dollar will generate the most return. The agent builds a marginal ROI model for each active channel and campaign by analyzing the relationship between incremental spend increases and incremental pipeline generation over rolling time windows. Channels that show diminishing returns at current spend levels are identified even when their average ROAS is still healthy. Channels with capacity for more efficient spend — where CAC is still declining as spend increases — are flagged as under-allocated. The model updates continuously as new performance data arrives, so allocation recommendations reflect current platform dynamics, not last quarter's equilibrium.

The gap between a performance insight and an actionable budget decision is often where value gets lost. The agent doesn't just identify over- and under-allocated channels — it generates a specific allocation recommendation with a projected pipeline and revenue impact. Each recommendation shows the current allocation, the recommended allocation, the spend delta, and the model-projected change in pipeline contribution and CAC. Recommendations are ranked by projected revenue impact so budget decisions can be prioritized. Scenario modeling allows teams to explore alternative allocations — for example, what happens to projected revenue if a channel is cut entirely or budget is added from a new source.

Budget allocation decisions are only as good as the execution that follows them. The agent monitors actual spend pacing against the approved allocation targets in real time, flagging deviations before they become significant. If a campaign is trending toward overspend in the first week of the month due to bid volatility or audience size changes, the alert fires with a specific recommended bid adjustment or budget cap change — not after the overspend has happened. Underpacing campaigns are flagged with the same urgency because underspend means missed pipeline opportunity. Pacing alerts are channel-specific and include the estimated impact on end-of-month pipeline if the current pace continues.

Budget allocation in a vacuum ignores the business context that should shape paid investment. The agent incorporates pipeline coverage data from the CRM to adjust allocation recommendations when the pipeline gap is large — a coverage ratio below 3× generates a signal to increase paid investment rather than optimize toward efficiency. Seasonal patterns in historical performance data are incorporated into projections, so allocations in Q4 account for the different competitive dynamics and buyer behavior of that period. Upcoming events, product launches, or campaign sprints flagged by the marketing team are factored as modifiers to the baseline model output. Allocation is a continuous optimization process, not an annual planning exercise.