Skip to main content

​
Overview

Budget Recommendations help you decide how to allocate spend across channels for the next planning period. The recommendations are based on:
  • the budget planning source of truth for the upcoming period
  • your approved budget for the upcoming period
  • the latest model’s response curves
  • saturation behavior by channel
  • channel guardrails and business constraints
The goal is not just to show what performed well historically. The goal is to recommend where the next dollar is most likely to create incremental value.

​
How the workflow works

In most accounts, the workflow looks like this:
  1. A total monthly budget is entered in Budget Planning for the upcoming period
  2. An operator reviews or approves that budget
  3. Channel guardrails define where spend is allowed to move
  4. The latest model run provides response curves and saturation signals
  5. Provalytics recommends an allocation across channels within those constraints
That means the recommendation is not a free-form suggestion. It is built from a defined spend envelope plus the rules and evidence available for that client.

​
What guardrails mean

Guardrails define what the optimizer is allowed to recommend for each channel. Common guardrails include:
  • Minimum spend: keep a channel active even if it is not the top short-term efficiency option
  • Maximum spend: prevent over-scaling when a channel is approaching saturation or should stay capped
  • Locked spend: hold a channel fixed so the remaining budget can be reallocated around it
Guardrails matter because they let the recommendation reflect business reality, not just math.

​
What the recommendation is optimizing

Budget Recommendations use the latest modeled response curves to estimate how channels convert additional spend into incremental outcomes. In plain English:
  • channels with room to scale may get more budget
  • channels near saturation may get less incremental allocation
  • channels with strong contribution and strong efficiency may remain priority channels
  • channels with halo effects may still matter even when their direct efficiency looks weaker

​
How to read the output

Read each recommendation as a planning signal, not as a verdict on the channel. Pay particular attention to:
  • Suggested spend: the recommended budget for that channel
  • Expected incremental outcome: what the model expects that spend level to produce
  • Percentage change vs. current spend: the reallocation signal, not a quality score

​
Important interpretation rule

A large suggested increase does not automatically mean “best channel.” It can also mean:
  • the channel is currently small
  • the channel has room to test
  • the model sees efficient headroom before saturation
Likewise, a recommended decrease does not automatically mean a channel is unimportant. It may mean:
  • the channel is already heavily funded
  • marginal returns are flattening
  • another channel has better room to grow within the same budget

​
How to use recommendations well

Use Budget Recommendations to support:
  • monthly planning conversations
  • finance reviews
  • channel reallocation discussions
  • test-and-learn planning
The best way to use the output is alongside:
  • incremental contribution
  • ROAS or CPA
  • channel role in the funnel
  • known halo effects
  • business constraints the optimizer cannot infer on its own

​
Budget Recommendations vs Scenario Planner

Budget Recommendations answer:
What does the model think is the best allocation within the current rules?
Scenario Planner answers:
What happens if we choose a different allocation on purpose?
Use recommendations when you want the model’s best constrained answer. Use Scenario Planner when you want to test alternatives before making a decision. For most teams, Budget Planning comes first, Recommendations comes second, and Scenario Planner comes third.