> ## Documentation Index > Fetch the complete documentation index at: https://docs.getprova.com/llms.txt > Use this file to discover all available pages before exploring further. # Recommendations > Learn how to use Recommendations to review model-led channel budget shifts, scenario views, and allocation detail. ## Overview Recommendations is the model-led budget allocation page. It helps answer: > Given the current model, budget level, and constraints, where should spend go next? This page turns the modeled response system into a usable channel-allocation view for planning and reallocation. ## What the page looks like in practice Recommendations is built in two layers: ### Rollup and recommended mix At the top, the page shows: * scenario budget * projected efficiency * projected KPI output * scenario budget options around the current plan * a channel rollup view with the recommended allocation mix Recommendations overview ### Budget scenario range Each time the platform creates a new recommendation and associated forecast, it also creates six additional budget-and-forecast levels around the current plan. That gives the team seven optimization views in total: * `Flat` * `+10%` * `+20%` * `+30%` * `-10%` * `-20%` * `-30%` Recommendations budget scenarios ### Adjustable spend controls Lower on the page, users can inspect or adjust channel-level spend inputs directly through sliders. This is where the recommendation becomes operational and comparable against current spend levels. Recommendations spend controls ## What you can do In Recommendations, you can: * choose a KPI * review scenario-based budget views * compare recommendation outcomes across budget levels * inspect channel rollups * drill into detailed recommended allocation rows * export both rollup and detail views * move into [Opportunity Days](/using-provalytics/opportunity-days) for timing ## What the page is showing Recommendations is designed around three layers: ### Scenario budget context The page shows the selected scenario budget and, when configured, projected efficiency and projected total KPI output. Those scenario levels are intentional. In real planning environments, budgets are often mostly fixed, but marketers still face surprise changes, especially near decision deadlines. Sometimes incremental budget becomes available late. Sometimes budget is reduced with little warning. Instead of forcing users to request a brand-new model run for every small budget change, Provalytics keeps these seven optimization views ready and refreshed each time new data updates the platform. ### Channel rollup This is the high-level answer to: > How is the recommendation reallocating budget across channels? At this level, the page shows: * current daily spend * max daily spend * the direction and size of the recommended change Recommendations channel rollup detail ### Recommended budget detail This is the lower-level view showing where recommended spend is being allocated inside the recommendation output. This is another meaningful Provalytics differentiator. Recommendations do not stop at the channel level. The platform can carry the recommendation down to the same level of granularity used in the underlying data. In many accounts, that means a hierarchy such as: * channel * campaign * ad group * creative That means the output can show not only recommended spend and max daily spend by channel, but also how the recommendation is being distributed within the actual working structure of the media plan. Recommendations detailed allocation ## How to interpret it well Recommendations is best used when: * you need a model-led answer within current constraints * you want to compare budget scenarios * you are deciding where additional or reduced spend should go This makes the page especially useful when leadership asks questions like: * What if we get 10% more budget? * What if we need to cut 20% immediately? * If the budget stays flat, what is the best constrained allocation right now? ### A practical reading rule Treat Recommendations as a planning signal, not a command. The page is strongest when read together with: * [Proof Report](/using-provalytics/proof-report) * [Incrementality Report](/using-provalytics/incrementality-report) * [Campaign Performance](/using-provalytics/campaign-performance) * [Spend Headroom](/using-provalytics/spend-headroom) That sequence helps connect: * model trust * business impact * current efficiency * next budget move ## Recommendations vs Scenario Planner Recommendations answers: > What is the model’s best constrained allocation view right now? Scenario Planner answers: > What happens if we intentionally try a different plan? Use Recommendations for the model-led answer. Use Scenario Planner for deliberate what-if testing. ## Important interpretation note If a channel receives a lower recommendation, that does not automatically mean it is unimportant. It may mean: * the channel is near saturation * another channel has better room to scale * the chosen scenario shifts the allocation logic