> For the complete documentation index, see [llms.txt](https://docs.roivenue.com/roivenue-resources/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/reports/budget-optimizer.md).

# Budget Optimizer

Budget Optimizer recommends how to split your weekly marketing budget across channels or campaigns, helps you plan a budget change, and shows how confident each estimate is.

Budget Optimizer looks at how your spend and results moved together week by week and fits a response curve for every channel, campaign, or other dimension value. From the curves, it estimates where extra spend still pays off and where it no longer does. It then suggests a new weekly budget split and tells you how confident it is about each row, so you can decide how far to follow it.

Use Budget Optimizer when you want to:

* find out which channels should get more budget and which should get less,
* split an extra amount, or a cut, between selected channels,
* check if the same total budget brings more when you split it differently.

The page has five parts:

1. **Header:** the period the curves are fitted on, currency selector, attribution model selector, and **Optimizer settings**.
2. **Tabs:** **Recommendations**, **Planning**, and **Redistribute**, three ways to plan the budget.
3. **Controls and KPI cards:** site, goal, dimension, and budget, with a summary of the current and the planned weekly spend and results.
4. **Plan table:** one row per dimension value with a fitted curve, its current and recommended spend, and the expected effect.
5. **Not Estimated:** dimension values the optimizer could not model, with the reason for each.

To open the page, go to **Reports → Budget Optimizer V2** in the left navigation. All amounts are weekly. **Current Weekly Spend** is the spend of the last complete week. The chip next to the page title shows the fit window, for example "Fitted on 26 weeks of data, 27 Apr – 25 Oct 2026". The results use the attribution model selected in the header, so the same channel can get a different recommendation under a different model.

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FOiyLOylgf224oCYNbJxV%2F01-budget-optimizer-overview-v2.png?alt=media&amp;token=4bb29cda-6943-4c7e-9c13-ed6e86de61b0" alt=""><figcaption></figcaption></figure>

***

#### How to set up the plan

The bar above the KPI cards has these controls.

| Control               | What it does                                                                                                                                                                                                             |
| --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Site**              | The site the plan is calculated for.                                                                                                                                                                                     |
| **Optimization Goal** | What the optimizer maximizes. See the table below.                                                                                                                                                                       |
| **Dimension**         | The level you plan on, for example **Platform Name**, **Campaign** or **Channel Name**. Each dimension value gets its own curve.                                                                                         |
| **Budget**            | **Keep current total** moves money between rows and keeps the total. **Increase budget** and **Decrease budget** plan a bigger or a smaller total. **No limit** lets the optimizer add spend wherever it still pays off. |

The goal decides what the optimizer maximizes.

| Goal                     | When to use it                                                                                                 |
| ------------------------ | -------------------------------------------------------------------------------------------------------------- |
| **Max marketing profit** | You want the most profit from marketing. Spend goes where it still earns more than it costs.                   |
| **Balanced**             | You want a compromise between profit and revenue growth.                                                       |
| **Max net revenue**      | You want growth. Spend goes where it brings the most revenue, even when the profit per unit of spend is lower. |

These three goals apply when the account optimizes for revenue and profit. Accounts without conversion values, for example, lead-based businesses, switch the **Optimization objective** in the settings to **Conversions**. The optimizer then plans for the most conversions, and the cards and columns show conversions and CPA instead of revenue.

{% hint style="info" %}
**Tip:** Start with **Platform Name** and **Keep current total**. Once the platform level makes sense, switch the dimension to **Campaign** to see the plan in more detail.
{% endhint %}

***

#### How to read the Recommendations tab

The KPI cards sum up the plan:

* **Current Weekly Spend**: the spend of the last complete week, with the share that the optimizer could model.
* **Recommended Weekly Spend**: the total the plan suggests, with the difference against the current spend.
* **Estimated Net Revenue** and **Estimated Marketing Profit**: the expected weekly result at the recommended spend, with the difference against the current state.

Hover over a card to see the current value, the estimate, and the difference.

The table has one row per dimension value:

| Column                | What it shows                                                                                    |
| --------------------- | ------------------------------------------------------------------------------------------------ |
| **Spend**             | The current weekly spend.                                                                        |
| **mROAS**             | Marginal ROAS: how much net revenue one extra unit of your currency brings at the current spend. |
| **Recommended Spend** | The weekly spend the plan suggests, with the change in your currency and in %.                   |
| **Est. Revenue**      | The expected net revenue at the recommended spend, with the change and a ± range.                |
| **Est. Profit**       | The expected marketing profit at the recommended spend, with the change and a ± range.           |
| **Confidence**        | How sure the optimizer is about the curve, from 1/4 (Low) to 4/4 (Very High).                    |

The icon before the name shows what the plan does with the row, for example **Invest more** or **Cut Spend**. Hover over it to see the label.

Every estimate has a ± range. It combines the uncertainty of the fitted curve with the week-to-week noise in the row's results. The range does not get smaller when the change gets smaller, so a wide range on a small change is expected.

Follow high-confidence rows first. Treat low-confidence rows as a hint to test, not as a plan.

***

#### How to check a single row

Click **Chart** or the info icon next to the name to open the detail of a row. It shows:

* the current and the recommended weekly spend and the estimated result,
* **Marginal return at current spend** and **Marginal return at recommended**, which is what one extra unit of your currency brings at each spend level,
* **Confidence** of the fit,
* the response curve with the weekly observations, the current, and the recommended spend, on the tabs **Net revenue**, **Gross profit**, and **Marketing profit**,
* **Weekly data points**, the weeks the curve is fitted on.

In **Weekly data points** you can change how much a single week counts:

* **Weekly weight** comes from the settings and is read-only here. Recent weeks count more than older ones.
* **Weight** gives one week more or less influence on the curve, from 0 to 5.
* **Exclude** leaves the week out of the fit, for example a big sale or a tracking outage.

Click **Save For Account** to keep the change, or **Discard** to drop it. Saved weights apply to everyone in the account.

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FC2g6qn1WI3xkIdbOuh79%2F02-row-detail.png?alt=media&amp;token=9f6cf7b7-f415-4251-b808-98edbd6a4960" alt=""><figcaption></figcaption></figure>

To leave a row out of the plan, click the cross at the end of the row. The row moves to **Not Estimated** with the reason "Manually excluded." and keeps its current spend. To bring it back, use **Restore to the plan**.

***

#### How to plan an extra amount or a cut

Use **Planning** when you know how much you want to add or remove:

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FS8F40rOZ8yFc04pnxtGV%2F03-planning-v2.png?alt=media&amp;token=51cd22e9-a273-49b4-b1e1-e0d9734e5102" alt=""><figcaption></figcaption></figure>

1. In **Dimension values**, pick the rows you want to work with.
2. Choose **Invest More** or **Save Budget**.
3. Enter the **Amount** in % or in your currency.

The optimizer splits the amount between the selected rows so that it brings the most for your goal. The other rows keep their current spend. The spend limits from the settings still apply, so the plan can place less than you asked. In that case it tells you how much stays unallocated.

***

#### How to redistribute the same budget

Use **Redistribute** when the total budget stays the same:

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FEVaCOgaIVhbh2JXBekSr%2F04-redistribute-v2.png?alt=media&amp;token=e162867b-00df-47ea-b1ca-69418b42bffc" alt=""><figcaption></figcaption></figure>

1. In **Dimension values**, pick two or more rows.
2. The optimizer re-splits their combined spend, shown in **Total Budget to Redistribute**, between them for the best result.

The total does not change. Only the split between the selected rows does.

***

#### Why a row is in Not Estimated

Some dimension values do not have enough reliable data for a curve. The optimizer lists them under **Not Estimated** with the share of spend that is not modelled and a reason for each row.

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FfBYEAR9MrUaXrrhN9Ecj%2F06-not-estimated.png?alt=media&amp;token=c1ed1f3f-096c-40b7-a520-5997df8613ba" alt=""><figcaption></figcaption></figure>

| Reason                             | What it means                                                                                           |
| ---------------------------------- | ------------------------------------------------------------------------------------------------------- |
| No investment over the last period | The row had no spend in the fit window, for example **Direct** or **Organic**.                          |
| Spend barely changed               | Weekly spend was almost the same every week, so the optimizer cannot see how results react to a change. |
| Too volatile to estimate           | Results moved from week to week in ways the spend does not explain.                                     |
| Flat response                      | The curve does not rise with spend, so it does not show what more spend would buy.                      |
| Below the required confidence      | The fit is not reliable enough to recommend a change.                                                   |
| Manually excluded                  | Someone removed the row from the plan.                                                                  |

Rows that cannot be modelled reliably can still be marked **cut only**. The optimizer uses a straight line through their average return. It never adds budget to them, but it can recommend a cut when they return too little. You control this with **Allow spend cuts on unmodeled values** in the settings.

What you can do:

* Wait until the row collects more data.
* Change its spend for a few weeks, so the optimizer can see how results react.
* Merge small rows into a bigger group, for example with a custom dimension. See [Data](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/data).

***

#### How to change the optimizer settings

Click **Optimizer settings** (the gear icon) in the header. The settings are shared across the account, so everyone sees the same setup. Only a Contributor can change them.

<figure><img src="https://4279879158-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F-MGvvrlVUvq4TNL0oYP6%2Fuploads%2FPqiu2z1OLqE3pIbWs2Wu%2F07-optimizer-settings.png?alt=media&amp;token=5f7faecf-c7b4-49e2-805b-110cfd1e1e49" alt=""><figcaption></figcaption></figure>

| Setting                                  | What it does                                                                                                                                                                                                                                                          |
| ---------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Optimization objective**               | **Revenue & Profit** for e-commerce with purchase conversions. **Conversions** for lead-based businesses.                                                                                                                                                             |
| **Limit maximum CPA**                    | Only for **Conversions**. Applies when **Budget** is **No limit**. Spend grows until one more conversion would cost more than the limit. **Automatic** uses 1.5× the site's current blended CPA. You can also set the same target for all sites or a target per site. |
| **Allow spend cuts on unmodeled values** | When on, rows that cannot be modelled can have their spend cut to free budget for others. When off, their spend stays unchanged.                                                                                                                                      |
| **Limit recommended spend increase**     | **Maximum increase (%)** caps how much the optimizer can raise a row's weekly spend. The default is 25%. The base is either **Based on your usual spend** (the last 10 weeks) or **Based on your last spend**. Decreases are not limited.                             |
| **Weeks used for curve fitting**         | How many recent complete weeks the curves use, from 5 to 100. The default is 26. More weeks give a broader history, fewer weeks follow recent behaviour more closely.                                                                                                 |
| **Period weight overrides**              | **Weight of the oldest week** sets how much the oldest week counts compared with the latest one. The table below lets you change the weight of a week or exclude it for all rows at once.                                                                             |

Click **Save** to apply the changes, or **Cancel** to leave without them. **Reset All Settings To Default** puts back the account defaults and clears the week weights and exclusions saved on single rows. It keeps each user's own goal, budget, and removed rows.

{% hint style="info" %}
**Tip:** Exclude a week in the settings when it was unusual for the whole account, for example a tracking outage. Exclude it in the row detail when it was unusual only for one channel, for example a one-off campaign.
{% endhint %}

***

#### How to work with the recommendations

Budget Optimizer gives statistical recommendations. It does not know about everything that affects your results, such as a new creative, a competitor's campaign, stock levels, or seasonality outside the fit window. You decide how to turn a recommendation into a real budget change.

1. Set the goal and check the settings.
2. Check confidence and the ± range. Start with high-confidence rows.
3. Check the row's curve in the detail. Make sure no unusual week drives it.
4. Validate the channel's role in [Model Comparison](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/attribution/model-comparison). Check that it looks similar across attribution models.
5. Change the budget step by step and write down what you changed and when.
6. Evaluate the results after a week or two. Compare them with your goal.
7. Repeat. Keep what worked and adjust what did not.

{% hint style="info" %}
**Tip:** Large one-time changes make it hard to see what worked. Smaller steps give the optimizer new data at different spend levels, so its next recommendations get more confident.
{% endhint %}

***

**FAQ**

<details>

<summary>1. Why is a channel missing from the recommendations?</summary>

It is probably in **Not Estimated**. Check the reason next to it. Most often the channel had no spend in the fit window, its spend did not change enough, or its results vary too much to fit a curve. A channel can also be missing because someone excluded it from the plan.

</details>

<details>

<summary>2. Why does a recommendation have low confidence or a wide range?</summary>

The channel's results vary a lot at similar spend, or it has few weeks of usable data. The optimizer then cannot tell precisely how results react to spend. Treat the recommendation as a direction to test in small steps.

</details>

<details>

<summary>3. Which goal should I choose?</summary>

Choose **Max marketing profit** if every unit of spend has to pay for itself. Choose **Max net revenue** if growth matters more than profit per unit of spend. **Balanced** sits in between. If your account has no conversion values, switch the **Optimization objective** in the settings to **Conversions**.

</details>

<details>

<summary>4. What is the difference between Recommendations, Planning and Redistribute?</summary>

**Recommendations** plans the whole budget for your goal. **Planning** splits an extra amount or a cut between rows you pick. **Redistribute** re-splits the current spend of rows you pick and keeps their total.

</details>

<details>

<summary>5. Why does the optimizer recommend less spend on a channel that performs well?</summary>

A channel can have a good average return and still be close to saturation. The optimizer looks at the return on the next unit of spend, the marginal return, not the average. If the same money brings more elsewhere, it moves part of the budget there.

</details>

<details>

<summary>6. Why can't the optimizer raise a channel more?</summary>

**Limit recommended spend increase** in the settings caps how much a channel can grow in one plan, 25% by default. Increase the limit if you want bigger moves. A channel with no spend in the fit window gets no recommendation at all.

</details>

<details>

<summary>7. Should I exclude a week with a big promotion?</summary>

Yes, if the week does not represent normal performance. Exclude it in the settings for the whole account, or in the row detail for one channel. You can also lower its weight instead of excluding it completely.

</details>

<details>

<summary>8. Why does the recommendation change when I switch the attribution model?</summary>

The optimizer fits the curves on the results of the selected model. Each model credits channels differently, so the curves and the plan change too. Use the model you report on, and compare the plan under another model in [Model Comparison](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/attribution/model-comparison).

</details>

<details>

<summary>9. What is saved for the whole account and what only for me?</summary>

The optimizer settings and the week weights are saved for the whole account, so everyone gets recommendations from the same curves. Your goal, budget, and the rows you removed from the plan are kept per user.

</details>

<details>

<summary>10. Why can't I change the settings?</summary>

Only a Contributor can change the optimizer settings for the account. Other users can read them. Ask a colleague with the Contributor role, or contact Roivenue support.

</details>
