> 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/account-configuration/attribution-models.md).

# Attribution Models

Attribution models decide how customer journeys are built and how credit for a conversion is split between touchpoints. Manage models and turn on attribution per site.

An attribution model decides two things. First, how Roivenue builds customer journeys from touchpoints. Second, how the credit for a conversion is split between the touchpoints in the journey. Different models give different answers to the question of which channel brought the conversion.

On the **Attribution Models** page, you see the models of your account, create new ones, and choose which sites have attribution calculated.

To open the page, go to **Account Configuration → Attribution Models** in the left navigation.

***

### The list of models

| Column         | What it shows                                                                                                                                                                                          |
| -------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Name**       | The full name of the model, for example **AI Data Driven**.                                                                                                                                            |
| **Short Name** | A short version of the name, used where space is limited, for example on [Model Comparison](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/attribution/model-comparison). |
| **Type**       | The base algorithm of the model.                                                                                                                                                                       |

Click the edit icon next to a model to change its settings, or the delete icon to remove it. To create a new model, click **Add Attribution Model**.

***

### Base algorithms

Every model starts from one base algorithm:

| Algorithm          | How it splits the credit                                                                               |
| ------------------ | ------------------------------------------------------------------------------------------------------ |
| **First Touch**    | All credit goes to the first touchpoint in the path.                                                   |
| **Last Touch**     | All credit goes to the last touchpoint before the conversion.                                          |
| **Linear**         | Credit is split evenly between all touchpoints in the path.                                            |
| **Time Decay**     | Touchpoints closer to the conversion get more credit.                                                  |
| **AI Data Driven** | Roivenue AI gives each touchpoint a fair share, based on patterns that led to conversions in the past. |

Read more in [AI Data Driven Attribution](https://docs.roivenue.com/roivenue-resources/roivenue-training/core-concepts/ai-data-driven-attribution) and [Last Touch vs. AI Data Driven Attribution](https://docs.roivenue.com/roivenue-resources/roivenue-training/core-concepts/last-touch-vs.-ai-data-driven-attribution).

***

### Model settings

When you add or edit a model, you choose the base algorithm and then adjust the settings below.

#### Base Settings

| Setting                          | What it means                                      |
| -------------------------------- | -------------------------------------------------- |
| **Attribution Model Name**       | The full name shown in reports.                    |
| **Attribution Model Short Name** | The short name shown where space is limited.       |
| **Attribution Model Code**       | A unique identifier. Use only letters and numbers. |

#### Attribution Path Settings

These settings decide where one customer journey ends and a new one begins.

| Setting                              | What it means                                                                                                                                             |
| ------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Maximum Days Between Touchpoints** | When more days pass between two touchpoints, a new journey begins. Leave it blank for no limit.                                                           |
| **Maximum Journey Length in Days**   | When a journey lasts longer, a new journey begins. Leave it blank for no limit.                                                                           |
| **Start New Path After Conversion**  | A new path starts after each conversion. Only touchpoints after the conversion get credit for the next one.                                               |
| **Attribute Not Tracked Orders**     | Orders that web tracking missed are modeled from tracked user behavior and attributed as complete journeys. Works only with a connected Transaction Feed. |

#### Path Stitching Settings

Path stitching connects touchpoints of the same user into one journey.

* When **probabilistic path stitching** is on, Roivenue uses an AI model to link interactions across devices and sessions based on all collected parameters. It works only with [Roivenue Measurement](https://docs.roivenue.com/roivenue-resources/implementation/web-data/roivenue-measurement).
* When it is off, only deterministic parameters, such as cookies and user IDs, build the path.

**Required Probability of Match** sets how sure Roivenue must be before it adds a touchpoint to a path. A higher value adds fewer touchpoints, but with higher confidence that they belong there.

#### Custom Touchpoint Weights

The AI Data Driven model calculates the share of each touchpoint. With **Add Rule**, you can change the weight of a touchpoint for a chosen Source/Medium and so influence how much credit it gets.

#### Synthetic Impression Settings

Walled garden platforms, such as Facebook, Snapchat, or TikTok, don't share detailed impression data. Synthetic impressions fill this gap. Roivenue creates them from the conversions and revenue reported by the ad platform and matches them with conversions tracked on your website. See [Click-based vs. Impression-based Touchpoints](https://docs.roivenue.com/roivenue-resources/roivenue-training/core-concepts/click-based-vs.-impression-based-touchpoints).

| Setting                                                                 | What it means                                                                                                             |
| ----------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- |
| **Synthetic Impression**                                                | Turns synthetic impressions on for the model.                                                                             |
| **Maximum Group Size**                                                  | The maximum number of conversions on one ad row that can be matched on exact revenue.                                     |
| **Weak Match for Conversions**                                          | Allows matching for rows with more conversions than the maximum group size, where revenue does not have to match exactly. |
| **Maximum Revenue Difference per Conversion**                           | How much the revenue may differ, for example when the platform receives revenue with tax and your website without it.     |
| **Use not tracked orders for synthetic impression conversion matching** | Includes orders that web tracking missed. Works only with a connected Transaction Feed.                                   |
| **Time Window for Matching Conversions**                                | Extra hours for matching, to cover delays between the conversion and the moment each platform receives it.                |

***

### How to turn on attribution for a site

The **Sites** section at the bottom of the page lists all sites of your account.

1. Switch on the toggle next to each site that should have attribution calculated.
2. Click **Save And Generate Attribution Data**.

The data becomes available 4 to 8 hours after you turn attribution on. Sites without attribution use last touch. The same setting is available as **Calculate Attribution Models** in [Sites](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/account-configuration/sites).

{% hint style="warning" %}
**Changing a model changes your results.** Path, stitching, and synthetic impression settings affect all reports, dashboards, and exports that use the model. Discuss changes with your Client Success Manager (CSM) first.
{% endhint %}

***

#### FAQ

<details>

<summary>1. Which model should I use?</summary>

**AI Data Driven** gives the most balanced view, because it takes all touchpoints into account. **Last Touch** and **First Touch** are useful for comparison. Compare the models side by side in [Model Comparison](https://docs.roivenue.com/roivenue-resources/roivenue-training/roivenue-app/attribution/model-comparison).

</details>

<details>

<summary>2. I turned on attribution for a site, but I don't see the data. Why?</summary>

The data is generated 4 to 8 hours after you click **Save And Generate Attribution Data**. Until then, the site uses last touch.

</details>

<details>

<summary>3. What happens to a site without attribution?</summary>

Its conversions are credited by last touch. The other models are not calculated for it.

</details>

<details>

<summary>4. Should I edit an existing model or create a new one?</summary>

Create a new model when you want to test different settings. You can then compare it with the original one, and your current reports stay the same.

</details>

<details>

<summary>5. What happens when I leave a field blank?</summary>

A blank field uses its default value. For **Maximum Days Between Touchpoints** and **Maximum Journey Length in Days**, blank means no limit.

</details>

<details>

<summary>6. Is this the same as the attribution window in my advertising platforms?</summary>

No. The [attribution window](https://docs.roivenue.com/roivenue-resources/roivenue-training/core-concepts/attribution-window) is set on the platform side and decides which conversions the platform reports. The settings here decide how Roivenue builds journeys from your own data.

</details>

<details>

<summary>7. Why do some settings work only with a Transaction Feed or Roivenue Measurement?</summary>

**Attribute Not Tracked Orders** needs a Transaction Feed to know which orders web tracking missed. Probabilistic path stitching needs the parameters collected by Roivenue Measurement. See [Transaction Data Implementation](https://docs.roivenue.com/roivenue-resources/implementation/connecting-backend-data).

</details>
