Integration Guidance: Predicted Lifetime Value (pLTV)

Updated: Jul 9, 2026

Copy for LLM

Overview

Predicted lifetime value (pLTV) value optimization lets you acquire high-value customers based on your own long-term value predictions. You make these predictions early in the acquisition funnel and send them to Meta, which optimizes delivery toward the users you expect to be most valuable over time.

pLTV is built around a conversion event: this is the event where you acquire the user. This event happens only once per customer and acts as the optimization event for the campaign. For example, if you can predict pLTV once a user starts a free trial, the free trial is your conversion event. You can only associate one conversion event with pLTV per dataset (pixel).

After the conversion event, you must send the pLTV signal within 7 days of sending the conversion event to Meta.

Requirements

To use pLTV value optimization, confirm all of the following:

You have an existing web Conversions API integration, or you are ready to build one. pLTV does not support app events at this time.

You have a pLTV model that is validated in some way, such as one already used for optimization, measurement, or budget decisions.

Your predictions must include at least 5 distinct pLTV values. All values must be positive (not binary or negative), and the highest value should be at least 3x greater than the lowest.

You have a minimum of 100 conversion events per week attributed to Meta for each of the last 4 weeks.

Your dataset is not in Core Setup. Datasets in Core Setup cannot use this product.

The following events support pLTV value optimization:

Purchase

Subscribe

StartTrial

CompleteRegistration

AddPaymentInfo

Custom events

Choose your integration method

There are two ways to send pLTV signals, depending on how quickly your model computes a value:

Real-time integration — use this if you can send the pLTV signal at the same time as the conversion event.

Delayed integration — use this if your model benefits from waiting up to 7 days to compute a more accurate value.

Unlike other Meta products, sending pLTV values faster does not currently improve performance. Choose the method that matches your model, and send the most accurate value you can within 7 days.

Keep the following in mind regardless of method:

Do not delay the conversion event to wait for pLTV. If your model is not real-time, send the conversion event as usual and use the delayed integration to send pLTV later.

You can integrate a pLTV signal with only one event per dataset.

Meta does not recommend creating a new Meta Pixel or a new custom event only for testing.

Send real-time pLTV signals

Use the real-time integration when you can compute pLTV at the moment of the conversion event. Add predicted_ltv and currency to the conversion event’s custom_data. Sending through the Conversions API is required. You can also send the same parameters through the Meta Pixel.

In Events Manager, go to Datasets and select your dataset. Select Value and Currency Metrics dropdown and select the conversion event you intend to optimize for. Valid events will display Set up pLTV with a pLTV icon.

Events Manager pLTV onboarding overview with an expanded event with 'Set up pLTV' action

A screen will display to allow you to select how you will send your pLTV data to Meta. Select With [Event Name] events to send pLTV real-time. Select Start integration to begin the onboarding process.

Events Manager pLTV integration modal displaying the available options to send pLTV data to Meta with the 'with Purchase events' selected

Add pLTV to your existing conversion event (any standard or custom event) in Conversions API (required) and Meta Pixel (optional). It is not recommended that a new custom event is created just for testing purposes.

Conversions API (required)

Include predicted_ltv, value and currency in the custom_data parameter in your Conversions API payload:

{
  "data": [
    {
      "event_name": "Purchase",
      "event_time": 1633552688,
      "action_source": "website",
      "custom_data": {
        "currency": "USD",
        "value": 142.52,
        "predicted_ltv": 678.9
      }
    }
  ]
}

Use the Conversions API Payload Helper to generate and validate a complete payload.

Meta Pixel

If you are sending values in both Meta Pixel and Conversions API, modify your existing event to include the value and currency and predicted_ltv parameters. Note: sending in Conversions API is required, Meta Pixel is optional.

fbq('track', 'Purchase', {
  value: 142.52,
  currency: 'USD',
  predicted_ltv: 678.9,
});

Send delayed pLTV signals with AppendValue

Use the delayed integration when your model benefits from more time to compute an accurate value. Send the conversion event as usual with proper identifiers, then send the pLTV signal within 7 days through a separate event named AppendValue.

Meta matches each AppendValue event back to its conversion event using the identifiers in original_event_data. Send the AppendValue event within 7 days of sending the conversion event to Meta. The window is measured from when you send the conversion event, so if you send events in batches, account for the time already elapsed. For example, if you calculate pLTV on day 5, you have at most 2 days to send the data back to Meta.

A screen will display to allow you to select how you will send your pLTV data to Meta. Select After [Event Name] events to send pLTV delayed. Select Start integration to begin the onboarding process.

Events Manager pLTV integration modal displaying the available options to send pLTV data to Meta with the 'after Purchase events' selected

Set up the AppendValue event following the sections below

Match your events

To match each AppendValue event to its conversion event, provide the following in original_event_data:

event_id and/or order_id (needed for matching): Meta uses these identifiers to find the conversion event that corresponds to your AppendValue event. Send both when possible to improve your match rate, and ensure each value is identical to the one sent on the conversion event.

external_id (optional, but recommended): A user-level identifier that helps Meta associate both events with the same user. Because it identifies users rather than events, it cannot replace event_id or order_id.

If you send the conversion event through both the Meta Pixel and the Conversions API, include event_id and/or order_id as a top-level parameter on both the Meta Pixel and the Conversions API, keep the payloads identical, and send external_id on both if you use it. To add external_id to the Meta Pixel, see advanced matching. For the Conversions API, see External ID.

Do not confuse the top-level event_id/order_id with the event_id/order_id in original_event_data. The top-level values uniquely identify the event and are used for deduplication, while the values in original_event_data are used to match an AppendValue event to its conversion event.

For a conversion event, the top-level event_id/order_id is required, and event_id/order_id in original_event_data should not be provided.

For an AppendValue event, the top-level event_id/order_id is optional (provide it if you need Meta to deduplicate the event for you), and event_id/order_id in original_event_data is required.

AppendValue event parameters

Top-Level Parameter	Nested Parameter	Description
`event_name` string		Required. The standardized name for the delayed pLTV event. Expected value: `AppendValue` Note: This name is standardized. Do not change it.
`event_time` integer		Required. A Unix timestamp in seconds indicating when you calculated the pLTV. It must be later than the conversion event’s `event_time`. Send the `AppendValue` event within 7 days of sending the conversion event.
`event_id` string		Optional. This ID can be any unique string chosen by the advertiser. The `event_id` and `event_name` parameters are used to deduplicate events sent by both web (via the Meta Pixel) and the Conversions API. Note that while `event_id` is marked optional, it is recommended for event deduplication. For deduplication, the `eventID` from a browser must match the `event_id` in the corresponding server event. Learn more about Handling Duplicate Pixel and Conversions API Events. An order number or transaction ID are two potential identifiers that can be used for `event_id`. For example, if a customer makes two purchases on your website with order numbers 123 and 456, each Conversions API call would need to include its respective order number for `event_id`. This allows us to properly distinguish these two purchase events as distinct orders. The two corresponding browser Pixel purchase events would need to also send the same order numbers in the `eventID` parameter for us to understand that there were only two events that took place, not four unique purchases. For other events without an intrinsic ID number, a random number (so long as the same random number is sent between browser and server events) can be used.
`original_event_data` object	`event_name` string	Required. The `event_name` of the conversion event you are appending to, for example `Purchase`.
	`order_id` string	Required for matching (provide `order_id` and/or `event_id`). The `order_id` of the conversion event. Must match the value sent on the conversion event.
	`event_id` string	Required for matching (provide `order_id` and/or `event_id`). The `event_id` of the conversion event. Must match the value sent on the conversion event.
`custom_data` object	`predicted_ltv` float	Required. The predicted lifetime value for this user. Must be a positive number.
	`currency` string	Required. The ISO 4217 three-digit currency code for `predicted_ltv`.
`user_data` object	`external_id` string or list	Optional, but recommended. Any unique ID, such as loyalty membership IDs, customer IDs, and external cookie IDs. You can send one or more external IDs. In order to make sure your `AppendValue` event is resolved correctly to the same user as the original event, it is recommended that you send consistent `external_id` which can help preserve identity, or Meta-generated click ID via `fbp`/`fbc`. Note: Hashing is recommended, but you must have the same method of hashing for both the original conversion event and `AppendValue`, to ensure they match.

Example payload

{
  "data": [
    {
      "event_name": "AppendValue",
      "event_time": 1633984688,
      "action_source": "website",
      "user_data": {
        "external_id": "<HASHED_EXTERNAL_ID>"
      },
      "original_event_data": {
        "event_name": "Purchase",
        "order_id": "<CONVERSION_EVENT_ORDER_ID>",
        "event_id": "<CONVERSION_EVENT_EVENT_ID>"
      },
      "custom_data": {
        "predicted_ltv": 20.1,
        "currency": "USD"
      }
    }
  ]
}

The example above omits other required Conversions API parameters. See the Conversions API Main Body Parameters for the full specification, and use the Payload Helper to validate your payload.

Check your integration health

After you start sending events, you can confirm that your integration is healthy via a number of methods:

In Events Manager, go to Datasets and select the dataset onboarding to pLTV to verify its status.

Events Manager pLTV integration progress widget at 0% with a 'Waiting for conversion events' status and View status button

Select View status to see which step you are on and to review any issue details and recommended fixes. Selecting the displayed issue(s) will show you more details.

Events Manager pLTV integration modal displaying a completed pLTV onboarding step and an onboarding step with an issue that displays a hyperlink of the issue description 'Some of your website AppendValue events are sending invalid pLTV data'

You can also confirm your integration health by toggling the Value and Currency Metrics dropdown, then selecting an event actively onboarding to pLTV and View issues.

Events Manager 'Value and Currency metrics' table displaying the 'AddPaymentInfo' event with a warning triangle indicating it has issues

Once you pass all checks, allow roughly 2 weeks for Meta’s models to train before you start your first campaign.

Test and measure your results

After your integration is healthy and trained, validate performance before you scale:

Warm up your campaigns before starting your first pLTV value optimization campaigns by running a pLTV value optimization for ~7 days to help the model learn and exit the learning phase.

Compare the pLTV campaign against your existing baseline campaign to gauge the impact of optimizing on pLTV.

Read results in Ads Manager by adding the pLTV column for your conversion event, such as Purchase pLTV, through Columns > Customize columns and selecting Results ROAS.

Return on ad spend (ROAS) is comparable only across pLTV campaigns. Comparing ROAS across different optimization settings is not accurate, because each value is calculated differently.

If you use delayed pLTV, wait for the campaign to end plus the number of days your pLTV is delayed. For example, if your pLTV model takes 5 days to compute and send an AppendValue event, wait 5 days after the campaign ends before pulling final numbers.

For campaign setup, budgeting, and scaling guidance, work with your Meta representative.