Skip to main content
Docs
Feature Requests Onboarding Trust Center Status Page

Community

Connector Improvement: Facebook Marketing API / Insights API: Conversion Lift Measurement

Planned
David Mor España User

Dear Fivetran Team,

We are writing to request a feature to integrate data from Facebook Marketing API Lift Studies into the Fivetran Facebook Ads connector. We believe this enhancement would significantly benefit your users, including ourselves.

Our Use Case:

We utilize Facebook Marketing API Lift Studies (https://developers.facebook.com/docs/marketing-api/guides/lift-studies/) to measure the incremental impact of our advertising campaigns. Accessing this data within Fivetran would streamline our analysis workflow considerably. Currently, we have to extract and process this data separately, which is inefficient.

Impact and Benefits:

Integrating Lift Studies data into Fivetran's Facebook Ads connector would provide several key benefits:

  • Improved Efficiency: Centralizing all Facebook Ads data, including Lift Studies results, within Fivetran eliminates the need for manual data extraction and processing, saving us significant time and resources.
  • Enhanced Analysis: Having Lift Studies data directly in our data warehouse, alongside other Facebook Ads metrics, would allow for more comprehensive and insightful analysis. We could easily combine Lift Study results with campaign performance data, audience demographics, and other relevant information.
  • Better Decision-Making: By providing a unified view of our Facebook advertising efforts, including the measured lift, Fivetran would empower us to make more informed decisions about budget allocation, campaign optimization, and overall marketing strategy.
  • Competitive Advantage: For Fivetran, offering this integration would provide a competitive advantage by catering to the growing need for robust marketing measurement solutions. Many advertisers are increasingly relying on Lift Studies to prove the effectiveness of their campaigns.

Specific Requirements:

Ideally, the integration would include the following:

  • Dedicated Tables: Creation of new tables or enhancement of existing ones within the Facebook Ads connector to specifically house Lift Study data.
  • Key Metrics and Dimensions: Access to the relevant metrics and dimensions available in the Facebook Marketing API for Lift Studies, including (but not limited to) test/control group indicators, incremental conversions, reach, and other lift metrics.
  • Documentation: Clear documentation on how to access and utilize the Lift Studies data within Fivetran.

We believe that integrating Facebook Marketing API Lift Studies data would be a valuable addition to Fivetran's capabilities, enabling users to gain a deeper understanding of their advertising effectiveness and drive better business outcomes.

Thank you for considering our request. We look forward to your response.

Sincerely,
David

Please sign in to leave a comment.

Comments

5 comments

Date Votes
  • Official comment
    Lucas Alexander Fivetranner

    Hi David,

    Luke from the Product team here. Thank you for submitting this request!

    Before we implement this, I want to clarify a few details to ensure we represent this data correctly in the schema. Here is my understanding of how this data is structured:

    Each ad_account can have many ad_studies. We can retrieve the ad_studies associated with an ad_account using the ad_studies edge on the ad_account endpoint (source). Each ad_study can have one or many objectives. We can retrieve the list of objectives from the objectives edge on ad_study (source). Once we have the list of study_objective_ids we can retrieve a list of results for each study_objective_id (source). We can also use the cell_id breakdown to get results broken down by cell_id (e.g. control vs. test) - source. We can retrieve results for a specific date; however, the date must be within the prior 30 days. 

    So I anticipate we'll need the following tables in our ERD:

    • AD_STUDY - This would have a FK relationship to the existing ACCOUNT_HISTORY table. 
    • OBJECTIVE - This would have a FK relationship to AD_STUDY
    • OBJECTIVE_RESULT - This would have a FK relationship to OBJECTIVE. It would be broken down by date and cell_id. It would represent the results for each date, cell_id, and objective. 

    This is an example of the results we'd provide in OBJECTIVE_RESULT. This is based on the sample provided in this doc:

    	"population_test":2334212,
	"population_control":123407,
	"population_reached":1862084,
	"impressions":19020874,
	"spend":26059,
	"conversions_control_raw_scaled":110918.27003534,
	"conversions_exposed":86961.044050743,
	"conversions_raw_pValue":0.12863848309723,
	"conversions_test":104412.89695396,
	"conversions_control_scaled":104575.81331581,
	"conversions_incremental":-162.91636184894,
	"conversions_notExposed":87123.960412592,
	"conversions_confidence":0.69291721817069,
	"conversions_multicell_confidence":null,
	"conversions_incremental_lower":-3470.6251396487,
	"conversions_incremental_upper":3235.0644420632,
	"conversions_multicell_rank":null,
	"conversions_incremental_share":-0.001873440730011,
	"conversions_CPiC":-159.95324044961

    Does this align with your expectations?

    Thanks,
    Luke

  • David Mor España User
    • Edited

    Hi Luke,

    First of all, thanks so much for looking into this! I really appreciate it.

    Yes, that aligns with my expectations.

    The only thing missing would be something that can be retrieved from the https://graph.facebook.com/v23.0/ endpoint.
    and the most important fields are:
    - id: The unique ID of the ad study cell.
    - name: The name of the cell (e.g., “Test Group”, “Control Group”).
    - description: A description of the cell
    - treatment_percentage: The percentage of the audience in this cell that receives treatment (i.e., sees ads). For a control group, this will be 0.
    - control_percentage: The percentage of the audience that is held out from seeing ads. For a test group, this would typically be 0, and for a control group, it would be the percentage of users not exposed. Note: Facebook documentation states that treatment_percentage for each cell should be at least 10, and the sum of treatment_percentage for all cells should be less than or equal to 100
    - adaccounts: A list of ad account IDs associated with this cell. These are the ad accounts whose ads are being included in this specific test or control group
    - campaigns: A list of campaign IDs associated with this cell
    - adsets: A list of ad set IDs associated with this cell.

    Could that be included?

    Thanks!

  • Abdullah Tahir User

    Hi Fivetran Team,

    We’re writing to add our support for this feature.

    We rely on Meta Lift Studies to measure marketing performance, and this data feeds directly into our internal Forecasting models.

    Currently, without access to ad_study_objective results (incremental conversions, iROAS, and test/control cell metrics), we have to retrieve these manually from clients. This adds operational overhead and limits our ability to scale workflows efficiently.

    If the proposed schema (AD_STUDY, OBJECTIVE, OBJECTIVE_RESULT, and AD_STUDY_CELL) were ingested, it would allow us to centralise lift test data alongside other metrics already flowing through the connector. As Meta Lift Studies become increasingly important for advertisers, supporting this data natively would strengthen Fivetran’s connector, particularly for teams focused on marketing measurement and incrementality.

    Could you share any guidance on where this addition sits in your roadmap, or whether it may be considered in an upcoming planning cycle? Even high-level insight would be greatly appreciated.

    Thank you for your time and consideration,
    Abdullah

  • Lucas Alexander Fivetranner

    Hi David and Abdullah,

    We're working on this feature now. In order to complete it, we need source access to validate the API response. I sent both of you access requests via our support team.

    Please check your inboxes and accept the request if you're willing to help out.

    Thanks,
    Luke

  • James Waller User

    We can give you the details if you're still looking to implement this?