The most practical way to connect LinkedIn Ads to BigQuery is to use a managed ETL connector. The connector authenticates with LinkedIn's Advertising API, retrieves campaign reporting data and loads it into BigQuery on a schedule.
There are 3 practical routes:
LinkedIn's official integration route is its Marketing API. BigQuery has its own loading and ingestion methods, so connecting the two usually requires either a third-party connector or a custom extraction pipeline.
| Requirement | Recommended approach | Why |
|---|---|---|
| Fast setup with minimal engineering | Managed ETL connector | Handles authentication, scheduling, loading and common schema changes |
| Multiple advertising platforms | Multi-source connector | Brings LinkedIn Ads, Google Ads, Meta Ads and other channels into one system |
| Full control over fields and transformations | Custom API pipeline | Lets you define requests, storage and data models |
| One-time export | CSV upload to BigQuery | Simple, but not automated |
| High-volume or regulated environment | Custom pipeline with governance controls | Gives you more control over access, logging and retention |
You can load account, campaign, campaign group, creative and supported demographic reporting into BigQuery. Common metrics include:
The available metrics depend on the reporting pivot. LinkedIn states that some metrics are unavailable with particular pivots, so the number of fields supported by a connector is only part of the evaluation. The connector must support the dimensions and metrics your reports need in combination.
A useful BigQuery model usually separates the data into tables such as:
linkedin_ads_rawlinkedin_campaign_performancelinkedin_creative_performancelinkedin_demographic_performancelinkedin_ads_sync_logKeep the original API response separate from the reporting tables. The raw layer helps you investigate discrepancies, replay failed loads and troubleshoot a sync without repeatedly querying LinkedIn.
A managed connector suits marketing teams that need reliable reporting without maintaining an API integration.
The connector should support:
Do not assume that every LinkedIn Ads connector supports the same API resources. Some tools load only basic campaign metrics. Others also support demographic reporting, conversion data or lead information.
A custom pipeline gives you more control, but your team must maintain the extraction, error handling and data model.
A typical architecture looks like this:
LinkedIn Advertising API
↓
Extractor service
↓
Raw storage or staging table
↓
Transformation layer
↓
BigQuery reporting tables
↓
Looker, Power BI, Tableau or internal analytics
LinkedIn's reporting API requires the appropriate advertising permissions. The r_ads_reporting permission is used to retrieve advertising account reporting data. LinkedIn Marketing API access also uses authenticated member authorisation rather than a simple unauthenticated server connection.
A custom extractor should:
LinkedIn's Ad Analytics endpoint supports finder methods for analytics, statistics and attributed revenue metrics. It also has restrictions around pagination, metric availability and reporting pivots. The extractor needs to handle those restrictions before the data reaches dashboard users.
CSV works for:
The process is:
BigQuery supports CSV, JSON, Avro, ORC and Parquet data through load jobs. Google recommends scheduled or automated loading methods for recurring ingestion rather than manual uploads.
CSV is a poor long-term choice because it does not provide reliable incremental extraction, automatic reauthorisation, error monitoring or consistent handling of campaign changes.
A campaign performance table should include fields such as:
| Field group | Example fields |
|---|---|
| Identifiers | account_id, campaign_group_id, campaign_id, creative_id |
| Names | account_name, campaign_name, creative_name |
| Reporting grain | report_period, time_granularity, currency |
| Delivery metrics | impressions, clicks, reach |
| Cost metrics | spend, cost_per_click, cost_per_lead |
| Conversion metrics | leads, conversions, conversion_value |
| Metadata | campaign_status, objective, campaign_type |
| Pipeline fields | extracted_at, source_system, sync_run_id |
Where possible, use one row for each reporting period, account, campaign and creative combination. Do not put campaign-level and demographic-level records in the same fact table. The two grains differ, so combining them can multiply spend, impressions or conversion totals.
A clean model often includes:
Account, campaign and creative performance metrics are described by LinkedIn as near real-time. Professional demographic metrics can be delayed by 12 to 24 hours. Some video watch-time metrics can be delayed by up to 48 hours.
Your pipeline should update existing records after the first extraction. Treating the first result as final can leave reports incomplete.
LinkedIn states that professional demographic metrics are approximate for privacy reasons. Totals can differ when you compare different reporting ranges, reporting levels or combinations of dimensions.
Do not treat demographic breakdowns as an exact additive allocation of campaign totals.
A metric may be available at campaign level but unavailable for a demographic or device pivot. Preserve null values when LinkedIn does not provide a metric. Do not convert missing values into zero.
A user may be able to view an account in Campaign Manager without the connected application having the required API access. LinkedIn controls access through application approval, OAuth permissions and advertising account authorisation.
Do not treat BigQuery as an unrestricted archive for LinkedIn API data.
LinkedIn's storage requirements set retention limits for different categories of Marketing API data. Advertising account administration and reporting data have one storage period, while professional demographic data has a separate requirement. LinkedIn also has additional conditions for member-related data and deletion requests.
Your pipeline should include:
A connector that stores data indefinitely without configurable retention controls can create a governance problem even when the sync itself works.
After loading LinkedIn Ads into BigQuery, compare:
Use a tolerance policy for metrics known to be approximate or delayed. Investigate large differences caused by:
Use a managed connector when you need to centralise LinkedIn Ads data quickly for dashboards, attribution models or marketing performance reporting.
Build a custom pipeline when you need non-standard API requests, custom attribution logic, control over retention, detailed observability, integration with an existing data platform or lower variable connector costs at significant scale.
Use CSV for temporary transfers and one-off imports. For a production setup, load LinkedIn Advertising API data into a raw BigQuery layer, then publish separate campaign, creative and demographic tables.