TikTok Ads data in BigQuery is best handled through the TikTok Marketing API, a raw landing layer, and a normalized reporting table.
For a recurring pipeline, choose one of these approaches:
| Method | Best for | Main advantage | Main limitation |
|---|---|---|---|
| TikTok Ads Manager export | One-off analysis | No development required | Manual or email-based workflow |
| Managed ETL connector | Teams that want scheduled syncs with little maintenance | Fast implementation | Connector fees and possible schema limits |
| Custom TikTok API pipeline | Data teams and agencies | Full control over fields, logic and refreshes | Requires development and ongoing maintenance |
Google's published BigQuery Data Transfer Service source list includes platforms such as Google Ads and Facebook Ads, but it does not list TikTok Ads. TikTok Ads therefore usually requires a CSV workflow, a third-party connector or a custom API integration rather than a standard Google-native transfer. Google Cloud documentation
A production integration should use this flow:
TikTok Ads API
↓
Raw JSON or CSV landing layer
↓
BigQuery staging tables
↓
Typed and deduplicated fact tables
↓
Marketing dashboards, attribution and warehouse models
The reporting grain is usually:
advertiser_id + ad_id + reporting date
For campaign-level analysis, use:
advertiser_id + campaign_id + reporting date
Do not combine campaign, ad group and ad rows in one fact table without a field such as report_level. Otherwise, the same spend can be counted at multiple levels.
TikTok Ads Manager lets users create custom reports with dimensions such as campaigns, ad groups and ads. Reports can include metrics such as spend, impressions, clicks, CPM, CPC, CTR and conversions. You can download the results for loading into BigQuery. TikTok Ads help
Manual exports work when:
TikTok also supports scheduled custom report exports delivered by email. This reduces manual work, but the process still depends on report files and email delivery. An API or warehouse connector is usually easier to monitor. TikTok Ads help
BigQuery supports batch loading from Cloud Storage or local files in formats including CSV, newline-delimited JSON, Avro and Parquet. BigQuery documentation
The TikTok Marketing API is the practical choice when you need scheduled, repeatable ingestion without relying on downloaded files.
TikTok's reporting API can return advertising data at advertiser, campaign, ad group and ad levels. Depending on the selected dimensions and report type, reports can include spend, impressions, reach, clicks, conversions, video views, engagement and website event metrics. TikTok Business API documentation
A custom integration requires:
TikTok requires advertiser authorization before an application can access Marketing API functionality and ad account data. TikTok Business API documentation
Use a synchronous report for a small or interactive query. TikTok states that synchronous reporting can return data for up to 20,000 ads. Larger requests should be split with campaign, ad group or ad filters. TikTok Business API documentation
Use an asynchronous report for scheduled warehouse extraction and larger recurring jobs. TikTok's asynchronous reporting supports Ad Reports and uses separate task creation and download steps. TikTok Business API documentation
A typical API workflow is:
A useful fact table might include:
CREATE TABLE `project.marketing.tiktok_ads_daily`
(
advertiser_id STRING NOT NULL,
report_level STRING NOT NULL,
campaign_id STRING,
campaign_name STRING,
adgroup_id STRING,
adgroup_name STRING,
ad_id STRING,
ad_name STRING,
stat_date DATE NOT NULL,
spend NUMERIC,
impressions INT64,
clicks INT64,
reach INT64,
conversions INT64,
video_views INT64,
source_updated_at TIMESTAMP,
loaded_at TIMESTAMP NOT NULL
)
PARTITION BY stat_date
CLUSTER BY advertiser_id, campaign_id, adgroup_id, ad_id;
TikTok report responses commonly expose metric values as strings in the API documentation. Cast those values during transformation rather than leaving spend and performance metrics as text. TikTok Business API documentation
A safe natural key should include:
advertiser_id
report_level
entity_id
stat_date
The value of entity_id depends on report_level:
campaign_id for campaign-level rowsadgroup_id for ad group-level rowsad_id for ad-level rowsThe source query should return only one row for each natural key before the merge runs.
The staging table should contain the latest extracted rows. The curated table can then be updated with a MERGE statement:
MERGE `project.marketing.tiktok_ads_daily` AS target
USING (
SELECT
advertiser_id,
report_level,
campaign_id,
campaign_name,
adgroup_id,
adgroup_name,
ad_id,
ad_name,
stat_date,
SAFE_CAST(spend AS NUMERIC) AS spend,
SAFE_CAST(impressions AS INT64) AS impressions,
SAFE_CAST(clicks AS INT64) AS clicks,
SAFE_CAST(reach AS INT64) AS reach,
SAFE_CAST(conversions AS INT64) AS conversions,
SAFE_CAST(video_views AS INT64) AS video_views,
CURRENT_TIMESTAMP() AS loaded_at
FROM `project.marketing.tiktok_ads_staging`
) AS source
ON target.advertiser_id = source.advertiser_id
AND target.report_level = source.report_level
AND COALESCE(target.ad_id, target.adgroup_id, target.campaign_id)
= COALESCE(source.ad_id, source.adgroup_id, source.campaign_id)
AND target.stat_date = source.stat_date
WHEN MATCHED THEN UPDATE SET
campaign_name = source.campaign_name,
adgroup_name = source.adgroup_name,
ad_name = source.ad_name,
spend = source.spend,
impressions = source.impressions,
clicks = source.clicks,
reach = source.reach,
conversions = source.conversions,
video_views = source.video_views,
loaded_at = source.loaded_at
WHEN NOT MATCHED THEN INSERT (
advertiser_id,
report_level,
campaign_id,
campaign_name,
adgroup_id,
adgroup_name,
ad_id,
ad_name,
stat_date,
spend,
impressions,
clicks,
reach,
conversions,
video_views,
loaded_at
)
VALUES (
source.advertiser_id,
source.report_level,
source.campaign_id,
source.campaign_name,
source.adgroup_id,
source.adgroup_name,
source.ad_id,
source.ad_name,
source.stat_date,
source.spend,
source.impressions,
source.clicks,
source.reach,
source.conversions,
source.video_views,
source.loaded_at
);
The merge updates previously loaded rows instead of appending duplicates. Reprocess recent completed reporting periods because TikTok reporting data can arrive late or be revised.
Start with a small field set that matches the reports your team will use.
advertiser_idcampaign_idcampaign_nameadgroup_idadgroup_namead_idad_nameAvailable metrics depend on the selected report dimensions. TikTok also notes that not every dimension available in Ads Manager is available through the API. Test the report definition before finalizing the warehouse schema. TikTok Business API documentation
TikTok states that asynchronous report data is refreshed periodically and that current-day data may be inaccurate. Treat completed reporting periods as the basis for dashboards and reconciliation rather than treating same-day numbers as final. TikTok Business API documentation
TikTok documents possible differences between synchronous reports and reports downloaded from Ads Manager. When reconciling totals, compare an Ads Manager download with its asynchronous API equivalent. Do not assume that every reporting method will produce identical results. TikTok Business API documentation
TikTok restricts which dimensions can be combined in one request. Report designs commonly use one entity ID dimension together with a time dimension. Test each combination of entity, date, targeting and conversion fields before using it in production. TikTok Business API documentation
TikTok Ads data measures platform-reported delivery and attribution. It does not replace first-party order, subscription or CRM data.
For ROAS analysis, join TikTok Ads data with:
Use campaign, ad group and ad IDs in tracking parameters when the destination and campaign setup support them. Keep TikTok-attributed conversions separate from warehouse-attributed conversions so the two measurement systems are not presented as the same metric.
Choose a managed connector when:
Choose a custom API pipeline when:
Choose manual exports only for occasional reporting. Manual files are not a strong foundation for automated dashboards, multi-account reporting or historical backfills.
For most companies, the setup should be:
This structure gives you automated ingestion while retaining the raw data needed for reconciliation, debugging and future reporting changes.