How to load TikTok Ads data into Snowflake with OWOX Data Marts

What you end up with

A table in your own Snowflake account, tiktok_ads_ad_insights, with one row per ad per day by default: impressions, clicks, spend, conversions, video views and engagement.

The connector reads the TikTok Marketing API, writes that table and repeats on a schedule. It does not model the data. Reports on top of it are SQL you write.

Before you start

In TikTok. A TikTok for Business user assigned to the advertiser account, and the numeric advertiser ID. You find it in TikTok Ads Manager.

In Snowflake. A storage of type Snowflake connected to your project, as described in the storage guide: account identifier, warehouse, and a key pair or a username with a programmatic access token. The role must be able to create a database, a schema and tables. A small dedicated warehouse with auto-suspend is enough, and Snowflake pricing explained shows what a daily load costs.

Step 1: create the Data Mart

Click New Data Mart, enter a title such as “TikTok Ads Insights”, select the Snowflake storage and click Create Data Mart. On the Data Setup tab, set Definition Type to Connector and choose TikTok Ads.

Step 2: sign in and enter the advertisers

Continue with TikTok is the short way. Sign in as a user who can access the advertiser account and approve access. It needs no developer app.

Manually means your own TikTok developer app: an access token, an App ID and an App Secret. TikTok reviews every app, which can take up to seven business days, so this route is slower. The credentials guide covers it.

Then fill in Advertiser IDs. Use the numbers only. Several advertisers go in one field, separated by commas or semicolons, and they all write into the same table.

Step 3: choose the data level

This is the setting to get right the first time. Data Level decides what one row of the performance table means.

• AUCTION_AD, the default. One row per ad per day.
• AUCTION_ADGROUP. One row per ad group per day.
• AUCTION_CAMPAIGN. One row per campaign per day.
• AUCTION_ADVERTISER. One row per advertiser per day.

A finer level means more rows and more detail. A coarser one means a smaller table that cannot be broken down later.

The level also decides the keys the connector merges rows on, which is why you should not change it after data has been loaded. New rows would be matched on a different set of keys than the rows already in the table. If you need a second grain, create a second Data Mart with its own table.

For most reporting, keep the default. Ad level can always be summed up to campaigns. The reverse is not possible.

Step 4: choose the endpoint and the fields

Each Data Mart loads one endpoint.

• Ad Performance (ad_insights). Daily performance at the data level you chose. The one for reporting.
• Ad Performance by Country (ad_insights_by_country). The same, with a country breakdown.
• Campaigns, Ad Groups and Ads. Names, objectives, budgets, statuses and creative details.
• Advertiser and Audiences. Account details and custom audiences.

Then select fields. The key fields for your data level are locked, and the panel says which ones.

Beyond impressions, clicks and spend, the TikTok-specific fields are worth a look: video views, two-second and six-second views, completion, and likes, comments, shares and follows.

The last screen asks for the database, schema and table. The setup proposes a database named after the source, the schema PUBLIC and a table named after the endpoint.

Step 5: publish, backfill and schedule

Click Publish & Run Data Mart. The first run loads data from the first day of the previous month and can take several minutes. Afterwards the Data Setup tab shows the source, the Snowflake table and its columns.

History. Click Manual Run and choose Backfill (custom period). One backfill run covers at most 31 days. For a longer history, run consecutive periods, each after the previous one has finished.

Schedule. A Data Mart without a trigger does not run again. On the Triggers tab, add a trigger of type Connector Run: daily, weekly, monthly or at an interval.

Each scheduled run re-requests the days set by the Reimport Lookback Window, two by default. Impressions, clicks and spend settle within a day or two. Conversions keep arriving for as long as your attribution window, so if you report on them, set the lookback at least that long.

One advanced setting to leave alone: Create Empty Tables. With it off, an account with no delivery yet gets no table, and later runs fail because the table is not found.

What the table looks like in Snowflake

Names are case-sensitive. The connector creates the schema, table and column names quoted. In Snowflake that makes them case-sensitive, and an unquoted name is read as upper case. ads_owox.public.tiktok_ads_ad_insights fails with “does not exist or not authorized”. ads_owox."PUBLIC"."tiktok_ads_ad_insights" works. The database is not quoted; the schema, the table and every column are.

The day is stat_time_day, a date column.

IDs, not names. The performance table carries campaign_id, adgroup_id and ad_id. Names are in the Campaigns, Ad Groups and Ads endpoints. Load the ones you need as separate Data Marts and join on the ID.

Metrics are numbers. impressions, clicks, spend, the video counts and the engagement counts can be summed directly.

Query the data

Both queries were validated on Snowflake against a table loaded by the connector. The examples use a shared database, ads_owox, for all ad platforms.

Spend, clicks and cost per click by campaign and day, for the last 30 days:

SELECT
  "stat_time_day" AS date,
  "campaign_id",
  SUM("impressions") AS impressions,
  SUM("clicks") AS clicks,
  SUM("spend") AS spend,
  SUM("spend") / NULLIF(SUM("clicks"), 0) AS cpc
FROM ads_owox."PUBLIC"."tiktok_ads_ad_insights"
WHERE "stat_time_day" >=
  DATEADD(day, -30, CURRENT_DATE())
GROUP BY "stat_time_day", "campaign_id"
ORDER BY date, spend DESC;

Engagement rate by campaign, a number TikTok reports are often read for:

SELECT
  "campaign_id",
  SUM("impressions") AS impressions,
  SUM("likes") + SUM("comments") + SUM("shares")
    AS engagements,
  (SUM("likes") + SUM("comments") + SUM("shares"))
    / NULLIF(SUM("impressions"), 0)
    AS engagement_rate,
  SUM("spend") AS spend
FROM ads_owox."PUBLIC"."tiktok_ads_ad_insights"
WHERE "stat_time_day" >=
  DATEADD(day, -30, CURRENT_DATE())
GROUP BY "campaign_id"
ORDER BY spend DESC;

What counts as an engagement is your definition. Here it is likes, comments and shares. Write it down once, in a query that is published, and not in each report.

Compute cpc, ctr and cpm from sums, as above. The columns with those names hold the ratio for one ad on one day and cannot be averaged across rows.

From table to report

A connector Data Mart can be reported from directly. On its Destinations tab you attach reports to Google Sheets, Data Studio, Slack or email, each with its own columns and schedule. Snowflake reports in Google Sheets walks through it.

For a report with campaign names and your own engagement definition, publish that query as a second Data Mart and report from it. To put TikTok next to the other channels in one table, see self-service marketing analytics on Snowflake.

Where to go next

The same setup loads the other sources into the same Snowflake account: Facebook Ads, Google Ads, Microsoft Ads, LinkedIn Ads, Reddit Ads and Shopify.

On Databricks the steps are the same and the table names have three parts: see TikTok Ads to Databricks. For the platform itself, start with what Snowflake is and how it works.

SnowflakeAutomate Snowflake reports in Google Sheets with OWOX Data MartsOctober 1, 2026→SnowflakeHow to load Facebook Ads data into Snowflake with OWOX Data MartsOctober 1, 2026→SnowflakeHow to load Google Ads data into Snowflake with OWOX Data MartsOctober 1, 2026→

See all articles →