Connect Airtable to BigQuery: Manual, Script, and Auto-Sync

Summary

  • For a one-time load: export a CSV from Airtable and upload it in BigQuery.
  • For full control if your team writes Python: a script on the Airtable API with a personal access token, paging and retries on 429 errors.
  • For scheduled no-code replication with incremental updates: Skyvia, once a Last modified time field is in each table.
  • For sending scores or enriched values from BigQuery back to Airtable: Skyvia Import with INSERT, UPDATE or DELETE.

For plenty of teams, Airtable is the first "database" they build without asking IT. Projects, a content calendar, maybe a scrappy CRM. Then finance asks how those projects line up with Stripe revenue, marketing wants GA4 sessions next to the leads, and Airtable's views and formulas weren't built for that kind of join. That's the point where people start looking at moving data from Airtable to BigQuery.

We'll show you three ways to do it: export a CSV and load it by hand, write a Python script against the Airtable API, or let Skyvia replicate the tables on a schedule. Before any of that, though, there's some Airtable housekeeping (an access token, your plan's record and API-call limits), because a sync can stall there before it moves a single row.

Full disclosure: Skyvia is our product. We'll still tell you when a CSV or a short script is all you need.

Why Export Data from Airtable to BigQuery?

Overcoming Limits

Airtable is generous up to a point — then the hard caps kick in. Records per base depend on your plan: 1,000 on Free, 50,000 on Team and 125,000 on Business. Once you’re bumping against that limit, there’s no elegant way around it inside Airtable itself. Pushing data into BigQuery removes that ceiling entirely. Your warehouse scales with you, not against you. 

Advanced Analytics 

Airtable’s built-in reporting is fine for day-to-day visibility, but it doesn’t hold up when you need to run complex SQL queries or join your project data with data from Google Analytics, Salesforce, or your payment system. BigQuery makes that kind of cross-source analysis straightforward — and fast, even at serious data volumes. 

Visualization 

Airtable Interfaces cover the basics, but they’re not built for professional BI work. Once your data lands in BigQuery, tools like Looker Studio, Tableau, and Power BI can connect directly and deliver dashboards that keep up with the questions the business is asking. 

Data Historization 

Airtable doesn’t natively track how records change over time — when a field was updated, what it looked like last month, what changed between two points. BigQuery lets you build proper snapshotting workflows, so you always have a historical log to go back to. For compliance, auditing, or just understanding trends, that’s harder to live without than most teams expect. 

What Should You Set Up in Airtable Before a BigQuery Sync?

Only after a one-off CSV? You can skip ahead, then. Things change when a script or a sync tool calls the Airtable API. Now it matters how it logs in, how many calls your plan includes, and how Airtable reacts to requests that arrive too fast.

Signing in. Airtable's old API keys stopped working on February 1, 2024, so a script today uses a personal access token. You make one at airtable.com/create/tokens. Give it the data.records:read scope, add schema.bases:read if your code needs table and field definitions, and pick which bases it may open. It can't see anything your own Airtable account can't. Skyvia works differently here: you sign in to Airtable and choose the bases and workspaces to grant.

Your plan's limits. Airtable caps both records per base and API calls per workspace:

Airtable planRecords per baseAPI calls per workspace per month
Free1,0001,000
Team50,000100,000
Business125,000Unlimited

Here's why the last column bites. Because the API caps each response at 100 records, one full pass over a 50,000-record table eats 500 calls. Run that hourly and a Team workspace is out of calls in about eight days. That's the quiet way a homemade sync fails: it reloads everything on every run.

Going too fast. The limit is 5 requests per second per base, and 50 per second for everything one user runs through personal access tokens. Cross it and Airtable sends back a 429 error, and nothing gets through for the next 30 seconds. Your script should back off and retry instead of falling over.

And if Skyvia's incremental replication is the plan, add a Last modified time field to each table while you're in there. Skyvia uses a field of that type to tell which records changed, and it needs to be in place before the first run.

Method 1: The Manual Approach (CSV Export) 

No native connection exists between the two platforms, so the simplest way to get data across is the old-fashioned way — export a CSV from Airtable and load it straight into BigQuery. No additional tools, no code, no setup. Just a few clicks and some patience. 

In Airtable 

  1. Log in to Airtable and open the base you want to export. 
Airtable Tasks Grid View with Status, Priority, and Deadline Fields
  1. Click Grid View and select Download CSV from the menu. 
Airtable Grid View Context Menu with Download CSV and Other Options
  1. Save the file somewhere you’ll find it in the next thirty seconds. 

In BigQuery 

  1. Log in to BigQuery with your Google account and head to BigQuery Studio. 
  2. Under Add your own data, click Load file. 
Adding Data from a Local File in Google BigQuery Console
  1. Select your CSV, fill in the required fields in the Create table window, and hit Create table. 
BigQuery Create Table Dialog – Uploading a CSV File from Airtable
  1. Navigate to the dataset where the table was created and click Preview to confirm the data landed correctly. 
BigQuery Table Preview of Airtable Data with Task Records

Best for 

One-time exports and quick ad-hoc analysis where freshness doesn’t matter and you’re not planning to repeat the process. 

Pros 

  • Completely free — no tools or accounts needed beyond what you already have. 
  • Zero setup time, works for any team regardless of technical ability. 

Cons 

  • No automation whatsoever — every update means repeating the whole process by hand. 
  • Data types frequently break on export: dates come through as plain text, integers lose their formatting, and linked records turn into a messy comma-separated string that needs cleaning before it’s usable. 

Method 2: The “Do-It-Yourself” Approach (Airtable API & Python)

If the CSV approach feels too manual but you’re comfortable with Python, pulling data directly from Airtable’s REST API and pushing it into BigQuery with Google’s client library is a solid middle ground. You get automation, scheduling control, and full flexibility over what gets moved and how. The trade-off is that you’re also owning every edge case that comes with it. 

Best for 

Engineering teams that want tight control over their pipeline logic and have the bandwidth to build and maintain custom scripts.  Less ideal for teams that need something that just runs without babysitting. 

Step-by-Step Guide 

You’ll need two libraries to get started: 

pip install requests google-cloud-bigquery 

Here’s a basic script that pulls records from an Airtable table and loads them into BigQuery: 

import requests 

import time 

from google.cloud import bigquery 

# Airtable config 

AIRTABLE_TOKEN = "your_personal_access_token" 

BASE_ID = "your_base_id" 

TABLE_NAME = "your_table_name" 

AIRTABLE_URL = f"https://api.airtable.com/v0/{BASE_ID}/{TABLE_NAME}" 

# BigQuery config 

BQ_PROJECT = "your_project_id" 

BQ_DATASET = "your_dataset" 

BQ_TABLE = "your_table" 

headers = {"Authorization": f"Bearer {AIRTABLE_TOKEN}"} 

records = [] 

params = {} 

# Paginate through all records 

while True: 

    response = requests.get(AIRTABLE_URL, headers=headers, params=params) 

    data = response.json() 

    records.extend(data.get("records", [])) 

    offset = data.get("offset") 

    if not offset: 

        break 

    params["offset"] = offset 

    time.sleep(0.2) # Stay within rate limits 

# Flatten records for BigQuery 

rows = [{"id": r["id"], **r["fields"]} for r in records] 

# Load into BigQuery 

client = bigquery.Client(project=BQ_PROJECT) 

table_ref = f"{BQ_PROJECT}.{BQ_DATASET}.{BQ_TABLE}" 

errors = client.insert_rows_json(table_ref, rows) 

if errors: 

    print("Errors:", errors) 

else: 

    print(f"Loaded {len(rows)} rows successfully.") 

The Hidden Challenges 

This is where things get interesting and not always in a good way. 

Pagination. Airtable returns a maximum of 100 records per API call. For any reasonably sized base, you’ll need to loop through offset tokens until there are no more pages left. The script above handles this, but it’s easy to miss if you’re prototyping quickly and testing on a small dataset. 

Rate limits. Airtable caps you at 5 requests per second. Hit that ceiling without handling it properly, and your script starts throwing errors mid-run. The right fix is exponential backoff. Wait a bit, retry, wait a bit longer if it fails again. Without it, your pipeline becomes unreliable the moment data volumes grow. 

Schema evolution. This one tends to sneak up on teams. The moment a marketing manager adds a new column in Airtable, which takes them about three seconds. Your static BigQuery schema breaks. Every new field needs to be reflected in your table definition, and if nobody’s watching for that, you’ll lose data silently until someone notices a report looking odd. 

Pros 

  • Full control over transformation logic, filtering, and scheduling. 
  • Free to run beyond standard cloud compute costs. 
  • Can be extended and customized to fit specific pipeline requirements. 

Cons 

  • Significant upfront build time and ongoing maintenance overhead. 
  • Schema changes in Airtable require manual updates to keep the pipeline running. 
  • Retry logic, pagination, and error handling all need to be built from scratch, and getting them right takes longer than most teams budget for. 

Method 3: Automating Airtable to BigQuery with Skyvia

Skyvia is a cloud data platform that covers integration, replication, backup, automation, and API management from one place. It supports 200+ connectors and is code-free. 

Where the CSV approach breaks down on scale, and the Python route demands ongoing engineering attention, Skyvia handles the pipeline for you:  

  • Connections. 
  • Scheduling. 
  • Schema creation. 
  • Incremental updates. 

For the Airtable–BigQuery connection specifically, the right tool inside Skyvia is Replication: an ELT scenario that bulk-loads and incrementally syncs data from Airtable directly into your warehouse. You can see the full scenario on the Airtable to BigQuery replication page.

Best for 

Teams that want a reliable, automated pipeline without the overhead of building and maintaining one. Works just as well for a solo analyst as it does for a data engineering team that has better things to do than babysit a sync script. 

Prerequisites 

Step-by-Step: Replicating Airtable to BigQuery 

  1. Log in to your Skyvia account. 
  2. Click +Create New → Replication in the upper menu. 
  3. Select Airtable as the source and BigQuery as the target. 
  4. Choose the tables you want to replicate. 
  5. Select Incremental Updates to sync only new or modified records going forward. 
New Integration in Skyvia
  1. Click Schedule, set your preferred update frequency, and hit Save. 
Scheduling in Skyvia
  1. Click Create, then Run to kick off the first replication. 
Airtable to BigQuery Integration Setup in Skyvia
  1. Monitor progress in the Monitor tab. Once it’s done, open BigQuery and confirm the data landed correctly. 
BigQuery Preview of Tasks Table with Airtable Record IDs

Note: The free plan supports one sync per day. For more frequent updates, paid plans run hourly on Standard and as often as every minute on Professional. 

Two Features Worth Calling Out 

Incremental Updates. Rather than dumping the entire Airtable base into BigQuery on every run, Skyvia moves only what’s changed since the last sync. It finds edited records through a field of the Last modified time type, so add one to each Airtable table before the first run; without it, new records still arrive but edits don’t. For large bases that update frequently, this makes a meaningful difference in BigQuery query costs because you’re not reprocessing data that hasn’t changed. 

Automatic Schema Creation. On the first replication run, Skyvia inspects your Airtable structure and builds the corresponding BigQuery table automatically. No manual DDL, no writing CREATE TABLE statements, no worrying about whether your field types match up. Later changes are covered by the Update Schema option: with it on, incremental runs add new Airtable tables and fields to BigQuery, though columns for deleted fields stay in place. That’s the problem that makes the Python approach so brittle over time.

Pros 

  • No code required. The entire setup takes minutes, not hours. 
  • Incremental sync keeps BigQuery costs in check by moving only changed records. 
  • Automatic table creation on the first run, and new Airtable fields added to BigQuery when Update Schema is on.
  • Flexible scheduling, from once a day on Free to every minute on Professional.
  • Detailed run logs and monitoring built in. 

Cons 

  • The free plan is limited to one sync per day. Fine for non-urgent use cases, but teams needing frequent updates will need a paid plan. 
  • As a managed cloud service, you’re working within Skyvia’s connector coverage rather than building fully custom transformation logic. 

Deep Dive: Handling Linked Records and Arrays 

This is where Skyvia genuinely pulls ahead of a flat CSV export, and it’s worth understanding why. 

In Airtable, Linked Record fields don’t store the actual values of the linked records. They store an array of record IDs — something like [‘rec123’, ‘rec456’]. When you export to CSV, those IDs come through as a messy concatenated string with no structure, and you lose the relationship entirely. 

Skyvia gives you a choice here. In a replication, a field that holds an array or nested JSON can be stored as JSON in a text column or unwound into a separate child table, set per field in the task editor. A child table’s primary key includes the parent record’s key, so you can join it back to the parent in BigQuery. Either way, the record IDs from Airtable survive the move in a usable form, which a CSV export can’t give you and a homemade script has to build by hand.

Bonus: Reverse ETL

Once data is enriched in BigQuery, joined with other sources, scored, transformed, you’ll often want to push those results back into Airtable so the teams working there can actually use them. That’s exactly what Skyvia’s Import scenario handles. 

Enriching Airtable from BigQuery 

Say your data team has added a Task Score column to the BigQuery table based on an analysis that isn’t possible inside Airtable. Here’s how to get it back: 

  1. Log in to your Skyvia account. 
  2. Click +Create New → Import in the upper menu. 
  3. Select BigQuery as the source and Airtable as the target. 
New Reverse ETL Flow in Skyvia
  1. Click Add task, select the object to import, and choose your DML operation: 
    • INSERT: copies records from BigQuery to Airtable as new entries. 
    • UPDATE:  updates existing Airtable records with enriched values from BigQuery. 
    • DELETE:  removes records in Airtable based on the primary key. 
  2. Map source and target columns to align the data structures. 
  3. Click Schedule to set up regular syncs, then Save. 
  4. Click Run and monitor progress in the Monitor tab. 
  5. Once the process is complete, see the results in Airtable. 
Airtable Tasks Grid View with Task Score Field Added

Note: For Airtable tables, Skyvia Import supports INSERT, UPDATE and DELETE, but not upsert. To refresh a Task Score on existing records, use UPDATE and map the Airtable record ID; to add new rows, use a separate INSERT task.

Comparison: Manual vs. Script vs. Skyvia 

Every approach gets the job done under the right conditions. Here’s how they actually stack up across the things that matter once you’re past the initial setup: 

Criterion Manual (CSV Export) Script (Python & API) Automated (Skyvia) 
Setup Time Minutes Hours to days Minutes 
Technical Skill Required None Python, APIs, BigQuery SDK None 
Automation None Custom CRON/scheduler Built-in scheduler 
Maintenance Effort Low (but repetitive) High Low 
Incremental Updates No Manual Automated 
Array / JSON Handling Poor — linked records become flat strings Complex — requires custom parsing logic Configurable: JSON text column or a separate child table 
Historical Data No Manual snapshotting Automated 
Real-time Latency None Depends on CRON frequency Scheduled: daily on Free, down to every minute on Professional 
Cost Free Free + engineering time Free tier; paid from $79/mo 

Conclusion

The manual route works once. The Python script works until something breaks. Skyvia just works: incremental updates, automatic schema handling, linked records preserved, pipeline running on schedule without anyone babysitting it. If you’re weighing other platforms too, see our comparison of data ingestion tools. 

Ready to centralize your Airtable data? Start your free Skyvia trial today and have your BigQuery pipeline up and running in under 5 minutes. 

FAQ for Connect Airtable to BigQuery

Loader image

Yes, with Skyvia, since every Airtable table comes with a matching comments table named after it, such as Tasks_Comments. Those tables support incremental replication too, and nobody has to add a Last modified time field to them, because Skyvia reads their built-in InternalUpdatedDate column. So the discussion on each task can sit next to the task in BigQuery.

Not for long. The download URLs Airtable gives out come from airtableusercontent.com and expire a couple of hours after they’re generated, for security reasons. So a URL column in BigQuery goes stale fast. Need the attachments next quarter? Then we’d copy the files themselves into storage you own, Cloud Storage for instance, and keep that path in the table.

It’s the piece right after airtable.com/ that starts with app, up to the next slash. The segment after it, starting with tbl, is the table ID. The Airtable apps for iOS and Android hide the address bar, though, so on a phone you’d open airtable.com in a mobile browser to see it.

Because the script reads fields by their display names, and anyone with permission to edit fields can change one. Airtable suggests keying extraction code on field IDs, which start with fld, rather than on names. We’d pull the ID-to-name mapping once from the Get base schema endpoint and store it alongside the BigQuery column names your tables use.

Share

Nata Kuznetsova

Nata Kuznetsova is a seasoned writer with nearly two decades of experience in technical documentation and user support. With a strong background in IT, she offers valuable insights into data integration, backup solutions, software, and technology trends.

One platform for all your data work

One platform for all your data work

Integration, automation, live data access, and backup. No code, 200+ connectors.

Start free