Salesforce to BigQuery connector pricing: what the native BigQuery transfer, Fivetran and Airbyte really cost

How much does the BigQuery Salesforce connector cost?

Up to about $1.20 per hour of transfer run time in us-central1, by Google's own planning guideline. The BigQuery Data Transfer Service bills paid connectors in slot-hours, at $0.06 per slot-hour on demand, $0.054 with a one-year commitment and $0.048 with a three-year one. Google suggests planning for up to 20 slot-hours per hour a transfer runs, and says real consumption may differ.

That last caveat matters, and it is honest. A slot is a unit of BigQuery compute, and a transfer consumes them to orchestrate the Salesforce extract and to load and merge rows into your table. How many it uses depends on object size and how much changed. The 20 slot-hour figure is a conservative ceiling, which is what you want for a budget. Measure the real number from your billing export after the first week and replace the estimate.

One detail surprises people who used the service years ago. Google now lists Salesforce, along with ServiceNow, Salesforce Marketing Cloud, Facebook Ads, MySQL, PostgreSQL and Oracle, as a generally available paid connector. Connectors still in Preview, which today include HubSpot, Stripe, PayPal, Klaviyo and SQL Server, are not charged until they reach general availability. Google's own sources such as Google Ads and GA4 carry no transfer charge.

What does a Salesforce to BigQuery transfer cost per month?

Between roughly $12 and $288 a month for the same ten objects, depending only on the schedule. The table applies Google's ceiling of $1.20 per run hour to four realistic schedules. The run times are illustrative, and the dollar column is our arithmetic, not a Google quote. Replace the run time with what your transfer history shows and the math holds.

Two things push real bills above the table. Google recommends no more than 10 objects per transfer configuration, so a 30 object model is three configurations, each with its own run time. And incremental transfers never apply deletes, so most teams add the fourth row: a slower full reload that clears out records deleted in Salesforce. It is cheap on a weekly cadence and expensive if someone sets it to hourly by mistake.

There is also a trap that makes the 15 minute row cheaper than it should be, and not in a good way. A configuration runs one transfer at a time, and a run that starts before the previous one finishes is skipped. If a run occasionally takes 20 minutes, you pay for the long run and lose the next one. The schedule says 96 runs a day and the table in BigQuery is older than that. Every limit of this kind is listed on Salesforce to BigQuery integration tools, together with the currency fields Google lands as FLOAT.

How do Fivetran, Airbyte and Google compare on price?

They count different things, so the same Salesforce org can be cheapest on any of them. Google counts run time. Fivetran counts rows that changed during the month, once each however often they changed. Airbyte counts rows synced, at 6 credits per million for API sources like Salesforce, and on a full refresh it counts every row on every run. The table sets the six counting rules side by side.

Put numbers on it with one example org: 2 million Salesforce records across the core objects, about 20,000 of which change on a normal day. On Airbyte Cloud the Standard plan is $20 a month with 5 credits included and extra credits at $5 each, which by our arithmetic is about $30 per million API rows. Incrementally, 600,000 changed rows a month is 3.6 credits and fits inside the $20 plan. Switch the same org to a daily full refresh and it becomes 60 million rows and 360 credits, roughly $1,795 a month. Same data, same vendor, a 90x difference from one setting. The current Airbyte plan table is on Airbyte pricing.

Fivetran does not publish a per-row rate you can plug in, so we will not invent one. What its monthly active row model does well is ignore churn: a record edited fifty times in a month counts once. What it does badly is the first month, when the entire history counts as active. Price that month separately. The details of how the meter behaves are in Fivetran pricing.

Google's connector sits between the two. It does not care about your change rate, so a noisy org costs the same as a quiet one for the same run time. That makes it a good deal for large, busy Sales Cloud orgs on an hourly schedule, and a poor one for teams who want 15 minute freshness on a small org, where run time is mostly overhead.

Is the BigQuery Data Transfer Service free for Salesforce?

No. Salesforce is a generally available paid connector, billed in slot-hours from the first run. The free list is Google's own sources, such as Google Ads, GA4, Campaign Manager and YouTube, plus object storage sources like Cloud Storage and Amazon S3. That split is useful for attribution work: the ad data lands free, and Salesforce is the paid piece that closes the loop from click to closed-won. If what the marketing team actually wants is one view of spend and pipeline across every ad channel rather than a warehouse project, a unified marketing dashboard gets them there without anyone writing SQL.

What else do I pay for besides the connector?

BigQuery storage for the landed tables and compute for the queries that read them. On-demand queries cost $6.25 per TiB scanned, with the first TiB each month free. Storage is billed per GiB and drops to a cheaper long-term rate for tables not modified in 90 days, which a table updated every 15 minutes never reaches.

Then there is a cost that never appears on a Google invoice. Every Bulk API call counts against the Salesforce org's daily API allocation, which is shared with every other connected app. A tight schedule across many objects can starve your CPQ or marketing sync on a busy day. Salesforce includes five integration user licenses on Enterprise, Unlimited and Performance editions, and running the pipeline under one of them keeps its usage visible. The details are in the Salesforce integration user license.

When is a flat price cheaper than a metered connector?

When the bill has to be the same number every month, and when someone would otherwise spend time estimating slot-hours or credits. Adapters charges a flat tier by records synced: Starter at $49 for 10,000 records a month with hourly syncs, Growth at $149 for 100,000 with 5 minute syncs, Scale at $399 for 1 million with 1 minute syncs. There is no overage fee, and reading records that did not change costs nothing.

Be clear about where that loses. For the example org above, syncing incrementally, Airbyte's $20 plan is cheaper on price alone, and Google's connector on a daily schedule is cheaper still. What a flat plan buys is predictability and control over what lands: you choose the target type per field, so Opportunity.Amount arrives as NUMERIC instead of the FLOAT Google's connector writes, and adding an object next quarter does not mean a new transfer configuration and a fresh backfill. If those matter more than the last $30, the Salesforce to BigQuery connector is set up in minutes. For every other meter in the market, see data integration pricing.

How do I keep the Salesforce to BigQuery bill down?

Match the schedule to the decision the data feeds. Pipeline reviews happen weekly and forecasts daily, so hourly is plenty for almost every Salesforce object, and it costs a fraction of 15 minutes on a run-time meter. Keep the fast schedule for the one or two objects a live workflow depends on.

Then prune. After a month, check which tables anything downstream queried and drop the rest from the transfer. Put the delete cleanup on a weekly full reload rather than on every run. And never use a full refresh on a row-metered tool unless the object is small, because that is where the 90x surprise comes from. The same logic on the Snowflake side is worked through in Salesforce to Snowflake connector pricing, and every other source into the same warehouse is on BigQuery ETL tools.