For a Shopify brand moving marketing data into a warehouse, Fivetran is usually the lowest-effort choice, Airbyte is usually the lower-cost and more flexible one, and a custom pipeline makes sense only for a source no connector handles well. Both tools connect Shopify, Klaviyo, Meta Ads and Google Ads to BigQuery, Snowflake or Redshift. They differ in how they charge, how much control a brand gets, and who fixes a connector when a platform changes its API. This page compares them on published prices and documentation, adds BigQuery's built-in transfers as a third option, and sets out when custom code is worth it. Brand Media Works builds marketing data pipelines for ecommerce brands and works with all three routes.
How do Fivetran and Airbyte compare?
| Fivetran | Airbyte | |
|---|---|---|
| How it charges | Monthly active rows (MAR): each row that is added or changed in a month counts once, however many times it syncs | Credits, billed by data volume; Airbyte's calculator equates 40 credits to about 6,640,000 API rows or 10 GB of database data |
| Free option | Free plan up to 500,000 MAR a month across connections | Airbyte Core, the open-source edition, self-hosted on the brand's own servers |
| Entry paid price | Usage-based, with a $5 base charge per connection between 1 MAR and 1 million MAR | Standard from $20 a month with 5 credits, extra credits $5 each; Plus from $189 a month for 40 credits |
| Published example | Median list price of $48.67 a month for Google Ads and $22.06 for Facebook Ads, on the Standard plan for a company of 1 to 200 employees | 40 credits a month cost $195 on Standard and $189 on Plus, per Airbyte's calculator |
| Sync frequency | 15-minute syncs on Standard, 1-minute on Enterprise | 15-minute syncs on Plus |
| Connectors | 700+ fully managed connectors | 700+ connectors, the same code in Core and Cloud |
| Hosting | Managed by Fivetran; a hybrid deployment option on Enterprise | Managed cloud, or self-hosted with abctl or Kubernetes; Enterprise Flex runs the data plane in the brand's own cloud |
| Trial | 14 days free on the account and on each new connection | 14 days with 400 credits ($2,000 of usage) |
The billing units are the main trap in any price comparison. Ecommerce pipelines re-read recent data every day because conversions are revised after the fact: Meta conversions are re-read for 9 days by default in Fivetran's connector and up to 28 days in Airbyte's, and Klaviyo keeps revising attribution for up to 5 days after a send by default, or up to 90 days. Under Fivetran's monthly active rows, a row synced 30 times in a month still counts once. Airbyte's Standard plan bills by data volume, so a brand should check how its re-read windows affect the credit count during the trial rather than compare list prices.
What does each cost for a Shopify brand?
Neither vendor publishes a median price for Shopify or Klaviyo connectors, because the cost depends on order volume, catalog size and how many events are synced. What is published:
- Fivetran: the two ad connectors at median usage come to $70.73 a month (our arithmetic from the $48.67 and $22.06 medians). A small brand may fit inside the Free plan's 500,000 monthly active rows; once over it, each paid connection with up to 1 million MAR carries a $5 base charge on top of usage.
- Airbyte Cloud: $20 a month covers 5 credits. A brand syncing around 6.6 million API rows a month would land at about $189 to $195 a month, per Airbyte's own calculator.
- Airbyte Core: no license fee, but the brand pays for the servers and for someone to run upgrades and fix failed syncs.
- BigQuery's built-in transfers: Google Ads is a free connector, Shopify and Klaviyo are free while in Preview, and Facebook Ads is billed by the compute it uses. How to build a marketing data warehouse sets out what each built-in transfer does and what it costs.
For most brands with four main sources, the connector bill is tens to low hundreds of dollars a month at published prices. The larger cost is the build and the upkeep around it.
When is a custom pipeline worth it?
Writing custom code against each platform's API removes the subscription but moves the maintenance to the brand. The sources an ecommerce brand depends on change often:
- Shopify deprecated its REST Admin API on 1 October 2024, and on 1 October 2026 began moving shops' shipping settings into Markets, a rollout planned to reach all shops by 1 July 2027.
- Meta removed the 7-day and 28-day view-through attribution windows from its Ads Insights API on 12 January 2026, and requires an app to make at least 500 Marketing API calls in the past 15 days, with an error rate below 15%, to keep full API access.
- Klaviyo allows 225 requests a day on its reporting API, which limits how much a custom job can pull.
Connector vendors update their connectors when these changes land; with a custom pipeline, the brand has to make each fix itself. Build prices are modest: one data analytics consultancy that sells Shopify reporting publishes about $4,000 and about two weeks for a pipeline from one or two sources, and one data engineering firm puts a focused pipeline connecting three sources to one warehouse at four to six weeks to production. Custom code earns its place when:
- A source has no connector, or the connector does not expose the fields the brand needs, such as a niche subscription, returns or loyalty app.
- Volumes are high enough that per-row pricing outweighs the cost of an engineer's time.
- Data has to stay inside the brand's own network, though Airbyte Core or Airbyte's Enterprise Flex may meet that need without custom code.
Which should an ecommerce brand choose?
| Situation | Usual choice |
|---|---|
| Small team, no data engineer, wants syncs that keep working | Fivetran, starting on the Free plan |
| On BigQuery, with Google Ads as the largest source | BigQuery's built-in transfers for Google sources, a connector tool for the rest |
| Cost-sensitive, with someone technical to watch the syncs | Airbyte Cloud |
| Strict data-residency rules or an in-house engineering team | Airbyte Core self-hosted, or Enterprise Flex |
| A source no connector covers well | A custom pipeline for that one source, a connector tool for the rest |
Mixing routes is normal. A brand can use a free built-in transfer for Google Ads, a connector tool for Shopify, Meta and Klaviyo, and custom code only where nothing else fits. Whatever the mix, the joins, definitions and checks described in data engineering for ecommerce matter more than the ingestion tool.
How should a brand test the options?
- Run both trials on real data. Fivetran gives 14 days per new connection; Airbyte gives 14 days with 400 credits.
- Load full history, then watch a normal week. The first sync is the expensive one; the steady-state week shows the real monthly bill.
- Compare the tables, not just the price. Check that orders, refunds, discounts and ad conversions arrive with the fields the reports need.
- Check who gets alerted when a sync fails, and how long a fix takes when a platform changes its API.
Marketing data engineering sets out how Brand Media Works scopes and builds this work, and how much marketing data engineering costs sets out published project prices.
Frequently asked questions
Is Airbyte free?
Partly. Airbyte Core, the open-source edition, is free to use: it has no license fee, but the brand hosts it on its own servers and pays for that hosting and upkeep. Airbyte Cloud is paid. Its Standard plan starts at $20 a month with 5 credits included, and extra credits cost $5 each. New Cloud accounts start with a 14-day free trial that includes 400 credits (our arithmetic: $2,000 at $5 a credit). Prices as shown on Airbyte's pricing page, read 18 October 2026.
Is Fivetran too expensive for a small brand?
Not necessarily. Its Free plan covers up to 500,000 monthly active rows a month, and its published medians for Google Ads and Facebook Ads connections are under $50 a month each. Costs rise with volume, so check the bill after a normal week.
Can a brand switch tools later?
Yes. Both load into the brand's own warehouse, so the data stays put. Switching means setting up the new connections and running a fresh initial sync, and checking that the new tables match the reports built on the old ones.