Fivetran (Codex draft)

Fivetran is a strong choice for businesses that want to move data into a warehouse without maintaining every connection themselves. Its 700+ connectors, automatic schema updates and managed infrastructure make it particularly appealing to teams with several sources to look after.

My verdict: I recommend Fivetran when reducing pipeline maintenance matters more than getting the lowest software bill. Its usage-based pricing is my biggest reservation, especially if you need several connections or keep a detailed history of changing records.

In this Fivetran review, I’ll take a closer look at its pricing, connectors and workflow, explain what I like and dislike, and compare it with Airbyte and Hevo.

Editor’s note: This review evaluates Fivetran’s pricing, documented capabilities and suitability for different data teams.

Key Takeaways 🔍

• Fivetran’s managed connectors are its biggest strength: the catalog covers 700+ sources, reducing the need to maintain your own extraction code.
• The Free plan is worth considering: it includes 500,000 connection MAR, 3,500 activation MAR and 5,000 transformation runs monthly.
• Pricing can be difficult to forecast: connections have separate pricing curves, and history mode can increase counted usage.
• Custom connectors still involve work: you own SDK code and fixes unless you buy the maintenance add-on.
• Faster scheduling costs more: one-minute intervals require Enterprise or Business Critical and aren’t supported by every connector.

Fivetran Pros and Cons

I like Fivetran’s combination of broad source coverage and managed infrastructure. I’m less keen on the number of billing details you need to understand before choosing a paid plan.

For a team maintaining several supported pipelines, those advantages are persuasive. For a small business with one straightforward source, I’d compare cheaper options before paying for capabilities it may never need.

How Much Does Fivetran Cost?

Fivetran has four plans: Free, Standard, Enterprise and Business Critical. The paid plans charge by usage, so there isn’t one monthly subscription price that tells you what your business will spend.

That makes pricing less straightforward than I’d like. You choose a plan for its features, then pay according to your connections and usage. Fivetran’s pricing also separates ingestion from Activations, which sends warehouse data back into business applications.

• Free ($0 within its limits): for smaller workloads that fit the ongoing allowances.
• Standard (usage-based): my starting point for paid ingestion with 15-minute scheduling.
• Enterprise (usage-based): for supported enterprise databases, one-minute scheduling or Hybrid Deployment.
• Business Critical (usage-based): for additional controls such as customer-managed encryption keys and private networking.

Plan	Connection usage	Scheduling	Who I’d recommend it for
Free	Up to 500,000 MAR monthly	15 minutes	Smaller teams within all free allowances
Standard	Billed on usage	15 minutes	Businesses that need more usage without enterprise features
Enterprise	Billed on usage	Down to one minute where supported	Teams with specific database or deployment requirements
Business Critical	Billed on usage	Down to one minute where supported	Organizations requiring its additional security controls

Understanding monthly active rows

The monthly active rows (MAR) measure generally counts distinct row identities inserted, updated or deleted during a month. Each connection is priced separately, including when you send the same source data to two destinations.

For example, suppose the same 10,000 rows change daily in a normal, non-history table with stable primary keys. That can count as 10,000 MAR for the month, not 300,000. Send those rows through another connection and that connection counts its own 10,000. This is a usage illustration, not a dollar quote.

The complication is history mode, where repeated changes can create additional billable versions. Captured deletes also count under current rules. These exceptions are why I wouldn’t estimate costs from a table’s size alone.

Small connections also have a spending floor. Fivetran’s 2026 pricing update specifies a $5 minimum spending threshold per qualifying pay-as-you-go connection, including Standard connections generating 1 to 1 million MAR. Free accounts and connections with no paid MAR are exempt; annual contracts adopt the change at renewal. I’d factor that into a setup with many low-volume connections rather than assuming each will cost pennies.

Additional costs to consider

I like that Free includes 500,000 connection MAR, 3,500 activation MAR and 5,000 transformation runs monthly. However, paid plans don’t keep the 500,000 free connection allowance. Repeatedly exceeding Free’s limits can also interrupt pipelines.

Initial historical loads are free of connection MAR charges, but you still need to budget for destination storage and compute. Paid plans also retain 5,000 free transformation runs.

Transformation charges count successful model executions, including intermediate models. For an illustrative paid account after trials, 20 models running 12 times daily for 30 days produce 7,200 runs. With no other transformation usage, the included 5,000 leave 2,200 charged at $0.01, or $22 in transformation fees. Ingestion, Activations and destination costs are additional.

Is Fivetran Good Value for Money?

I would pay more for less recurring maintenance, but only where the time saved matters to the team. A higher software bill is easier to justify when engineers can spend that time on reporting or product work. For a single straightforward pipeline, I’m less convinced.

• Free is my first choice for smaller workloads, provided every required feature fits.
• Standard is my recommendation for most eligible paid workloads. Higher usage alone doesn’t justify Enterprise.
• Annual contracts deserve caution: the published minimum commitment is $12,000. I’d establish realistic usage before committing.

Reviewer’s Notes

I’d choose Standard when its connectors and 15-minute scheduling cover the job. Enterprise becomes more compelling when you need a specific database connector, faster eligible syncs or Hybrid Deployment. Pay for a feature your team needs, rather than a higher plan name.

Getting Started With Fivetran

Fivetran’s setup centers on connecting a source to a destination, rather than building an extraction script. I like that approach for teams whose main aim is to make existing business data available for reporting.

The onboarding process takes you through account verification, source authentication and destination configuration. Save & Test checks the setup before the initial sync. You can add transformations once data reaches the destination.

Connecting your data

You choose the source application or database, provide the required access and configure where the data should go. The destination could be your existing warehouse; it still needs its own setup and budget.

I like having a defined setup path for a supported connector instead of a new extraction project. That won’t remove source permissions or warehouse administration, but it can reduce the code your team has to own.

I see the strongest fit in a team that already knows what it wants to report on. If your analysts are waiting for data from several supported applications, managed connections address a recognizable problem. If you’re still deciding what to collect or how to model it, Fivetran won’t make those decisions for you.

The first load is only part of the experience. I would want to see how a connection handles ordinary business activity: changed customer records, deleted entries and new fields. A static sample can make a pipeline look more suitable than it is.

Making the most of the trial

Fivetran’s 14-day account trial begins around the transition into incremental syncing, rather than simply when you register. It includes Business Critical features, which gives you room to explore but can make the cheaper plans seem more capable than they are.

Reviewer’s Notes

I’d use the trial with one source that reflects your normal workload. Keep track of the features you rely on, then compare them with Standard before choosing a plan. A feature available during the trial may require a more expensive subscription afterward.

Fivetran’s Connectors and Customization

Fivetran’s 700+ connectors are a major reason I’d consider it, particularly when a team has several business applications feeding the same warehouse. The potential benefit is fewer separate integrations to maintain.

However, a large catalog doesn’t guarantee that your exact workload is covered. A connector can support an application while missing an object or field that matters to your reporting. I’d put coverage of your essential data ahead of the headline connector count.

What I like about managed connectors

Managed connectors make Fivetran most attractive when your requirements match what it already supports. You can focus more of your effort on using the data, while the vendor handles its connector’s ongoing operation.

There are still plan restrictions. Specified Oracle and SAP connectors require Enterprise or Business Critical. That’s disappointing if Standard otherwise covers everything you need: one database requirement can change the whole purchase.

Where customization adds work

Connector SDK lets you run custom Python extraction code on Fivetran’s infrastructure. You own the source logic, state handling and fixes unless you buy the maintenance add-on. SDK also lacks native History Mode; incremental extraction and table-level re-sync depend on your implementation.

I like having that option for an unusual source, but it weakens Fivetran’s main selling point if most of your connectors need custom code. You would still be maintaining extraction logic, even with someone else operating the runtime.

I’d compare Airbyte when control is the priority. Its self-hosted route gives you another option, although operating the infrastructure becomes your responsibility. For a team trying to reduce maintenance, Fivetran’s prebuilt coverage is the stronger reason to buy.

How Well Does Fivetran Keep Data Updated?

Fivetran offers 15-minute scheduling on Standard and one-minute intervals on eligible higher-tier connections. I like having that choice, because a daily finance report and an operational dashboard need different levels of freshness.

The restriction is that upgrading doesn’t unlock one-minute syncs everywhere. Fivetran’s scheduling rules exclude Lite connectors and specified standard connectors, including Google Ads and Google Analytics 4. The default interval is six hours, so don’t assume a new connection is already running at your plan’s fastest setting.

A schedule also isn’t a delivery guarantee. A one-minute interval doesn’t prove that every record appears in the destination within one minute. I would judge freshness by the reporting task it supports, rather than the smallest number on the plan comparison.

Handling changes to your data

I like Fivetran’s automatic schema handling, which helps keep a connection aligned when the source structure changes. The Schema tab also lets you choose whether to allow new data, allow new columns only, or block new data.

Those controls matter when a newly added field shouldn’t automatically appear in a reporting warehouse. However, flexibility varies by connector: some schemas are only partly customizable, and certain tables can’t be blocked completely. Adding previously excluded data can also require a re-sync.

Compared with Airbyte’s cloud plans, Fivetran Standard’s 15-minute scheduling is attractive: Airbyte Standard lists hourly scheduling, while Plus lists 15 minutes. I’d compare the eligible plans for your actual source before deciding which offers better value.

Fivetran’s dbt and AI Features

Fivetran and dbt Labs completed their merger in June 2026, bringing data movement and transformation under the same company. I like the closer connection with dbt for teams that already use it, although it doesn’t make every service part of one price.

Transformations turn loaded source data into useful reporting models. Fivetran’s hosted approach runs them after loading, so you need to consider both model-run fees and the destination’s compute costs.

The September 2026 releases include a mix of available products and beta features:

Feature	Availability	My view
dbt v2	Generally available	Worth evaluating with an existing dbt project
dbt State	Generally available	Interesting for avoiding unnecessary model work
Fivetran Context Layer	Private Beta	Promising for AI context, but confirm access
Lake Compute	Private Beta	A separate evaluation from everyday ingestion

I’m most interested in features that make an existing workload easier to operate. dbt State, for example, checks what changed before deciding which work to run or skip. That is a more concrete benefit to assess than a broad promise of being ready for AI.

I wouldn’t choose Fivetran primarily for beta features. Buy it for ingestion and modeling you can use now, with new capabilities as a potential addition.

How Does Fivetran Handle Security and Support?

Fivetran has useful options for businesses that need more control over where data is processed. Hybrid Deployment runs processing in your environment, while Fivetran provides orchestration. It requires Enterprise or Business Critical and supports selected sources and destinations.

I like that compromise for teams that want managed coordination while keeping the data-processing environment inside their network. However, metadata and logs still go to Fivetran, and your team hosts the agent. The Hybrid Deployment model therefore brings its own operational responsibilities.

Business Critical adds controls such as customer-managed encryption keys and private networking. Those can justify the upgrade for a specific requirement, but I wouldn’t recommend paying for them simply because the plan sounds more comprehensive.

Free and Standard users start with an AI Support Assistant when contacting support through the portal, then have the option to open a ticket. Other plans go directly to the support request flow. Tickets can cover product support, account support or billing.

I like having billing and technical help in the same support flow, but I’m less keen on the extra assistant step for lower-tier users. If an interrupted connection holds up your reporting, access to the right person matters more than a convenient chat interface. I’d weigh the support arrangements alongside connector coverage when choosing a plan.

How Does Fivetran Compare to Competitors?

Fivetran would be my pick when managed source coverage is the priority. I’d put Airbyte alongside it for deployment choice, and Hevo on the shortlist when its event-based packages fit the workload.

Platform	Pricing approach	Who I’d recommend it for	Main drawback to weigh
Fivetran	Connection usage plus separate service charges	Teams reducing supported-pipeline maintenance	Usage forecasting and connector-specific limits
Airbyte	Entry cloud credits; enterprise capacity pricing	Teams comparing cloud and self-hosted deployment	Operating work and plan restrictions
Hevo	Event-based packages	Teams whose sources fit its managed catalog	Event allowances and source coverage

• Airbyte is the alternative I’d consider first for deployment flexibility. Managed Standard starts at $20 monthly with five credits; Plus starts at $189 with 40. Both are volume-metered, while Pro and Enterprise Flex use capacity pricing. I like that deployment choice, although the different billing models make direct price comparisons harder.
• Hevo is worth considering if you prefer to compare event-based packages. Its Starter plan lists $299 monthly, or $265 monthly with annual billing, at the entry selection. There’s also a Free plan for limited connectors. I’d compare its coverage with your actual sources before treating the headline price as a saving.

My biggest caution is that MAR, credits and events are different billing units. Matching the same number across three pricing pages won’t tell you which service is cheaper.

I’d compare one representative pipeline across all three, with the same source data, destination and freshness requirement. Then include the engineering work each option leaves behind. Fivetran can justify a higher software bill if it removes recurring work, but a simpler alternative may be better value when it already supports everything you need.

How I Reviewed Fivetran

I evaluated Fivetran’s published pricing, connector documentation, deployment options and support process, then compared its fit with Airbyte and Hevo. My recommendations weigh the recurring work a team wants to reduce against the costs and responsibilities it keeps. Pricing examples illustrate billing rules, rather than measured account charges. Prices and product availability were checked in September 2026.

Fivetran Review: Should You Use It?

I recommend Fivetran for teams that want to spend less time maintaining supported pipelines. Its connector range, automatic schema handling and deployment options give it a convincing role in a growing data stack.

Pricing is the reason I’d hesitate. Separate connection curves and history-mode behavior can make costs harder to predict, so I wouldn’t commit based on a quiet trial week or the Free plan’s allowance.

Start with the free trial and choose Standard if your paid workload fits its features. Consider Enterprise for a specific requirement. If your priority is deployment control, compare Airbyte; if Hevo covers your sources, include its event-based plans in the cost comparison.

Frequently Asked Questions

Is Fivetran free?

Yes, it has an ongoing Free plan with usage limits, separate from the 14-day account trial. You need to select Free explicitly, and repeated breaches can interrupt pipelines.

Does every update count again toward MAR?

Ordinary repeated changes to the same row identity generally count once per month. History mode and source-specific behavior introduce exceptions, while captured deletes can count.

Is Fivetran ETL or ELT?

ELT describes its workflow of loading data before transforming it in the destination. You’ll still need models that turn those source tables into useful reporting data.

Can I add an unsupported source?

Connector SDK provides a Python route, with development and maintenance responsibilities. I’d first check whether the work involved still makes Fivetran good value for your team.