MAR isn't row count, and it isn't sync frequency. It's distinct rows that changed in a calendar month, counted once. Here's Fivetran's own definition, with real numbers from its pricing page.
No credit card. 14 days. Cancel in one click.
Quick answer: Fivetran bills on Monthly Active Rows (MAR): distinct primary keys that see an insert, update, or delete in a calendar month, counted once no matter how many times that row changes again. Fivetran's pricing page adds a $5 base charge per standard connection, a free tier up to 500,000 MAR, and published connector examples from $15.88/month (Google Analytics 4, 21,993 MAR) to $424.54/month (Marketo, 847,574 MAR). Sync frequency doesn't change your MAR; how many distinct rows actually change does.
Fivetran's own documentation defines a Monthly Active Row (MAR) plainly: "the number of distinct rows synced from the source system to your destination system in a given calendar month," tracked by distinct primary key. A row becomes active the moment it sees an insert, update, or delete, and Fivetran counts it once for the whole month no matter how many more times it changes. Their own worked example makes the mechanic clear: a table with rows a, b, and c starts at 3 MAR; if c updates repeatedly that same month, it's still 3 MAR, because c was already counted. MAR only climbs when a row that hadn't changed yet, like a, changes for the first time that month.
That single design choice is why Fivetran's bill doesn't move in lockstep with how often a pipeline runs. Running a sync every five minutes instead of once a day doesn't multiply your MAR; what drives the number up is how many distinct rows actually change in a month, and how big your tables are to begin with.
| Plan | What it includes | Row limit |
|---|---|---|
| Free | Connections | 500,000 MAR/month, no charge |
| Free | Activations (reverse ETL) | 3,500 MAR/month |
| Free | Transformations | 5,000 monthly model runs |
| Standard | Usage-based, most common paid tier | $5 base charge per connection, then per-MAR usage across a 1–1,000,000 MAR band |
| Enterprise | Standard features plus advanced capabilities | Custom, quote-based |
| Business Critical | Highest protection tier | Custom, quote-based |
Fivetran's pricing page also lists real, published per-connector examples at Standard-tier rates: Google Analytics 4 at 21,993 MAR runs $15.88/month; Google Ads at 88,240 MAR runs $48.67/month; Facebook Ads at 34,479 MAR runs $22.06/month; and a heavier connector like Marketo at 847,574 MAR runs $424.54/month. Those four numbers, taken straight from Fivetran's own page, show how directly MAR drives the bill: roughly 4x the rows on Marketo versus Google Ads produces roughly 9x the monthly cost, because the per-MAR rate itself isn't flat across the whole usage curve.
Take the Google Analytics 4 example directly from Fivetran's pricing page: 21,993 MAR billed at $15.88/month works out to roughly $0.00072 per MAR at that connector's rate and volume. Compare that to Marketo's 847,574 MAR at $424.54/month, which works out to roughly $0.0005 per MAR, a lower per-row rate at higher volume, consistent with usage-based pricing that gets marginally cheaper per row as volume climbs within a connector. What stays constant across every connector is the $5 base charge Fivetran applies to each standard connection before any usage is added, so a dozen small connectors each generate their own base charge even if none individually crosses much MAR.
-- Rough monthly Fivetran bill for N connectors, each with its own MAR and rate -- (illustrative, not a Fivetran formula; consult Fivetran's pricing page for exact rates) SELECT connector_name, monthly_active_rows, 5.00 AS base_charge, monthly_active_rows * per_mar_rate AS usage_charge, 5.00 + (monthly_active_rows * per_mar_rate) AS estimated_total FROM connector_usage;
The honest answer is that you can't get an exact number without knowing your own tables' actual change rate, because MAR is driven by how many distinct rows change in a month, not by table size, sync frequency, or row count at rest. A 50-million-row dimension table that's almost entirely static, updated a few thousand times a month, generates a small MAR count. A 500,000-row transaction table where every row updates multiple times a day can generate close to its full row count in MAR every single month. Before committing to a connector, Fivetran's own usage dashboard (available once you've connected something, even on the free tier) is the only reliable way to see your actual MAR, not a guess based on table size.
Annual contracts get a published discount too: Fivetran's pricing page states annual discounts "start at 5%, with greater discounts available," reaching up to 22% at the numbers they publish, which matters if you're comparing a monthly Standard estimate against what an annual commitment would actually cost.
QueryFlow doesn't meter by row at all. Pipelines is $29.99/month or $199.99/year flat, whether a sync moves a thousand rows or several million, because the pricing model is a Mac app license, not a managed hosted service billed on data volume. That's a real tradeoff, not a strictly better deal: Fivetran's MAR model buys continuous managed infrastructure, connector maintenance across API changes, and (on paid tiers) support, none of which QueryFlow's flat price includes, since QueryFlow's syncs run on a schedule on your own Mac. For teams whose data volume is unpredictable or growing fast and want their infrastructure cost to scale with it, MAR pricing is defensible. For teams with a known, modest set of sources who'd rather have a fixed line item than a bill that moves with data volume, a flat price removes that uncertainty entirely.
The $5 base charge Fivetran applies per standard connection, stated on its own pricing page, is easy to overlook when comparing a single connector's usage cost, but it compounds quickly once a team is running many small connectors rather than a few large ones. Ten lightly-used connectors each generating a modest usage charge still carry ten separate base charges before any per-row cost is added, which is one reason consolidating several small sources behind fewer, larger connectors (where that's architecturally possible) can meaningfully change the total bill independent of MAR itself.
Fivetran's Transformations product, its dbt-based transform layer, is billed on a completely different unit than MAR: monthly model runs, not rows. Its published tiered rate on the pricing page runs free for the first 5,000 model runs a month, then $0.01 per run from 5,001 to 30,000, dropping to $0.007 per run from 30,001 to 100,000, and $0.002 per run beyond 100,000. A team evaluating total Fivetran cost needs to add this on top of connector MAR separately; the two don't share a billing pool, and a heavy dbt user with modest connector volume can end up with Transformations as the larger line item.
Fivetran's Activations product (its reverse-ETL offering, moving data out of the warehouse into SaaS tools) has its own MAR pool again, separate from connector MAR and separate from Transformations. The free plan's Activations allowance is capped at 3,500 MAR/month per Fivetran's pricing page, a much smaller free allowance than the 500,000 MAR free tier for regular connections, worth noting if you're planning to use Activations for anything beyond a small pilot.
None of the worked numbers on this page substitute for Fivetran's own usage dashboard once you've connected a real source; MAR depends entirely on your own data's change rate, which no outside estimate can know in advance. What these numbers do establish, directly from Fivetran's published materials, is the shape of the pricing model: a per-connection base charge, a per-row usage rate that varies by connector and volume tier, and two entirely separate MAR pools for Transformations and Activations if you use either.
No. Fivetran counts a row once per calendar month regardless of how many times it re-syncs, per its own definition. Running hourly instead of daily doesn't multiply the row count, though it can surface changes sooner.
It varies by connector and by where you sit in Fivetran's usage bands; Fivetran doesn't publish one flat per-row rate across all connectors. Its pricing page shows several worked connector examples with their own implied rates, but your own number depends on the specific connector and volume.
Fivetran's free plan covers 500,000 MAR for connections at no charge, which comfortably fits a handful of small tables with modest change activity. A single busy transactional table can exceed that on its own.
Yes, Fivetran's stated definition counts a row active on insert, update, or delete, the same way for all three.
QueryFlow charges a flat $9.99/month (Studio) or $29.99/month (Pipelines) regardless of row volume, with no MAR concept at all. Fivetran charges per connection plus usage that scales with how many distinct rows change each month.
14-day free trial, no card. QueryFlow Pipelines: $29.99/month, no row metering.
No credit card. 14 days. Cancel in one click.