MAR pricing vs event pricing: Which one actually costs you more

One Fivetran customer on Gartner Peer Insights said it in one line. The team “raised costs by 2x despite us being on an unlimited plan.” That is the MAR model working the way it was built to work. MAR pricing and event pricing both charge for the rows that change in your destination. But they charge for that change in very different ways. 

MAR, short for Monthly Active Rows, charges per connector, per million rows. The rate resets every time you add a new source. Event pricing bundles the same row changes into flat, simple tiers. One MAR is not the same as one event. Modeled across real usage levels, event-based billing comes out 40 to 60 percent cheaper. This gap has nothing to do with how 

This blog explains why MAR pricing and event pricing charge different amounts for the same data. It shows when each one works best. And it gives you a quick way to check which one costs less for you.

What MAR pricing actually counts

Fivetran’s pricing model defines a Monthly Active Row as any row that gets added, changed, or deleted in your destination in one month. Update the same row five times in thirty days and it still counts as one. That part makes sense.

The confusing part is how the bill gets built. Since Fivetran’s 2025 pricing update, MAR is counted per connector, not per account. This ended the volume discount teams used to get by combining usage across sources. A company running twelve connectors now has to hit the discount level on each one by itself, and that almost never happens. Every connector, even a small one, also carries a $5 monthly minimum. Twenty small sources cost $100 a month before a single useful row moves.

Then there is normalization. Fivetran breaks nested and semi structured data, like JSON files and deeply nested API data, into many destination rows. A million source records with nested fields can turn into ten million or more MAR after this process. HubSpot is the classic example. One contact update can trigger row changes across contacts, deals, marketing emails, and every custom field tied to that record. You are not paying for the data. You are paying for how many rows Fivetran’s system creates from it.

What event-based pricing actually counts

An event means the same thing: one insert, update, or delete landing in your destination. The difference is how it gets billed. Instead of a live per-row meter that resets per connector, event pricing sells flat monthly tiers. You pay for a block of events and use it across every connector and every table, with no per-source math. Transformations run inside the same plan instead of billing separately. When you add a new connector, you are not starting a new discount curve from zero. You are just using more of the tier you already bought.

The predictability does not come from your data changing less. It comes from the pricing model not punishing you for where the change happens.

Difference between MAR Pricing and Event Pricing

What changes	How it affects MAR pricing	How it affects event pricing
More connectors	Every connector gets billed on its own, so it is harder to get a lower rate even when your total data volume is high.	All your connectors share the same bucket of usage, so adding more of them does not cost you a discount.
Nested data (JSON, HubSpot, Salesforce, Stripe)	When your data has nested fields, one record can turn into several rows, and you get charged for every one of those rows.	No matter how complicated the data looks, it still counts as just one event.
Same volume moved	You end up paying a higher price for moving the exact same amount of data.	You pay 40 to 60 percent less for moving that same amount of data.

Humi’s data team saw this firsthand. After moving off MAR pricing, they doubled their data volume and cut their bill by 25 percent. 

“We\'re handling twice the data at 25% lower cost, and my team finally gets to work on the interesting stuff instead of babysitting pipelines.”

Sean Froese

Director of Data and Analytics

Read the Full Case study

Which is Cheaper for ETL: MAR or Event Pricing? 

Event pricing is cheaper in almost every case of ETL pricing. Modeled across usage levels from 500K to 1B rows moved, it costs 40 to 60 percent less than MAR pricing for the same data. The only time MAR can match it is when your data barely changes, and you run just a few connectors. Once you add more connectors, or your data starts changing a lot, MAR pricing costs more, and the gap keeps growing.

How to Evaluate Which Pricing Model is Best for your Stack

Skip the vendor pricing calculators for a minute. Walk through these points first, using your own data, not a sales estimate. 

Connector count

Count your connectors and think about how many you will add later. More connectors added over time works against you under MAR pricing, since every connector is billed on its own. 

Data change rate

Check how often your data actually changes, not just how much you have. A table that updates constantly costs far more under MAR than one that barely changes. 

Nested data share

Look at how much of your data is nested or has many fields, like JSON files or data from HubSpot, Salesforce, or Stripe. This kind of data costs more under MAR because of how it gets split into extra rows. 

Sync speed

Check how fast you need your data to sync. Faster sync speeds on Fivetran can force you onto a pricier plan before your MAR bill even comes into play. 

Data stability

Think about how stable your data is overall. If your sources are mostly quiet, low-change tables on a fixed schedule, MAR pricing can work fine and land close to a flat tier. 

Workload type

Think about your workload type. If you run CDC, streaming, or high-frequency SaaS sources, event pricing will almost always cost less as you scale.

Real usage check

Pull your actual usage from the past three months. Compare both models using that real number, not a quote from a sales call.

Event or MAR Pricing: If you are a Startup, Mid-Market, or Enterprise

Factor	Startup	Mid market	Enterprise
Typical connector count	3 to 6, mostly SaaS tools like Stripe and HubSpot	10 to 20, with new ones added most quarters	20 or more, across databases, SaaS, and internal systems
Data change rate	Low to moderate, most tables update in batches	Rising fast as more of the business runs through the warehouse	High, with CDC and real-time sources common
Compliance and security needs	Minimal	Growing, but rarely the deciding factor yet	High, can push you to a pricier tier regardless of pricing model
How MAR pricing affects	Won’t affect until you are on Free tier	With more connectors and an increase in volume of data, the bill becomes exponentially high	Compounds across connector cost, resync or backfill costs, and unpredictable MAR pricing
Best fit	Event pricing with no connector costs	Event pricing that bills only for the rows modified, excluding backfills or resync	Event pricing wins because it avoids all the compounded costs in the MAR pricing model

When to Choose the MAR Pricing Model? 

MAR pricing works well when your data barely changes. Think stable, low-change tables, like a warehouse fed by a few internal databases on a fixed schedule. In that case, MAR’s per-row cost can land close to a flat tier, and you still get Fivetran’s full connector list and support. The model punishes change, not volume, so quiet data does not get hit hard. 

When to Choose the Event-based Pricing Model? 

Choose event pricing when your data changes a lot. This includes CDC from live databases, streaming sources like Kafka or Kinesis, fast-moving SaaS APIs, and sources heavy in nested or JSON data. It also wins when you run many connectors and keep adding more, since MAR’s per-connector discount reset works against you at that point. Teams growing both connector count and data change see the biggest gap, often past that 40 to 60 percent baseline.

Why Hevo is the Best Alternative to MAR Pricing 

Event pricing is the model Hevo runs on, and it is built to fix the exact problems covered in this piece. 

• Consistent pricing: Every event, an insert, update, or delete, counts the same no matter which connector it comes from or how the source data is shaped. 
• No stacked minimums: There is no $5 minimum stacking on small connectors. 
• No connector penalty: Adding a new source does not reset your discount level the way it does under MAR pricing. 
• Transformations included: Transformations run inside the plan instead of billing separately. 
• Safe migration: Hevo replicates the exact schema Fivetran was producing, same table names, same column names, same data types, so your dbt models and BI dashboards do not need to change. 
• Low risk switch: Fivetran can stay live the whole time you are testing, and most teams complete the switch in about two weeks. 

FAQ