That’s what every CFO says when your Fivetran bill inexplicably doubles. Or when that “free” feature suddenly requires a $50K annual add-on. Or when your monthly data integration costs balloon from $5K to $40K with no warning.
If you’ve been through a SaaS pricing surprise, you know the feeling: betrayal mixed with resignation. Of course they raised prices. Of course the “simple” pricing model turned complex. Of course the features you need are now in a higher tier.
It’s the enterprise software playbook, executed with ruthless efficiency.
But the Fivetran-dbt merger takes this to a new level. Understanding their pricing trajectory isn’t just about historical grievances—it’s about predicting what comes next and protecting your budget.
Let’s walk through exactly how modern data integration pricing “evolves” (read: escalates), using Fivetran as our case study, and then talk about how to protect yourself.
Fivetran’s early pitch was seductive: “Pay only for what you use. No hidden fees. Transparent pricing based on Monthly Active Rows.”
Monthly Active Rows (MAR) was brilliantly simple: count the unique rows that were added, modified, or deleted in your data warehouse each month. More data = more money, but it scaled predictably.
Early customers loved it:
The pricing calculator on their website lets you estimate costs in minutes. Sales cycles were short. CFOs approved budgets. Everyone was happy.
The genius was this: as you built more pipelines, added more connectors, and integrated more of your data ecosystem, you became deeply dependent on Fivetran.
Your analytics dashboards relied on their pipelines. Your data warehouse refresh cadence assumed their reliability. Your team built skills and workflows around their platform.
Switching costs grew every month, even though you barely noticed.
And then, once they had you hooked…
In March 2025, Fivetran fundamentally changed their pricing model. Out went account-wide MAR with volume discounts. In came per-connector billing with separate cost curves for each data source.
What This Actually Meant:
Before: If you had 10 connectors and moved 50 million MARs total, you got volume discounts on the entire 50 million.
After: Each connector has its own cost curve. Those 10 connectors moving 5 million MARs each? You’re now in the expensive part of the curve on ALL of them instead of the cheap part of one combined curve.
The Impact:
Some customers reported price increases of 4-8x virtually overnight. The same data, the same workload, suddenly cost 4-8 times more because of how the pricing was calculated.
One user on G2 complained: “Customers often reported discrepancies between actual usage and the amount Fivetran billed them for.”
Another wrote: “Predictability and transparency have been the foundation of many users’ satisfaction with Fivetran, but with this new model, users are finding that their monthly bills are far harder to estimate.”
The Justification:
Fivetran claimed this gave customers more “granular control” and “transparency into connector-level costs.” In reality, it shifted risk from Fivetran (who could predict aggregate usage) to customers (who couldn’t predict per-connector usage).
Not content with one pricing change, Fivetran introduced another just two months earlier.
Starting January 1, 2025, Fivetran began charging for:
These were previously free (or at least included). Now they’re billed by “successful model runs”—with a “generous” 5,000 free runs per month before charges kick in.
The Justification:
“This change allows us to continue enhancing the Fivetran platform while maintaining the ease of use our customers expect.”
Translation:
“We found another thing to charge you for.”
Let’s call this what it is: The SaaS Pricing Escalation Playbook
It’s not unique to Fivetran. But they’ve executed it with particular efficiency.
Which brings us to October 2025 and the dbt merger.
When a vendor merges with or acquires complementary products, customer negotiating leverage decreases. Here’s why:
Reduced Competition: Fivetran and dbt were independent. If you were unhappy with one, you could switch just that piece. Now they’re unified. Switching either requires replacing both.
Bundling Pressure: Expect future pricing that “incentivizes” (read: forces) you to use both products. “Oh, you want to use dbt with a different ingestion tool? That’ll be the expensive tier.”
Consolidated Renewal Cycles: Instead of negotiating two separate vendors, you’re negotiating one larger vendor with more leverage.
Cross-Subsidy Pricing: Profitable products subsidize less profitable ones. If dbt transformations are more commoditized than ingestion, guess which product’s pricing will rise to “cover the costs” of the other?
The data engineering community remembers what happened with previous acquisitions
Salesforce + Tableau (2019):
Google + Looker (2020):
One Reddit user summed up the concern: “History isn’t encouraging. Remember when Salesforce bought Tableau? Or Google acquired Looker? Both saw stagnant innovation and climbing prices. The community is worried this merger follows the same playbook.”
Another put it more bluntly: “Fivetran recently raised prices 4-8x for some customers. Now they’re acquiring dbt. You don’t need a crystal ball to see where this might go.”
The pattern is clear: M&A reduces customer leverage and accelerates pricing escalation.
When evaluating data integration platforms, most teams focus exclusively on license fees. But that’s like buying a car based only on the sticker price while ignoring gas, insurance, maintenance, and repairs.
Let’s look at the total cost of ownership:
Platform Licenses: What you actually pay Fivetran, dbt, or other vendors. This is visible and quantifiable.
Compute and Storage: Data integration requires processing power and storage. With Fivetran’s model, you pay for their processing plus your warehouse costs for storing the data they move.
Support Tiers: Basic support is inadequate for production workloads. Enterprise support with actual SLAs costs significantly more.
Training: Your team needs training on each tool. Multiply this across all the point solutions in your modern data stack.
For a mid-sized company, direct costs might be:
Total direct: $600K/year
Maintenance and Integration: Someone needs to keep all these tools working together. For a typical modern data stack, estimate 2-3 full-time engineers just on maintenance.
At $150K fully-loaded cost per engineer, that’s $300-450K/year just keeping the lights on.
Incident Response: When pipelines break at 2am (and they will), someone needs to diagnose which of your eight vendors is causing the problem and fix it.
Performance Optimization: Each tool requires ongoing tuning. Fivetran sync frequencies, dbt model efficiency, Snowflake query optimization, Airflow resource allocation—it never ends.
Duplicate Efforts: Different teams building similar capabilities because the tooling doesn’t enable reuse.
Coordination Overhead: Meetings to align on changes across tools, rollout schedules that don’t conflict, incident post-mortems involving multiple vendors.
Add it up: Another $500K+/year in indirect costs.
Delayed Projects: “We’d love to build that AI use case, but we need to fix our data pipelines first” is a sentence that costs millions in lost revenue or competitive advantage.
Competitive Disadvantage: While you’re wrestling with tool sprawl, your competitors with better infrastructure are moving faster.
Talent Retention: Good engineers don’t want to spend 70% of their time on maintenance. They leave for companies with better tooling.
Failed Initiatives: How many projects got started but abandoned because the data infrastructure couldn’t support them?
These costs are harder to quantify but often larger than direct and indirect costs combined.
For our mid-sized company example:
Total: $1M+/year minimum
And that’s before Fivetran’s pricing changes add another 4-8x multiplier.
If you’re already locked into contracts with vendors who have a history of pricing surprises, here’s how to protect yourself:
Monitor the Market: Join data engineering communities on Reddit, Slack, LinkedIn. When pricing changes hit, they’ll talk about it before official announcements.
Track Your Usage Patterns: Don’t wait for the bill. Monitor your MAR growth, connector count, model runs, and other usage metrics continuously.
Build Pricing Models: Create spreadsheets that model different scenarios. “If our data grows 50%, what happens to our Fivetran bill under the new pricing?” “What if we add 10 more connectors?”
Follow M&A Activity: When your vendor acquires someone or gets acquired, assume pricing changes are coming within 12-18 months.
If you’re coming up for renewal, negotiate aggressively:
Multi-Year Price Locks: “We’ll commit to 3 years, but only if you guarantee our per-unit cost won’t increase.”
Volume Commitments with Discounts: “We’ll commit to 100M MARs annually. Lock in our per-MAR rate for the contract term.”
Most Favored Nation Clauses: “If you offer better pricing to similarly-sized customers, we automatically get it.”
Grandfather Clauses: “Any pricing model changes don’t apply to us during our contract term.”
Performance Guarantees: “If uptime falls below 99.9%, we get credits.”
Exit Clauses: “If pricing increases beyond X% at renewal, we can terminate without penalty.”
Most enterprise vendors will negotiate on these points if you’re large enough and willing to walk away.
The best negotiating position is having credible alternatives:
Vendor-Independent Schemas: Don’t let vendors transform your schemas in ways that make you dependent on their specific format.
Abstraction Layers: Use data products or intermediate representations that work with multiple systems.
Parallel Pipelines: For critical data, run pipelines through two different systems. Expensive but gives you exit options.
Modular Architecture: Design so you can swap out individual components without rewriting everything downstream.
Document Your Dependencies: Know exactly what would break if you switched vendors. The act of documenting usually reveals ways to reduce dependencies.
Treat your data integration vendors like the businesses they are:
Annual Vendor Reviews: Formally assess pricing, reliability, innovation velocity, and satisfaction.
Cost Per Value: Don’t just track total costs. Track cost per pipeline, cost per data source, cost per transformation. When these metrics climb, investigate.
Alternatives Analysis: Every year, seriously evaluate 2-3 alternatives. Even if you don’t switch, knowing what’s available keeps you informed.
Relationship Diversification: Don’t become dependent on one sales rep or one support engineer. Build relationships across the vendor organization.
Backup Plans: Have a documented “what if we had to switch in 90 days” plan. Update it annually.
Sometimes the best move is to recognize that incremental optimization of a fundamentally broken architecture isn’t worth it. Consider platforms that solve the pricing problem structurally:
Transparent Tier-Based Pricing: Instead of unpredictable usage-based models, some platforms offer straightforward tiers. “Under 1TB/month: $X. 1-10TB: $Y.” No surprises.
All-Inclusive Features: No “premium connectors” or “transformation fees” or “extra charges for orchestration.” Everything’s included.
Predictable Growth Curves: As your usage grows, you know exactly when you’ll cross into the next tier and what it will cost.
No Retroactive Pricing Changes: Your rate is locked for your contract term. New customers might pay different rates, but your grandfathered in.
What does good pricing look like in data integration? Here are the principles:
You should be able to calculate your monthly bill on a napkin. If you need a pricing calculator with six input variables, the model is too complex.
Your bill shouldn’t vary by more than 10-15% month to month unless your usage patterns significantly change. If normal growth causes wild cost swings, the model is broken.
Costs should scale with value. If you’re getting 10x more value from the platform, paying 10x more is reasonable. But paying 10x more for the same value? That’s just extraction.
You should understand exactly what you’re paying for and why. No hidden fees, surprise charges, or “contact sales for pricing.”
Core capabilities shouldn’t be arbitrarily split across pricing tiers. Charging extra for “premium connectors” or “advanced transformations” is artificial segmentation.
Some platforms take a fundamentally different approach:
Tier-Based Pricing: Based on total data volume and number of sources, not complex per-connector calculations.
All Features Included: Transformations, orchestration, quality checks, governance—all part of the base platform.
Transparent Tier Transitions: Clear documentation of when you’ll cross into the next tier and what the new cost will be.
Multi-Year Discounts: Commit for longer, get better rates—but your rate is locked for the term.
No Retroactive Changes: Current customers keep their pricing even when new pricing is introduced.
Organizations that switch report dramatic savings:
Based on historical M&A patterns and Fivetran’s recent pricing behavior, here’s what to expect:
Bundling Pressure: “Great news! We’re offering a special package for Fivetran + dbt Cloud customers.” The bundle will be cheaper than buying separately—but more expensive than what existing customers currently pay for both.
Feature Migration: Key features from one product will be “enhanced” by requiring the other. “To use Fivetran’s new continuous sync feature, you’ll need dbt Cloud for quality validation.”
Tier Consolidation: Current separate product tiers will be merged into new “platform” tiers. Most customers will find themselves pushed into higher tiers.
Legacy Pricing Elimination: “We’re retiring our legacy pricing models. All customers must migrate to our new unified platform pricing by [date].”
dbt Core “Prioritization”: While remaining open source, dbt Core will get minimal investment. Key innovations will be “dbt Cloud exclusive.”
Cross-Product Requirements: Certain Fivetran connectors will require dbt Cloud. Certain dbt features will require Fivetran ingestion.
Material Price Increases: Once most customers are locked into the new pricing, rates will increase. “Due to continued innovation and market conditions, we’re adjusting our pricing.”
Acquisition Integration Costs: “As we complete our platform integration, some price adjustments are necessary to reflect the value of the unified solution.”
This isn’t speculation. It’s pattern recognition based on decades of enterprise software M&A.
If you’re a Fivetran customer, a dbt customer, or use both:
The Fivetran-dbt pricing story isn’t about one company being uniquely evil. It’s about understanding how enterprise software businesses work.
They land with attractive pricing. They expand by building dependencies. They extract by changing the rules. And they rationalize it as necessary for sustainability.
M&A accelerates this pattern by reducing competition and customer alternatives.
For data leaders, the lesson is clear: Price today is not price tomorrow. The question isn’t whether your vendor will raise prices—it’s by how much and how quickly.
Your job is to:
The modern data stack failed partly because unpredictable, escalating costs made it economically unsustainable. The Fivetran-dbt merger doesn’t fix this problem—it makes it worse.
Choose vendors and architectures that solve the pricing problem structurally, not ones that promise to fix it despite decades of industry evidence to the contrary.
Your CFO will thank you.
In the News: Watch Nexla CEO Saket Saurabh analyze the Fivetran-dbt Labs merger with John Furrier on theCUBE + NYSE Wired, exploring its strategic impact on the evolving data ecosystem.
Fivetran and Nexla are leading data integration platforms, but they take different approaches. Learn how they compare on features, deployment, and governance to find the right fit for your data strategy.
The modern data stack has failed. The Fivetran–dbt merger highlights tool sprawl, rising costs, and integration complexity, forcing data leaders to rethink their infrastructure strategy. Choose wisely.
Webinar 
