Paylocity scales trusted data and AI with Fivetran + dbt

“Fivetran saves us almost a full headcount of someone having to write these pipelines and especially maintain them. When Fivetran finishes, our dbt jobs kick off, so we’re able to get the data in, transform it, apply the business logic, and get that final product to our analytics teams.”
— Farin Fukunaga, Manager of Data Engineering at Paylocity

Trusted by tens of thousands of companies, Paylocity provides HR and payroll software that helps organizations manage their people and operations.

Paylocity’s data journey began with a familiar challenge: as the company grew, critical customer data and business logic were spread across teams and systems:

• Marketing needed to understand campaign performance and customer acquisition.
• Sales needed visibility into the funnel.
• Customer Service needed insight across the customer journey.
• Finance needed consistent metrics it could trust.

Fivetran initially helped Paylocity solve the data movement side of that problem by automating pipelines from its go-to-market systems into BigQuery. What began as a marketing analytics initiative has since evolved into a broader enterprise data foundation powered by Fivetran and dbt. By automating data movement and standardizing business logic, Paylocity can give more teams faster access to trusted insights without adding equivalent engineering overhead — while creating the governed foundation its emerging AI use cases need.

Building a trusted view of the customer with Fivetran

When Andrew Wahl, Director of Marketing Analytics and Operations, joined Paylocity in 2019, he was a data team of one. He was tasked with helping marketing and sales understand performance, optimize spending, and build a more complete view of the customer journey.

At the time, spreadsheet-based processes and custom integrations made that difficult to scale. Wahl realized that manually building and maintaining pipelines from CRM and advertising platforms would consume the time and resources he needed to actually use the data to improve marketing performance.

Paylocity chose Fivetran to automate ingestion from systems including Google Analytics, LinkedIn Ads, Facebook Ads, and Microsoft Dynamics 365 into BigQuery. Instead of spending months building integrations, Wahl could connect new sources in minutes — giving a 1-person data function the leverage to support a rapidly growing go-to-market organization.

Centralizing that data gave marketing visibility into MQLs, SQLs, attribution, funnel performance, and customer acquisition costs. More importantly, teams could use those insights to change how they invested. When Paylocity identified rising acquisition costs, it reallocated resources and reduced customer acquisition costs by 20%.

Connecting product and billing data with marketing data also gave teams a more complete view of customers, helping Paylocity better understand who it was reaching and deliver more relevant campaigns.

“Fivetran has been a core part of allowing us to scale and a force multiplier in our data-driven processes across the organization. We would have spent six figures on other tooling and resources to achieve the same single source of truth for customer data that Fivetran enables.”
— Andrew Wahl, Director of Marketing Analytics and Operations at Paylocity

Scaling from marketing analytics to trusted metrics across the business

As Paylocity’s use of data expanded, moving data into one place was no longer enough. Different teams had their own code, processes, and definitions. A metric such as revenue could appear in 4 reports with different logic behind it, making it harder for teams to align on performance and increasing the work required to answer seemingly simple business questions.

Paylocity introduced dbt as the logic layer for its growing data environment. Definitions such as revenue and customer metrics can now be codified, documented, tested, and reused instead of being recreated independently by each team.

That turns business logic into a reusable company asset. Instead of searching across reports or relying on tribal knowledge to determine which number is correct, teams can work from shared definitions. It also reduces duplicated analytics work as Paylocity expands its data foundation across marketing, sales, customer service, and finance.

“dbt has really been the logic layer. When we’re sitting down thinking about revenue or thinking about what a customer is, we take that and then we get to put it into dbt as code. Whenever someone needs to find a metric, the first place that we point them to is dbt.”
— Farin Fukunaga, Manager of Data Engineering, Paylocity

That shift has helped Paylocity evolve from federated analytics teams with separate processes toward a centralized foundation serving marketing, sales, customer service, and finance. Instead of teams looking across multiple places for the right data or definition, they can build from shared business logic.

Reclaiming engineering capacity, saving costs with Fivetran + dbt

Today, Fivetran and dbt automate the path from source systems to trusted, business-ready data. Fivetran handles ingestion, and once a sync completes, downstream dbt jobs transform the data and apply Paylocity’s business logic before making it available to analytics teams.

That automation has a significant impact on engineering capacity. Fivetran saves Paylocity nearly 1 full headcount that would otherwise be dedicated to writing and, critically, continually maintaining pipelines. Instead of spending that capacity keeping data moving, Paylocity can direct engineering resources toward delivering data products and solving new problems for the business.

Paylocity is also using dbt State to eliminate unnecessary recomputation. Previously, reviewing another engineer’s work could mean rebuilding an entire schema and waiting as long as 30 minutes. By reusing previous work, Paylocity has reduced compute costs by 20–30% while accelerating development.

The payoff is more than lower infrastructure costs. Engineers spend less time waiting for jobs, rebuilding models, and maintaining pipelines, giving them more capacity to get trusted data to stakeholders faster and explore higher-value experiences such as chatbots and AI agents.

Turning trusted business logic into context for AI

TThe same foundation Paylocity built to make analytics more consistent is now becoming a critical building block for AI.

Paylocity had already begun developing its semantic layer to establish one approved definition for metrics such as revenue and pipeline. Instead of requiring every team to reconstruct those definitions from underlying tables, joins, and filters, Paylocity can codify the business logic once.

For AI agents, that changes the problem entirely. An agent no longer has to independently determine what “revenue” means or how to calculate it. It can access the same approved logic Paylocity’s teams already trust — helping the company scale AI without creating a new layer of inconsistent definitions.

“Instead of having to define, if you want revenue, go to this table and do this join and filter by this other thing, it’s, ‘Hey, look at the semantic layer, search something called revenue.’ It has all of our logic, all of our thinking there, and then any agent that plugs into the semantic layer has all of that right at its fingertips.”
— Farin Fukunaga, Manager of Data Engineering, Paylocity

For Paylocity, the fundamentals of trusted AI are similar to the fundamentals of trusted analytics: structured data and alignment with the business. The difference is scale. Instead of a marketer asking for a query, the consumer may now be an agent capable of accessing and acting on data much faster.

With Fivetran automating access to source data and dbt turning it into governed business logic, Paylocity can give both people and agents the same trusted foundation — helping AI move faster without leaving business context behind.

Keeping the data foundation flexible as AI evolves

The AI landscape is moving quickly, and Paylocity doesn't want the technology choices it makes today to limit what it can adopt tomorrow.

That makes open data infrastructure increasingly important. Paylocity sees value in being able to choose when to buy technology, when to build internally, and which agents, APIs, or platforms best fit the business as new capabilities emerge.

“If a year down the line, someone releases something new that’s super cool and interesting, we aren’t locked into that one vendor. We have the options to choose what we want to invest in and choose what’s best for our business.”
— Farin Fukunaga, Manager of Data Engineering, Paylocity

That flexibility helps Paylocity protect the investments it has already made in trusted data and business logic. Rather than rebuilding its foundation around every new AI tool, the company can evolve the technologies around it and choose the capabilities that deliver the best combination of performance, cost, and business value.

[CTA_MODULE]