How The Philadelphia Inquirer increases productivity and enables self-service with the dbt Semantic Layer

For the data team at The Philadelphia Inquirer, answering simple data questions fast, like, “Can I see a chart for Homepage traffic by day?” is essential for shaping editorial decisions, subscription strategies, and long-term growth initiatives.

But getting to those answers was far from simple. To get quick insights, business users relied heavily on the data team to translate their questions into SQL to query the data warehouse. Unfortunately, this led to organizational bottlenecks, inconsistent metrics, and barriers to scaling.

Rather than doubling down on custom dashboards, the data team took a new approach. Let’s explore how they used the dbt Semantic Layer and MetricFlow to enable true data self-service—and what other data teams can learn from their journey.

Too many dashboards, not enough clarity

Years ago, the Inquirer’s data team created denormalized data marts and dashboards, each stitched together with custom SQL.

The problem was, the codebase wasn’t flexible or easy to maintain.

Even small changes—like adding a new column to filter data—required going back into the SQL models, modifying the code, and deploying changes. Core metrics like “Total Pageviews” or “subscriptions” could be calculated in different ways (i.e. sum vs count), leading to confusion and a loss of trust in the data.

“Our denormalized layer became the most brittle part of our system,” says Brian Waligorski, Lead Data Engineer at The Philadelphia Inquirer. “Copying and pasting large blocks of SQL across models opened the door to countless errors and inconsistencies.”

Without a centralized source of truth, business users often resorted to exporting data into their own spreadsheets to do their calculations manually, and even added their own business logic. The result was fragmented reporting and misaligned metrics.

Ultimately, the process was a drain on resources.

“We’ve all seen the dreaded after-hours Slack ping to ‘pull a number real fast,’” comments Waligorski. “But every message interrupts what you’re doing to dive into SQL troubleshooting. We spent a lot of time reacting, instead of focusing on strategic work.”

A single source of truth for scalable self-serve

To break out of this cycle of ad-hoc requests and inconsistent reporting, the data team needed to centralize metrics to enable true self-service.

“Self-serve doesn’t just mean ‘analysts building dashboards, faster,’” Waligorski emphasizes. “It’s gaining direct access to trusted data faster. A solution like the dbt Semantic Layer reduces organizational bottlenecks and empowers users to interact with the data directly without the wait.”

By adopting the dbt Semantic Layer, the team built a centralized, governed single source of truth for their business metrics and logic that powers self-service across the organization tooling and systems. Here’s how:

• Centralized metric definitions. Metrics and business logic are now defined directly in dbt, right alongside the data models they rely on. This eliminates metric and logic discrepancies while simplifying maintenance: update a metric definition once, and it’s updated everywhere.
• Standardized naming conventions. The data team partnered with key stakeholders to define clear, consistent naming conventions and canonical metrics like “total users” vs “total viewers.” to reduce confusion and reporting discrepancies.
• Enabled self-service across every tool. The team integrated metrics into tools that meet users where they already work—including Steep (for product managers), Hex (for analysts), Google Sheets (for the finance team), and Exports/Saved Queries (for traditional BI consumption).

“Now, with the dbt Semantic Layer, we have a single, enforceable source of truth for our business metrics and logic,” says Waligorski. “It’s like a holy grail of analytics engineering, and it’s possible because of dbt.”

Faster insights, greater trust

By implementing the dbt Semantic Layer, the data team at The Philadelphia Inquirer has become more productive than ever.

For example, when they get a request, analysts no longer have to rewrite SQL or reverse-engineer metric logic. They can simply pull trusted metrics from their suite of governed tools, speeding up delivery and reducing friction.

“With the dbt Semantic Layer, our time-to-delivery for dashboards has gone down significantly,” reports Waligorski. “By reducing the back-and-forth between data engineers and analysts, we’ve become more efficient.”

The dbt Semantic Layer has also unlocked a new level of flexibility. Because it’s tool-agnostic—with 10+ out-of-the-box integrations and a robust API—the data team can deliver metrics wherever stakeholders need them. Rather than being tied to a single BI platform, the data team can deliver insights in whatever format best suits each stakeholder, whether that’s spreadsheets, visual dashboards, or notebooks. The Inquirer’s data team also saw a significant drop in metric errors and inconsistencies. Before, different teams often reported different numbers for the same metric, causing confusion, duplication of work, and long Slack threads trying to reconcile the truth. Stakeholders can finally rely on the numbers they see, increasing confidence in decision-making, trust in the data, and alignment across teams.

Takeaways and the road ahead

After a year of working with the dbt Semantic Layer, Waligorski has four pieces of advice for data teams looking to follow a similar path:

1. Normalize your models. Denormalized models limit flexibility—but refactoring for modularity pays dividends in the long run.
2. Avoid redundant joins. Let MetricFlow handle join logic and transformations to reduce code duplication. (But allow exceptions for specific, mission-critical edge cases.)
3. Be deliberate with naming. Agree on metric and business terminology early. It’s an investment of cross-functional work, but it will prevent confusion and rework later.
4. Start small. Pilot with key use cases, prove value, then expand adoption across departments and tools.

Looking ahead, the team is getting ready to launch AI chatbots for real-time metric queries, making it even easier for stakeholders to get answers faster directly in their AI systems. The Inquirer is also exploring anomaly detection and metric observability to proactively understand when key metrics spike or dip and why.

“Today, we can pull a strong set of metrics with low overhead with the dbt Semantic Layer,” says Waligorski. “That has freed up our capacity to focus on high-impact work and scale the business.”

Check out the full story on YouTube.

Curious how the dbt Semantic Layer can empower your team to move faster and trust their metrics?, Book a demo. We’d love to answer any questions about what dbt can do. You can also sign up for dbt to connect your data warehouse and start building today.