PetScreening's dbt Pipeline Ends in a Printed Pet Tag, Not a Dashboard

Most data pipelines finish their work on a screen. PetScreening's ends in a mailbox. At dbt Summit 2026, Nick Fernandez, the company's manager of data analytics, and William Guicheney, principal analytics engineer at Aimpoint Digital, described a build that turns modeled data into a physical pet tag.

PetScreening runs pet policy management software for property managers, handling vaccination records, risk assessments and assistance animal requests. Its FIDO Alert service, often called an "Amber Alert for pets," gives members a QR-coded tag tied to their phone number. When a pet goes missing, alerts go out to everyone within a 10-mile radius. The service has reached 1.6 million users and 2.2 million pets and has reunited more than 75,000 lost pets with their owners.

Tag fulfillment began with outside vendors and manually downloaded CSVs passed back and forth by email with no verification, creating a three-month backlog. Fernandez called the setup "unreliable, expensive, and slow, which is a very bad trifecta of qualities." The team brought manufacturing in-house and plugged it into its dbt project, treating the printing machine as just another exposure.

FIDO Alert has been through three iterations, moving from a storefront bolted onto PetScreening's core product to a dedicated Shopify storefront to a setup that now prints more than 1,000 tags a day on two machines. A standard medallion architecture let models be swapped without interrupting production. "It was really an overnight switch of updating the stream of data going into the actual machine," Fernandez said.

Reliability carries unusual stakes here. "For us, bad data doesn't result in bad numbers in a dashboard. It results in missed prints, reprints," Fernandez said. A duplicate join that reprints a tag 100 times wastes hundreds of tags and plenty of postage, so dbt tests for uniqueness, non-null values and custom checks at each stage, wired into Slack for fast diagnosis.

Reuse drives efficiency: the same gold-level model feeding the printer also powers the growth team's dashboard. "It's truly one source of truth," Fernandez said. PetScreening now runs more than 1,000 dbt models, has seen a 32x increase in data freshness and cut the time from a new data request to an actionable answer by more than 90%. The team is now building toward a semantic layer, with an eye toward letting AI agents work against well-documented, well-tested pipelines.

Source: dbt Labs Blog