DonorBureau helps nonprofits while cutting costs and improving ML efficiency with BigQuery ML and SADA

Engaging key social advocates with data-rich precision

For nonprofits and civics-based organizations, deepening social impact hinges on reaching the right supporters—those who will not only donate but also advocate, volunteer, or take action. With vast amounts of donor and volunteer data to sift through, pinpointing the most committed activists is both expensive and complex, with uncertainty about which engagement methods will drive meaningful, long-term support. That’s where DonorBureau comes in. 

“We use rational thought to take the time, cost, and risk out of finding the best advocates for organizations,” says Brian Johnson, Founder and COO of DonorBureau. “That could mean soliciting donations, recruiting volunteers, or getting people to call their congressperson.” 

By leveraging data science, DonorBureau helps organizations connect with their most active supporters in a cost-effective way. 

Business challenge

Escaping rising costs and vendor lock-in

While DonorBureau was optimizing outreach for their clients, they faced growing internal challenges. Machine learning (ML) models, critical to their data-driven approach, relied on a third-party software platform with rising costs and operational inefficiencies.

DonorBureau’s ML process required moving data between BigQuery and an external platform, leading to inefficiencies and potential failures. “It involved a lot of moving files, spinning up compute resources, and managing servers,” says Johnson. “More steps for things to break.”

Costs were another major issue. “Every contract negotiation, the price kept going up,” says Johnson. “We were captive to that solution.” 

This unpredictability complicated long-term planning, but when Google Cloud introduced ML capabilities directly into BigQuery, DonorBureau saw an opportunity. “We’ve used BigQuery for years, and every year, it gets better and cheaper,” says Johnson. “We wanted to see if BigQuery ML could match our existing solution while reducing costs.”

Solution

Transitioning to BigQuery ML for cost efficiency and performance

For DonorBureau, BigQuery ML offered a compelling alternative to their existing ML platform. “The cost of using BigQuery was effective and predictable,” says Johnson. “We didn’t have to worry about managing servers or indexing. We just wrote SQL code, and it worked.” 

The ease of integration meant that within a day, DonorBureau was able to migrate their existing SQL-based workflows with minimal disruption. But switching platforms wasn’t just about cost. BigQuery ML had to match the performance of their previous solution. 

“The models we built in BigQuery ML were just as good, and the evaluation metrics were on par,” says Johnson. “And because it was fully embedded in BigQuery, we could integrate it directly into our workflows.”

Navigating technical challenges with SADA’s expertise

Despite the smooth transition, DonorBureau quickly realized that unlocking the full potential of BigQuery ML required deeper technical expertise. Enter SADA, An Insight company and multiple-time Google Cloud Partner of the Year. Initially referred through Google Cloud, SADA’s BigQuery gurus helped DonorBureau navigate challenges that weren’t always addressed in standard documentation. “Some of this stuff isn’t in the official docs—it’s passed along through experience,” says Johnson.

Additionally, SADA’s participation extended beyond an advisory role. As DonorBureau scaled their operations, SADA:

• Provided key insights into optimizing query performance
• Refined infrastructure
• Ensured resources were allocated efficiently

When DonorBureau encountered bottlenecks—such as memory constraints or production scoring matters—SADA’s team worked closely with them to optimize code, adjust configurations, and fine-tune processes for stability and efficiency.

“SADA was invaluable when we needed technical help,” says Johnson. “They assisted us in rethinking our approach, provided concrete examples, and even connected us with Google Cloud’s internal BigQuery experts who guided us through advanced cost-saving strategies like reserved instances.”

With SADA’s support, DonorBureau was able to fine-tune their operations and set the stage for broader adoption of BigQuery ML. As they expanded their use of machine learning, new considerations emerged—particularly around long-term efficiency, production scalability, and maximizing cost-effectiveness across their entire data ecosystem. SADA was there for them.

Impact

Accelerating ML processing while cutting costs

As a result of SADA helping them migrate to BigQuery ML, DonorBureau significantly reduced costs while maintaining the quality and accuracy of their predictive models. The transition cut processing times in half, enabling customers to receive results much faster. “We were able to service our customers faster for lower costs while maintaining the quality that we had before,” says Johnson. 

In terms of cost savings, DonorBureau eliminated an annual $100,000 expense from their previous ML provider, replacing it with a far more cost-efficient, usage-based model in BigQuery. “Our cost of BigQuery modeling is about $10,000 per annum. It’s obscenely low,” says Johnson. 

While their BigQuery bill went up slightly, the savings from removing third-party software vastly outweighed any additional usage expenditures—amounting to a 10X cost reduction.

Streamlining data pipelines

Beyond cost reduction, the shift to BigQuery ML simplified DonorBureau’s infrastructure. Previously, data had to be transferred between BigQuery and an external ML platform, adding complexity and potential failure points. Now, everything runs within BigQuery, streamlining operations and reducing overhead.

“From our customers’ perspective, nothing changed operationally except that they get results faster,” says Johnson. “But on the back end, we removed a ton of unnecessary steps and reduced the risk of things breaking.”

Fostering a solution-finding collaboration

Coders at heart, DonorBureau’s team prides themselves on self-sufficiency. However, SADA played a crucial role in navigating key technological pathways that could have delayed or curtailed the migration. 

“If we didn’t get over those technical speedbumps, we would have never been able to get stuff into production,” says Johnson. “Having SADA there as a collaborator was key.”

Unlike traditional consulting engagements, DonorBureau approached SADA as a solution-finding cloud expert—engaging their experience when they encountered unknowns. 

“First, we play with new technology ourselves,” says Johnson. “Then when we get to a point of uncertainty, we push the SADA button and ask, ‘Hey, how do we handle this?’” SADA’s deep background with Google Cloud meant that no query was too obscure. 

“No one has seen more unique situations in Google Cloud than SADA has,” says Johnson. “Whatever you’re doing in Google Cloud, they’ve probably seen it before, and someone in the company has encountered that exact use case and can help you resolve it.”

Looking ahead: scaling with Google Cloud

With the migration complete, DonorBureau is now exploring reserved instances to optimize costs further. Additionally, they are experimenting with Google Cloud’s Gemini API to enhance their ML processes, integrating new AI capabilities into their data-driven models.

“SADA does what they say they’ll do and they’re very responsive,” says Johnson. “That’s really all you could ask for from a solution provider in this space.”

Achieving key outcomes

Overall, by working with SADA to expand their usage of BigQuery ML, DonorBureau was able to:

• Cut ML processing times by 50%, decreasing time to deliver results to customers  
• Eliminate a $100K third-party ML software expense—a comparative 10X cost reduction
• Simplify data pipelines by keeping everything within BigQuery
• Gain on-call access to SADA’s deep Google Cloud expertise
• Position themselves for future AI/ML advances, including Google Cloud’s Gemini API