Billtrust Customer story
How Billtrust manages cloud, SaaS and AI costs across a multi-provider infrastructure
Billtrust runs a multi-cloud infrastructure supporting $1 trillion in trade credit volume. Mavvrik is where the company goes to understand what that infrastructure costs, proactively stay ahead of where it is going, and ensure every dollar is working as efficiently as possible across cloud, SaaS, and a rapidly expanding AI footprint.
2x
Saving at least twice what they spent on the platform in the first year
$1T
In annual trade credit volume
2,600+
Suppliers, on a network of more than 20 million buyers
The challenge: a modern stack without a unified view
Billtrust is the B2B Cash Generation Platform. It runs invoicing, payments, credit, collections, and cash application as one credit-to-cash cycle for more than 2,600 suppliers, on a network of more than 20 million buyers and over $1 trillion in annual trade credit volume. Its infrastructure runs across AWS and Azure, including containerized workloads on EKS, AKS, with significant spend in Snowflake, MongoDB, Datadog, and Confluent alongside its cloud providers.
As the business grew through acquisition and data center migration, so did the challenge of managing what that infrastructure cost. Getting meaningful cost data out of Azure was particularly challenging and time-intensive, requiring manual work across spreadsheets with hundreds of thousands of lines just to understand what was being spent, and whether any of the optimization work was moving the number. Aggregating findings across dozens of AWS accounts and multiple Azure subscriptions into a picture leadership could act on was equally difficult.
Billtrust needed one place where all of it was visible, attributed, and actionable. That conversation started with Mavvrik.
Establishing the business case
Like any fiscally responsible organization, Billtrust required a clear, demonstrable return before committing to a new platform. The business case was grounded in what Mavvrik made visible immediately: consolidated cost findings across dozens of accounts that had never been aggregated before. The savings potential across those accounts was substantial. Billtrust committed to saving at least twice what they spent on the platform in the first year and delivered on that commitment.
Beyond the savings, the team used the platform to create a structured, repeatable program for cost management. Findings could be surfaced, assigned, implemented, and validated entirely within the platform, integrated with JIRA, without requiring senior involvement at every step. Cost management became something the team could run at scale, with accountability built in.
The foundation
From day one, Mavvrik gave Billtrust a single pane of glass across AWS and Azure, consolidating spend data that had previously lived in disconnected exports and manual reports. Reserved instance savings, savings plan coverage, contractual discounts, and hybrid benefit calculations across both cloud providers were surfaced automatically and accurately. For the first time, the infrastructure team had a reliable view of what the business was spending and what it was saving, across every account and subscription, in one place.
Cost optimization findings were aggregated across the full account footprint and made actionable through a structured workflow: identify, assign, implement, validate. Jira integration meant findings moved through the right teams without manual coordination. Savings were tracked and confirmed, not estimated. That foundation, cloud cost clarity at scale, is what made everything that followed possible.
Where the business goes to make decisions
Mavvrik has grown well beyond its starting point. Engineering leads use it to understand what specific microservices cost to run. Product teams use it to evaluate cost to serve and unit economics. Executives use it to track cloud and SaaS spend against annual commitments and monitor PPA agreement burndown across both AWS and Azure.
Forecasting is one of the clearest examples of that evolution. Billtrust used to rely on Excel’s pattern-based forecast function, a model with no awareness of what the business was actually planning. Now, when leadership asks whether spend is on track for the year, the answer comes from Mavvrik: machine learning forecasts layered with real operational context, planned optimizations, new services coming online, reserved capacity burndown. The forecast reflects what the business is doing, not just what the trend line suggests.
As Billtrust expanded its SaaS footprint, Snowflake, MongoDB, Datadog, Confluent, that spend came into the same model. Cost signals from across the stack, rationalized into one view, available to the people who need them.
Turning unit economics into confident pricing strategies
One of the most consequential shifts has been in how Billtrust uses cost data to make smarter decisions for its customers. When evaluating different pricing models and deal structures, the ability to understand what it actually costs to deliver a product at a given volume or customer profile changes the quality of the conversation entirely.
Cash Application was the first Billtrust product to have a full cost-to-serve model built in Mavvrik, combining cloud spend with SaaS costs across Datadog, MongoDB, Snowflake, and Confluent. With that foundation in place, the team can identify comparable customer profiles, pull actual unit economics, and approach pricing conversations grounded in real infrastructure cost rather than estimates. That changes the quality of those conversations for everyone involved.
The goal for the year is to bring all products in the Billtrust suite to the same place. As each product is onboarded into Mavvrik, costs become visible at the customer level, so every business decision starts from a complete and accurate picture of costs.
“Cost allocation models let us answer the question that actually matters: what does it cost to deliver each of our products? When you have that data, across cloud, data platforms, and AI, you can have a real conversation about pricing and investment. Mavvrik is what makes that possible without having to build it ourselves.”
Troy HeffernVP Infrastructure & Cloud Engineering, Billtrust
A growing AI footprint, and the governance to match
Billtrust is building AI into both its products and its internal operations, Claude Enterprise for developer tooling, OpenAI-powered production agents, and a growing fleet of agents being brought online across the organization. Each new agent, each new model, each new AI service is another cost signal that needs to be captured, attributed, and connected to the products and teams consuming it.
Mavvrik now ingests AI vendor spend alongside cloud and SaaS, Anthropic, OpenAI, AWS Bedrock, Microsoft Foundry Tools, and code assist tools, in the same cost model that already governs the rest of the stack. Where developers have dedicated API keys, that spend is attributable to specific workflows today. As agentic workloads grow and more agents are brought into the platform, the same unit economics framework being built for cloud, cost by product, cost by customer, will extend to every AI workflow Billtrust runs.
The stakes are only rising. As AI spend scales and more of Billtrust’s product value is delivered through AI-powered workflows, understanding what those workflows cost to run at the customer level is the same business question as understanding the underlying cloud infrastructure. The cost model has to cover everything.
Spend from Anthropic, OpenAI, AWS Bedrock, Microsoft Foundry and code assist tools sits in the same cost model as cloud and SaaS.
Where developers have their own API keys, spend is tied to specific workflows.
Cost by product and by customer for every AI workflow Billtrust runs.
“As our infrastructure investment grows, cloud, SaaS, and now AI, we owe it to our internal teams and our customers to make sure every dollar is being spent as efficiently as possible. Mavvrik gives us the foundation to do that across the entire stack.”
Farai AlleyneSVP Engineering & Infrastructure, Billtrust
Looking ahead
AI means the infrastructure picture is always evolving, with new models, new agents, and new cost sources coming online faster than any static approach can keep up with. The goal is not a finished view, but the most complete and current one possible, so that engineering, product, and finance are always working from the same numbers when they decide where to invest, where to optimize, and what to build next.
About Mavvrik
Mavvrik is an AI and hybrid cloud cost governance and optimization platform. It unifies AI, cloud, and GPU cost and usage data across public cloud, private AI, and on-premises environments, surfaces prioritized insights, and drives action, so enterprises can control spend and measure the value it delivers. SIs and MSPs embed Mavvrik in the AI factories, managed services, and advisory engagements they deliver, putting cost governance in place from day one.