Case Study · DPDzero · Fintech
DPDzero

How DPDzero used Scout to spend less and get more

DPDzero replaced a patchwork of CloudWatch, Sentry, and Performance Insights with one OpenTelemetry-native standard on base14, turning system reliability from one engineer's burden into team-wide infrastructure. Key Outcomes →

Industry
Fintech
HQ
Bengaluru, India
Employees
~249
01

About DPDzero

DPDzero is an AI-powered, full-stack debt collection and recovery infrastructure platform built for financial institutions and lenders in India. They have raised $10.8 million in funding across several rounds. Their clients include Banks, NBFCs, and fintechs, such as IndusInd, Aditya Birla Capital, and L&T Finance.

02

The challenge

DPDzero's engineering team was running observability through a stack of CloudWatch, AWS-native tooling, Sentry, and Performance Insights.

Publishing metrics was difficult, as was tracing.

CloudWatch

CloudWatch's custom metrics require explicit instrumentation for anything beyond default AWS service metrics, and its dimension-based pricing discourages the high-cardinality, per-request data that serious debugging needs.

Tracing & Sentry

Tracing was split across two disconnected systems: AWS-native tracing propagates its own trace IDs across service boundaries, while Sentry generates a separate set of transaction and error IDs for application-level issues.

Performance Insights

Performance Insights, meanwhile, operates at the database query level, with no concept of an application trace ID to tie a slow query back to the request that triggered it.

It all traced back to the same gap: these tools were never built to share a common data model or trace context. Each captured its own slice, metrics somewhere, errors somewhere else, with no thread connecting them. Reconstructing what happened for one request meant dashboard hopping manually, and checking on system health became a time sink.

Part of the time-consuming nature was tied to the fact that system reliability rested on the shoulders of one engineer. For one person to try and work across a distributed, inefficient stack was incredibly difficult. Naturally, the CTO wanted system reliability to be every engineer's responsibility. This meant cost was increasingly part of the picture too.

Incumbent tools' per seat pricing model meant that having every engineer responsible for system reliability would quickly balloon costs. A high cost is one thing, a high cost that isn't justifiable against the outcomes is another. Even so, the real frustration was operational: not enough database visibility, and a setup that made observability a siloed chore.

There was a deeper pattern underneath this, too. Features shipped, teams moved on, and if something degraded post-launch, DPDzero typically found out from a support ticket or from someone happening to eyeball an existing dashboard, rather than from any built-in way of knowing.

03

Why they chose base14

Open Standard

base14's OpenTelemetry-native foundation was a big draw, no black box, just an open standard DPDzero could build unified observability on.

Usage-Based Pricing

Cost came up, naturally. Our usage-based pricing meant that team-wide system reliability responsibility became feasible.

The Walkthrough

What sealed it was the insights we provided in the walkthrough, a technical breakdown of DPDzero's setup and how Scout would fit into it, paired with a trial that delivered on its promise.

04

Key outcomes

Unified observability, better reliability

All on one platform0% samplingFull visibilityImproved reliability

Call abandonment rate more than halved, going from above 50% to less than 25%. That reliability now shows up in day-to-day engineering habits, too: engineers check the dashboard the morning after a release as a matter of course, rather than waiting to be told something broke, which has caught regressions within the first hour of release that would previously have gone unnoticed for days.

base14 transformed our observability! We gained a deep and unified visibility across our stack - API, RDS, tracing, Celery, and more - which empowered us to quickly trace issues and proactively optimise performance.

Ranjith B. R.

Founder & CTO, DPDzero

Fully Customized Instrumentation

DPDzero wanted to expand their monitoring both vendor and client-side. For their vendors, to monitor performance and evaluate the cost-benefit of each vendor. For their clients, DPDzero wanted to monitor their own performance towards the KPIs they used to benchmark their work in collections. Working with base14, they were able to add customized instrumentation that provided the team visibility on both of these - helping them view their cross-functional systems more holistically, empowering themselves to make more informed decisions.

DPDzero even brought in different business units into their observability system - their operations team is now able to monitor their outbound service more efficiently, and operations leaders can see what is working well via metrics.

MCP Integration

Scout MCP has been a major reason DPDzero continues to value our partnership. The MCP has helped engineers easily analyse what's happening in the system, helped narrow issues to a particular point, and respond quicker - improving system reliability and nipping issues in the bud. Dashboard creation itself has become part of how DPDzero ships: a feature doesn't go live until its dashboard, covering the metrics, traces, and business counters that show whether it's doing what it was built to do, exists and is on the release checklist. To keep that from becoming manual overhead, DPDzero set up a GitHub workflow that uses LLMs to help generate these dashboards on an as-needed basis for every new feature they deploy.

Using MCP to create dashboards, and for debugging and preventing future issues. Letting AI do the work has been going really well. It's definitely unique

Yogeshwar Trehan

Senior Software Engineer, DPDzero

05

The service model

Setting up or transferring to a new observability system can be a daunting task, which is why we at base14 provide hands-on support not just at onboarding, but continually after that. base14 engineers worked closely with the team at DPDzero to set up custom instrumentation, and we continue to have fortnightly meetings.

Ongoing collaboration is built into how base14 works with our customers, not a one-time step at handoff.

Technical

On the technical side, custom instrumentation was worked out jointly, so tracing and metrics could share the same context from the start, tailored specifically to how DPDzero's systems run, and further business needs.

Efficiency

On the efficiency side, having base14 engineers involved past onboarding meant instrumentation could keep evolving alongside DPDzero's stack, rather than freezing in place after initial setup.

Cultural

On the cultural side, the fortnightly meetings keep both teams in the loop, always, so priorities and changes on either end surface early.

This is the shape the partnership takes by design: base14's service model treats continued engagement as part of the product itself, not a support function layered on top of it.

06

Summary

DPDzero came in with a cost problem downstream of a tooling and responsibility problem. What kept them was the support, the product itself, a workflow built around OpenTelemetry standards, an MCP integration that gave them something no one else did, and a level of cross-functional customization that turned observability from a chore into infrastructure they trust.

See how Scout can help you halve your MTTR while keeping costs under control