WorkRamp
Case Study — SaaS & Technology

WorkRamp: Building Customer-Facing Analytics Customers Actually Trust

How Astrodata helped relaunch an embedded analytics product in under three months — with zero downtime and a foundation built for AI.

The situation

WorkRamp powers employee and customer learning for high-growth SaaS companies — Airtable, Notion, Sprout Social, and others. For years, they'd built their own customer-facing reporting from scratch. The homegrown product worked, but every quarter it demanded more: new reports, new dashboards, more ad-hoc requests routed through engineering and customer success. What had started as a competitive advantage was becoming a tax on the team.

Arsh Mand, WorkRamp's CTO and co-founder, knew the answer wasn't to build more. It was to find a partner that could give their customers the flexibility and power they were asking for — without compromising the clean, intuitive experience WorkRamp had spent years perfecting.

After evaluating the embedded analytics landscape, WorkRamp chose Omni. Then they brought in Astrodata to make sure the foundation was right.

Why this engagement mattered

Embedded analytics projects fail in one of two ways. Either the semantic layer gets rushed — metrics are ambiguous, joins are fragile, and customers lose trust in what they're seeing. Or the implementation moves so cautiously that the rollout stretches into a year-long project and customer momentum dies.

WorkRamp needed neither. They needed to move fast, launch to thousands of customers with zero disruption, and build a data model clean enough to power both self-service exploration and AI-driven insights on day one.

That's the exact intersection of work Astrodata is built for.

Our approach

We partnered with Arsh and his team across two workstreams that ran in parallel: semantic model architecture and phased rollout planning.

Semantic model architecture

The semantic layer is the most consequential decision in any embedded analytics product. It determines whether every customer sees the same numbers for the same question. It determines whether AI responses are grounded in real business logic or just statistical guesses. It determines whether the team building on top of it can move fast or gets stuck relitigating definitions.

We worked alongside WorkRamp to structure their Omni semantic layer from the ground up:

  • Topics designed around customer mental models, not internal data organization. Customers search for what they care about — comprehension, completion, performance — not database table names.
  • Every measure documented, tested, and governed so that a metric calculated in a dashboard produces identical results in an AI-generated query or a scheduled delivery.
  • Joins and relationships validated against real-world customer use cases to prevent the subtle data-quality issues that erode trust over time.

Phased rollout planning

We advised on a rollout strategy that started with a small group of trusted early-adopter customers, validated the experience, then expanded to full adoption within a month. Zero downtime. No customer disruption.

Figure 01 — Three-Month Engagement Timeline
  1. Weeks 1–2
  2. Weeks 3–8
  3. Weeks 9–10
  4. Weeks 11–12Launch
  1. Discovery & semantic model architecture planning
  2. Semantic layer build, governance setup, Topic modeling
  3. SDLC workflows, testing, internal validation
  4. Phased rollout to trusted customers → full customer base

The outcome

Three months from kickoff to relaunch. A customer-facing data product that WorkRamp had spent years iterating on — rebuilt on a modern platform, with a cleaner experience, without a single customer seeing disruption.

Customer feedback exceeded expectations:

"I wish I had this when I signed on, it's like having a reporting assistant."
"My statistical brain is firing on all cylinders."

Measurable engineering impact: WorkRamp recovered approximately 10% of engineering time previously spent maintaining analytics features — roughly two engineers per quarter redirected to core product work.

Figure 02 — Engineering Time Reallocation
  • Before Astrodata engagement
    ~15% analytics features & maintenanceCore product
  • After Astrodata engagement
    ~5% analytics features & maintenanceCore product
≈10% of engineering time recovered = ~2 engineers per quarter redirected to core product.
  • Analytics maintenance
  • Core product

AI-ready foundation: Because the semantic layer was designed with governance and clarity from the start, WorkRamp's customers can now use Omni's conversational AI to ask questions in natural language and trust the answers. No retrofitting required.

Why Astrodata

This project succeeded because the work was done with discipline, not just speed. Three Astrodata values shaped every decision.

Figure 03 — Three Values in Action
  • Curiosity

    Learned customer usage patterns before proposing architecture

  • Excellence

    Every measure, join, and Topic documented and tested

  • Sustainability

    Governance patterns the team extends without us

"Astrodata was a great partner. They helped us think through the right way to structure our data in Omni. Their support helped us accelerate our launch and set us up for long-term success."
— Arsh Mand, CTO & Co-founder, WorkRamp

Industries, technologies, and capabilities

Industry
SaaS & Technology
Client size
High-growth SaaS serving thousands of enterprise customers
Technologies
Omni embedded analytics, semantic modeling, AI-ready data architecture
Astrodata capabilities
Data Product Strategy, Advanced Analytics & AI, Experience Design for Data Products

Ready to build analytics your customers trust?

If you're rebuilding a customer-facing data product, launching embedded analytics for the first time, or designing a semantic layer that needs to support AI from day one — we'd like to talk.

Request a consultation