Freed up 95% of engineering time with self-service data pipelines
Hevo Data
Hevo Data is a no-code ETL platform, helping businesses automatically move data from all their different sources/applications into one central database for analysis.
This case study details how I contributed to Hevo Data's mission by redesigning their core pipeline setup flow, giving users the control and confidence they needed to build reliable data pipelines.
A look at the new, simplified controls giving users full command of their data pipelines.
Impact
Freed up 95% of engineering time previously spent on manual data pipeline maintenance.
Enabled customers to scale their business up to 10x by providing a reliable and robust data infrastructure.
Boosted overall customer satisfaction by 50% by delivering a more transparent and predictable user experience.
Users & their needs
DATA ENGINEERS
Pain: Failing pipelines with little or no control to fix.
Consequence: They become a roadblock for the entire team.
Need: Confidence to dictate how pipelines handle errors and schema changes proactively, eliminating surprises.
DATA ANALYSTS
Pain: Can't trust the freshness of the data.
Consequence: They risk giving bad advice based on faulty information.
Need: Confidence that the data would arrive reliably and on time for their critical reports and dashboards.
NON-TECHNICAL USERS
Pain: Making decisions using unreliable data.
Consequence: They might miss opportunities due to faulty information.
Need: Confidence to create/manage their data pipelines without assistance or the fear of breaking something.
Team & my role
As the senior product designer, I worked closely with the design lead, engineers, product manager, data engineer and data analyst.
Led the research and found the true problem: it wasn't just broken tech, but broken trust in the data.
Designed the final solution, turning a complex system into a simple, visual tool that put users back in control.
Tested the first solution, focusing on how much it improved a user's feeling of confidence and control.
Constraints & roadblocks
We had to fix a complex old system without rebuilding it.
The product's poor quality was blocking us from selling to enterprise companies.
The design had to be simple for business users yet powerful for engineers.
Key decisions
Deliver value in small, safe steps instead of one big, risky launch.
Prioritize shipping features that would build the most trust.
Give users more direct control, not an oversimplified system that hides the details.
Design solution
We rebuilt user confidence through a series of deliberate, focused phases.
PHASE 1
Diagnosing the confidence gaps
Our customer support channels were flooded. Users complained about failing pipelines and broken dashboards.
When I dug deeper, the competitive research revealed that platforms like Fivetran and Airbyte gave users granular control and proactive feedback, features we lacked.
I audited our flow and partnered with engineers to understand the "why" behind every confusing error and unexpected failure.
This video diagnoses the old pipeline creation flow, highlighting the key friction points.
PHASE 2
Designing for user confidence and trust
Armed with this diagnosis, I designed the ideal solution: a guided, multi-step flow that gave users full transparency and control before they created a pipeline.
My new design introduced three critical confidence-building features:
Update 1: Introduced sync frequency
We switched to a single 'Sync Frequency' for the pipeline from two different frequencies for data ingestion at source and load at destination
This led to simplified experience, trading off cost/load for lower latency and no data loss.
This video shows how we replaced two separate, confusing pipeline sync settings with one simple "Sync Frequency", giving users clear control and confidence in their data delivery schedule.
Update 2: Intelligent object selection
We introduced a guided step that pre-validates data, asks for required keys upfront, and lets users select specific fields.
This prevents common setup failures and gives users confidence that they are moving the right data correctly.
This video demos our new intelligent object selection step. It guides users to a successful setup by providing key context, disabling inaccessible objects, and requiring primary keys upfront to prevent common pipeline failures.
Update 3: Proactive error control
We gave users granular control over how their pipeline should handle errors or schema changes.
This shifted them from reacting to failures to proactively defining their pipeline's resilience.
This video demos the new pipeline settings, where we gave users granular control over error handling, load modes, and schema evolution to build more resilient and predictable pipelines.
PHASE 3
Validating and communicating our commitment
This clickthrough demo shows our redesigned pipeline creation flow. The object and pipeline configuration steps prevent common errors upfront and give users full control, building confidence that their pipeline will succeed from the very start.
Outcome
Business: We successfully reduced customer churn and unlocked the ability to sell to enterprise-level clients, a key company goal.
Users: Data-related support tickets were reduced by 60%, signaling a significant increase in user confidence and self-sufficiency.
Learnings & growth
My biggest learning was the need to go deep into the engineering domain; it was the only way to understand the core system limitations and design an effective solution.
The project's success ultimately came down to creating a shared understanding between design, engineering, and business, which was more critical than the design itself.
