- Transferz, a fast-growing travel tech platform, eliminated its data team bottleneck by deploying AI/BI Genie — achieving 89,5% adoption across business users achieving 60% operational improvements.
- Their four-phase rollout — Curation, Calibration, Enablement, and Evolution — offers a proven, repeatable playbook for any organization looking to democratize data access.
- By treating Genie as a cultural initiative rather than a technology project, Transferz transformed even hesitant non-technical users into confident, self-service data consumers.
"Genie has been a game-changer for our data maturity. It is 10 times faster and 100 times more accurate for daily ad-hoc tasks, essentially acting as the digital analyst we needed to bridge the gap between business questions and technical data."
— Lesley Silbernberg, Head of Data, Transferz
How do you empower a high-performing data team to focus on what truly matters?
That was the challenge at Transferz, a travel technology platform connecting global travel brands with local ground transportation across 150+ countries. Ranked #5 in Deloitte's Technology Fast 50 for its innovative, tech-driven approach, Transferz was scaling fast but its data operations were victims of their own success. The challenge was painfully familiar to data leaders everywhere. Business users needed answers. But non-technical employees hesitated to ask questions at all, either daunted by the complexity of doing it themselves or worried about burdening an already stretched data team. The data team had the skills to deliver them. But with analysts spending up to 25% of their time fielding simple, repetitive queries and the rest of the time on operating the platform, strategic work was constantly deprioritized. All of this led to data-driven decisions being left on the table. Transferz needed a way to put data directly into the hands of every employee without requiring them to learn SQL, wait in a queue, or feel like they were bothering someone.
They found it in AI/BI Genie.
What followed was not just a tool deployment. It was a cultural transformation. And the way they did it offers a blueprint that any organization can follow.
The playbook: four phases from foundation to production
Transferz realized early on that you cannot just flip a switch and hope for the best. While the early iterations of Genie performed well on simple queries, the lack of accuracy for more complex business questions damped adoption amongst business users. Convinced of Genie’s capabilities, the data team then followed a deliberate, phased approach to drive adoption and cultivate a data-driven culture at Transferz. Here is how they did it:
Phase 1: Curation — Building a Precise Foundation
Instead of pointing Genie at every available table, Transferz focused on high-impact data to ensure reliability and to guide customer behavior.
- Aggressive Scoping: The team began by identifying high-impact tables within their Unity Catalog semantic layer, selecting only core data assets that users queried most frequently surrounding a specific business topic.
- Metadata Enrichment: Analysts wrote detailed descriptions and comments for every column. They utilized AI-generated suggestions to accelerate documentation for tables with up to 100 columns, followed by a peer-review "cross-check" process to ensure clarity and correctness.
- Strategic Instructions: This critical layer bridges business jargon and technical schemas. Transferz encoded logic rules—such as directing Genie to return airport names (IATA codes) instead of internal IDs—and defined complex joins to reduce the number of individual instructions needed.
Key Takeaway: Context is everything. Genie’s performance is directly proportional to the quality of your instructions and metadata.
Phase 2: Calibration — Validating with the Right Users
The team utilized a structured, two-stage testing cycle to refine the system before a company-wide rollout.
- Internal Stress Test: A two-week validation period where the data team used Genie intensively to verify responses against known answers and identify gaps in metadata.
- Power User Pilot: A one-month pilot involving data-savvy "ambassadors" from various departments. These users were briefed that the tool was in a testing phase, which was crucial for managing expectations and preventing early frustration. With a key point that this is not like generic AI tooling that gives answers to everything.
- Active Feedback Loops: Users were instructed to flag incorrect responses using the built-in feedback mechanism. Analysts reviewed these daily during the first month, and once corrected, users received a "red dot" notification to rerun the refined query.
Key Takeaway: Start with users who understand they are testing, not just using. Rapidly closing the feedback loop builds the trust necessary for long-term adoption.
Phase 3: Enablement — Creating Organic Pull
Rather than mandating adoption through training sessions, Transferz positioned IRIS (internal branding of Genie) as a high-value "data translator" to generate organic demand.
- Intentional Branding: Naming the Genie implementation as IRIS and framing it as a professional assistant rather than a "magic oracle" like ChatGPT helped manage user expectations regarding its specific capabilities.
- Prompt Guidance: The team gathered the most frequently asked historical data questions and published them as Sample Questions. This gave users a clear starting point and reduced "blank prompt" anxiety.
- The "FOMO" Strategy: Power users shared screenshots of their successful insights in a dedicated Slack channel, before general users got access. Seeing real-world value created a "fear of missing out" (FOMO) effect, leading colleagues to request access before the general release.
Key Takeaway: Do not launch broadly until you have organic ambassadors. When users pull for a tool because they've seen it work for their peers, adoption becomes self-sustaining.
Phase 4: Evolution — Evolving into Proactive Intelligence
Transferz treats IRIS as a living product that requires continuous refinement as the business evolves.
- Systemic Audits: The team performs bi-weekly audits of query logs to identify missing context, new business terminology, or areas where users are struggling.
- Specialized Spaces: As adoption grew, Transferz began building department-specific Genie Spaces (e.g., for Product or Operations) to provide deeper context awareness without cluttering a single environment.
- Proactive Alerts: IRIS now runs on pre-set prompts to deliver daily Slack notifications, flagging anomalies such as sudden 50% revenue drops. Users can respond directly with follow-up questions, turning analytics into an always-on conversational assistant.
Key Takeaway: Plan for ongoing investment. The organizations that get the most value from Genie are those that treat metadata and instructions as living documentation rather than a one-time setup task.
From efficiency to intelligence: The Transferz success story
Transferz’s transition to a self-service data culture powered by IRIS has delivered a 89,5% adoption rate among business users including the CFO, with usage growing 222% month-over-month following recent Genie updates. This shift has eliminated the "request-response" bottleneck, allowing the data team to reclaim the equivalent of multiple full-time analysts in monthly capacity.
The business impact is significant: monthly active users report a 60% productivity increase, saving 10–15% in effort across 12 major projects. Beyond standard queries, IRIS now provides proactive intelligence. For instance, it now automatically flags anomalies—such as sudden 50% revenue drops— through integration with Slack, enabling immediate intervention and directly impacting revenue generation. By removing the technical and social barriers to data access, Transferz has empowered its entire workforce to make rapid, high-precision decisions.
The cultural shift: the real ROI
The real game changer for Transferz was dismantling the psychological and social barriers to data. By providing a neutral, conversational interface, the company eliminated the "social friction" where employees felt hesitant to bother analysts with routine questions. This shift transformed data from an intimidating, gatekept resource into an everyday utility, leading users to check ten metrics where they previously checked two. Today, non-technical teams use IRIS to independently validate hypotheses and uncover hidden trends in real-time. This democratization has fundamentally raised the company’s data maturity, evolving the culture from a reactive "request-response" model to one of proactive, self-sufficient discovery.
Get started
Ready to bring the power of conversational analytics to your organization? TransferZ started with a single curated Genie Space and scaled to company-wide adoption in a matter of months.
Related Databricks Blogs
