CASE STUDY
Delivering fresh insights, Fast:
How a grocery tech leader scaled insights with PromptQL
Customer
Leading online grocery delivery platform in North America
Challenge
The company had a sophisticated data stack and a large, highly skilled data science and analytics team that serviced requests from business users across the organization.
For a business that prides itself on data-driven operations, keeping up with growing demand meant either continuously expanding the analytics team or requiring SQL proficiency across a wide range of business roles. Neither path was scalable or sustainable.
This led to a key question: could conversational analytics with AI break this bottleneck in a sustainable way?
Solution
After a rigorous evaluation process, the company selected PromptQL as its AI analytics platform.
The pilot began with data and analytics experts, who used PromptQL to accelerate data discovery, exploration, and querying–eliminating the need to manually craft queries. Following the pilot’s success, adoption expanded to business and operations users, who could now generate insights directly from data instead of waiting on the analytics team.
The result: improved operational efficiency, faster access to insights, and better data-driven decision-making across the business.
PromptQL has transformed how quickly and easily teams across the company get data-driven answers – without needing deep SQL or tooling knowledge.
Data Connected
Over 800 tables in the enterprise data warehouse, covering key business domains such as operations and delivery performance, product and pricing data, workforce effectiveness, customer engagement, and spend analytics.
Why PromptQL
High accuracy
For an org-wide rollout, trust in results was essential. PromptQL delivered low hallucination rates, deterministic responses, confidence indicators, and explainability, ensuring users could rely on every answer.
Domain-aware AI
With hundreds of custom KPIs and metrics, PromptQL quickly adapted to the company’s unique language and workflows, making it useful across a wide range of business tasks.
Continual learning at scale
When PromptQL missed a question, improving context was fast and repeatable. Its ability to learn from users (with governance) enabled ongoing improvement and large-scale adoption.
Time savings for data teams
PromptQL answered complex questions in minutes that previously took data science and analytics teams hours, freeing experts for higher-value work.
Example User Questions
How do retention rates differ when shoppers activate during low-demand vs. high-demand periods? Use data from the last 12 months, U.S. only.
Plot weekly GMV of orders for the last 8 weeks (use bar chart)
What % of batches are completed by STAR shoppers? Break this down by TTD ranges (<10min, 10-20min, 20-30min, >30min) and show ACR for each segment.
How many pickup orders were fulfilled at P1 warehouses last week?