AI-powered, self-service data product
Objective
Build the data foundation powering Blink IQ 2.0, a self-service, AI-driven analytics product for Blink’s enterprise customers.
Obstacle
Blink’s customer analytics product was running on an undocumented, untestable legacy stack that the team had outgrown and could not change with confidence.
Outcome
Designing every model specifically to power ThoughtSpot’s AI query engine and unlock self-service analytics for Blink’s enterprise customers.
Background
Blink is a frontline employee communications platform, serving enterprise customers across sectors including logistics, hospitality, and retail. Blink IQ is Blink’s customer-facing analytics product; giving organisations visibility into employee engagement, platform adoption, and communications reach.
With Blink IQ Power BI-based approach producing static dashboards, poor NPS scores, and a growing backlog of custom reporting requests, Blink committed to a full rebuild: migrating to ThoughtSpot for embedded, self-service analytics with AI-powered natural language querying. 173tech were engaged to design and build the underlying data models and semantic layer to power the new platform.
Challenges
Held Together With Stored Procedures: Blink’s existing analytics ran on a proprietary transformation tool Kleene, that stored SQL logic as text inside database tables, executed via cron-scheduled stored procedures. With no tests, no documentation, and fragile dependencies between transformations, the team had effectively outgrown their own infrastructure but were unable to change it with confidence.
No Shared Definition: Metrics were inconsistently defined across dashboards, and users could not trust what they were seeing. Before any model could be built for 2.0, Blink needed a single agreed definition of every metric, dimension, and business concept across the entire platform, something that had never existed before.
Solution
Data Dictionary: 173tech produced a comprehensive data dictionary covering every metric, dimension, and filter across the full Blink IQ 2.0 dashboard suite, built collaboratively with Blink’s product and analytics stakeholders. This was not just documentation, it became the contract between business intent and technical implementation, ensuring that what was built matched what customers would actually trust.
Semantic Layer: The entire modelling effort was shaped around one end goal: powering ThoughtSpot’s natural language query engine (“Spotter”), so that Blink’s customers could ask questions in plain English and get reliable answers. 173tech structured naming conventions, relationships, and metric definitions towards this, meaning the quality of the data models directly determined the quality of the AI experience that Blink’s customers would receive.
Impact
From Static Dashboards To Self-Service Intelligence: Blink IQ 2.0 gives enterprise customers something they never had, the ability to answer their own questions. Rather than waiting on the Blink team to build custom reports, users can now build their own dashboards, slice data by department, location, and manager, and query their workforce data in natural language. The shift from reactive reporting to self-service analytics removes a significant ongoing burden from Blink’s internal team while materially improving the product for customers.
A Scalable Foundation: By migrating from Kleene and onto a properly structured dbt codebase, Blink’s data team can now iterate with confidence. Models are tested, documented, and version-controlled, meaning new dashboard areas can be added without the risk of breaking what already exists.
Product-Led growth opportunity Blink IQ 2.0 is a core part of Blink’s commercial offering, with analytics tiering built into their customer pricing. A stronger, more reliable analytics product directly strengthens Blink’s ability to retain customers, upsell to higher tiers, and win new business. The data foundation 173tech built is the engine behind that commercial opportunity.
Creating Value For Blink...
7 data domains modelled,
Full data model suite designed, built, tested, and documented in just 8 weeks
And a data product built for the future.
Get In Touch
Our friendly team are always on hand to answer questions, troubleshoot problems and point you in the right direction.
