Consumer Electronics
Product intelligence from a connected appliance fleet
A leading global consumer appliances company
IoT telemetry, competitive pricing intelligence and anomaly detection in production
- Industry
- Consumer electronics / connected appliances
- Scale
- 70,000+ telemetry records per minute
- Engagement
- IoT analytics + pricing intelligence + anomaly detection
- Architecture shown
- Google Cloud reference design
The challenge
- A large connected fleet produced telemetry no non-technical team could actually query
- Competitive pricing had to be tracked manually, and did not exist at all for some markets
- Device anomalies were found by manual investigation, usually after customers noticed
- Product teams needed SQL to answer basic questions about fleet behaviour
What it had to do
- Ingest fleet telemetry continuously without dropping records
- Let product and analyst teams ask questions without SQL
- Detect and classify anomalies before customers feel them
- Track competitor pricing across retailers and SKUs daily
What we built
We built the streaming path for fleet telemetry, then put a natural-language layer over it so product teams could ask questions directly. Anomaly models classify issues by severity and route alerts to the right team, and a separate pipeline tracks competitor pricing across retailers and SKUs with multiple refreshes a day.
Streaming telemetry
Continuous ingest from the device fleet at 70,000+ records per minute.
Plain-English queries
Analysts ask questions about fleet performance without writing SQL.
Anomaly detection
Models classify anomalies by severity and route alerts before customers notice.
Pricing intelligence
160 retailers and 1,150+ SKUs tracked live, with several refreshes a day.
Reference architecture
Fleet
- Device telemetry
- Retailer feeds
Stream
- Pub/Sub
- Dataflow
Store
- Bigtable
- BigQuery
Act
- Vertex AI
- Gemini NL query
- Looker
Results
- Competitive pricing intelligence live in production across 160 retailers and 1,150+ SKUs in three countries
- Fleet telemetry made accessible to non-technical product teams through a natural-language interface for the first time
- Real-time anomaly detection against the live fleet, classified by severity and routed to the right team
- Predictive maintenance schedules generated from actual usage patterns rather than fixed intervals
- Automated daily and weekly reporting replaced manual investigation
For product teams
Fleet questions get answered directly, without a data request.
For support
Issues surface before customers report them.
For commercial
Pricing position is visible daily, including in markets that had no coverage.
- Pub/Sub
- Dataflow
- Bigtable
- BigQuery
- Vertex AI
- Gemini
- Looker
Facing something similar?
Tell us where you are now. A senior engineer replies, usually within one business day.
A senior engineer reads every message and replies within one business day.