AI data analyst for telecom
Trace the jump in port-outs before the retention review.
Port-outs, churn and availability by region. Bruin flags what moved.
Base health
Week of Sep 14 · all regions
- Port-outs
- 3,420
- +38%
- Churn
- 1.56%
- +0.06 pts
- ARPU
- $38.20
- +$0.10
- Availability
- 99.66%
- −0.29 pts
Port-outs per week, last 12 weeks
Port-outs up 38%. Most of the jump is in the North.
billing.subscribers + porting.requests + noc.incidents · refreshed 06:00
Trusted by forward-thinking teams
Ask it anything
Retention review at 10. The cause is in Slack.
Your numbers
Defined once. Used in every answer.
- Churn
disconnects / avg_subscribers- ARPU
service_revenue / avg_subscribers- Port ratio
port_ins / port_outs- Network availability
cell_minutes_up / cell_minutes_total- Activation time
median(activated_at − ordered_at)- Contact rate
care_contacts / avg_subscribers × 100
From a customer
“If something crashes, I have a root cause analysis done in 15 minutes and can act on it right away. A single investigation can go all the way to the root, rather than stopping at a table someone else owns that we know nothing about.”
Adam MusialVP of Analytics, ProphetX- runs all data and analytics
- 1 person
- idea to production data source
- Month → 1 day
- more pipelines, same headcount
- 3x
Built on the Bruin platform
What runs behind the answer.
Sources
DatabasesWarehousesApps & APIsFiles & storageStreams & webhooksWeb scrapingMove
Data IngestionFrequently asked
Questions from telecom teams.
How do I know Bruin’s churn numbers are right?
Every answer shows the query and the billing and porting tables behind it. Churn, ARPU and port ratio have one definition each, tested on every run.
Which telecom systems does Bruin connect to?
Oracle, PostgreSQL, SQL Server, SAP HANA, Salesforce, Zendesk, ServiceNow, Talkdesk, Twilio, PagerDuty, Datadog, Kafka, Kinesis and your warehouse. Porting files and CDRs load through SFTP or S3.
Can Bruin handle large billing tables and call detail records (CDRs)?
Yes. Bruin runs its queries in your warehouse (BigQuery, Snowflake, Redshift or Databricks), so large tables stay where they are.
Can Bruin link network outages to customer port-outs?
Yes. NOC incidents, care tickets and porting requests join per customer, so you can see which outages come before a rise in port-outs.
Can Bruin flag a port-out spike the morning it starts?
Yes. Set a rule, like port-outs 20% above the four-week average in any region, and Bruin posts it to #retention with the customers whose contracts end soon.
Does Bruin need access to our production billing system?
No. It loads from read replicas, exports or incremental loads, so nothing touches live billing or network systems. Bruin can also run in your own VPC.
We already have a data team. Where does Bruin fit?
Pipelines live in SQL, Python and YAML in your own Git repo, with checks and column-level lineage. Business teams ask in Slack on top of the models your team owns.
Can retention and care teams use Bruin in Microsoft Teams?
Yes. Bruin answers in Slack, Microsoft Teams, Google Chat, WhatsApp, Discord, Telegram, email or the browser. Nobody has to learn a new tool.
How is Bruin different from Tableau or Power BI?
Those show the dashboards someone already built. Bruin answers the next question in Slack, runs the query itself and shows its sources, and can build the dashboard too.
How is Bruin different from asking ChatGPT about our subscriber data?
A chat assistant only sees what you paste in. Bruin answers from your connected billing, porting and care data, runs the query itself and shows its sources.
How much does Bruin cost, and is it per seat?
Bruin Cloud is free to start: $100 in credits and 50 AI tasks, no credit card. After that, compute is billed per second (about $2.50 an hour for SQL, $10 on a standard instance) and AI tasks cost $1 to $3 each, depending on complexity. Committed-use discounts lower the rate as usage grows. No seats.
Is our subscriber data safe with Bruin?
It stays in your warehouse, with access by role. Bruin is SOC 2 Type 2 attested and ISO/IEC 27001:2022 certified, and your data is never used to train AI models.
Catch the next port-out spike the morning it starts.
$100 in credits and 50 AI tasks. No credit card.
A demo walks through your own data.