Capabilities
Data and analytics: the path behind every number
Every figure on a report arrived from somewhere, was shaped into something, was checked, and was put to use. These twelve capabilities are that path. When a number is wrong, it went wrong at one of these steps, and a well-built path can tell you which.
Two kinds of data, one path
The data a business runs on and the data it's measured by are the same facts in different shapes. Most reporting problems start by treating one as the other.
Operational dataAnalytical data
ForRunning the work: today's jobs, this customer's invoiceMeasuring the work: jobs by month, revenue by source
ShapeOne record at a time, always currentMany records at once, with history
Breaks whenA record is wrong or missingDefinitions differ, or a load is incomplete
The path turns the first into the second deliberately: it brings operational records in, shapes them into a model built for measurement, checks them, and only then lets reports and dashboards read them.
The twelve capabilities
Grouped by where they sit on the path. Checking isn't a step at the end; it happens at every stage.
Arrive
Getting the data in, with a record of what came.
- Data IngestionIn-depth pageBrings each source in on schedule and records every load.
- ETL / ELT PipelinesIn-depth pageRuns extraction, loading and transformation as one repeatable job.
- Analytics InstrumentationIn-depth pageCaptures the events and properties your own sites and apps should record.
- Data EnrichmentIn-depth pageAdds attributes from other sources to the records you already have.
Shape
Turning records into something you can measure.
- Data TransformationIn-depth pageWritten rules that turn every source into one shape.
- Data ModelingIn-depth pageDecides what one row means and how every metric is calculated.
- Database DesignIn-depth pageSchemas, constraints and indexes for the systems that store it.
- Data WarehousingIn-depth pageOne home for analytical data and its history, when there's enough to need one.
Check
At every stage.
Use
Putting the numbers in front of people.
Start with the symptom
Data Ingestion
“Someone downloads the numbers every Monday”
Replace the routine with loads that run on their own and leave a receipt.
Data Modeling
“Every report says something different”
Agree what's being counted, once, and every report inherits it.
Data Quality & Validation
“A wrong number reached a meeting”
Put checks where data crosses boundaries, each with a consequence.
Related pages
Where they come together
- Automated ReportingSix of these capabilities, end to end, behind one recurring report.
- Executive DashboardModeling, BI and dashboarding for the few numbers leadership watches.
- Data UnificationIngestion, transformation and quality aimed at one record per customer.
- Integrated PipelineThe arrive and shape stages as a product of their own.