Connectors
airbyte
- feeds
- Normalise
All case studies
SaaSData PipelinesDashboards4 months
Plug-and-Play Commerce Analytics
Signing up a new customer took three weeks of engineering time. It now takes two days and no engineer.
Tenant onboarding
2 days
Connector + template, self-serve
Was
3 weeks
Bespoke integration per tenant
In plain terms
What this actually was.
No jargon in this section. The technical write-up is further down.
What was going wrong
Every new customer needed custom integration work before they could see a single dashboard, so the sales team queued behind engineering. Meanwhile the revenue figures on those dashboards quietly drifted away from what finance was reporting.
What we built
We built one standard connector and a ready-made template a customer can set up themselves, and tied every revenue figure to the definition finance actually uses.
What changed
Onboarding dropped from three weeks to two days and no longer needs an engineer at all, so sales stopped waiting. Support questions of the "why do these two numbers differ" kind fell sharply.
Architecture
How it fits together.
Scroll through the stages. Anything marked as added is a component that did not exist before this project.
Normalise
canonical
- receives from
- Connectors
- feeds
- Quality gatetest
Quality gate
contracts
- receives from
- Normalisetest
- feeds
- BigQuery
BigQuery
multi-tenant
- receives from
- Quality gate
- feeds
- Metrics layer
Metrics layer
dbt
- receives from
- BigQuery
- feeds
- Dashboards
Dashboards
templated
- receives from
- Metrics layer
Connectors
airbyte
- feeds
- Normalise
Normalise
canonical
- receives from
- Connectors
- feeds
- Quality gatetest
Quality gate
contracts
We added this- receives from
- Normalisetest
- feeds
- BigQuery
BigQuery
multi-tenant
- receives from
- Quality gate
- feeds
- Metrics layer
Metrics layer
dbt
We added this- receives from
- BigQuery
- feeds
- Dashboards
Dashboards
templated
- receives from
- Metrics layer
The write-up
How it was actually built.
Dataset, approach, measured results and the stack. Written for whoever has to review it.
SaaSData PipelinesDashboards4 months
Plug-and-Play Commerce Analytics
Multi-tenant ingestion and revenue analytics that speed SMB onboarding with stable KPI definitions.
Problem
Every new tenant needed bespoke integration work, so onboarding took weeks of engineering time and dashboards silently drifted from the finance definition of revenue.
Dataset
- rows
- 58M orders
- features
- 112
- sources
- Shopify, Stripe, Ads, CSV
- window
- 18 months
Approach
- 01Built reusable connectors that normalise to a canonical commerce schema
- 02Added automated data quality checks at the ingestion boundary
- 03Modelled revenue logic once in a governed metrics layer
- 04Templated dashboards with per-tenant admin overrides
Metrics
- 2 days
- onboarding time from 3 weeks
- 94%
- schema coverage auto-mapped
- 312
- QA failures caught pre-load
- 100%
- metric parity vs finance
Business impact
- Self-serve onboarding removed engineering from the sales cycle
- One source of truth for subscription and commerce KPIs
- Support load on reporting questions down sharply
Tech stack
- TypeScript
- Next.js
- PostgreSQL
- BigQuery
- Temporal
- dbt
- Airbyte
- Metabase
Next
Need a system like this?
Answer six questions and we will tell you whether this shape fits your problem — including when it does not.