Case Study — MyArtBroker

From multiple reporting platforms to a single source of truth.

We rebuilt MyArtBroker’s data stack from the ground up — consolidating multiple systems into BigQuery, giving finance and commercial one report they can trust, and putting engineering guardrails around the numbers the business runs on.

ClientMyArtBroker
SectorOnline art marketplace
EngagementEmbedded data partner
FocusFoundation · Analytics · CRM
BigQueryDataformLooker StudioAttioZepicCI/CD
Gets under the skin of a business incredibly quickly, understands where the frustrations lie, and finds practical solutions. If you're looking for someone who can bridge technical complexity with strong commercial understanding, communicate brilliantly with multiple stakeholders and translate data into visually readable insight, I couldn't recommend more highly.
Charlotte Stewart
Charlotte Stewart
Managing Director, MyArtBroker
At a glance
ReportingOne source of truth

Hand-kept spreadsheets replaced by BigQuery + Looker Studio — finance and commercial finally read from the same numbers.

AttributionContent → conversion

Clear visibility into which content draws customers in and actually converts them — discrepancies fixed at the source.

CRMStack migrated

Moved from HubSpot to Attio + Zepic with no disruption to live reporting or priority projects.

FoundationProduction-grade

dbt → Dataform transforms, shared definitions, and CI/CD so business-critical dashboards never break in production.

01 / Single source of truth

One clean report for finance and commercial.

Our first priority was to replace the collection of Excel spreadsheets being maintained by hand, pulling data from disparate systems such as Airtable. We consolidated these sources into BigQuery and built a suite of Looker Studio dashboards on top — giving both finance and commercial one clean, reliable report that surfaced the figures most important to each.

Before → After · reporting architecture
Before
ExcelAirtableManual exportsAd-hoc CSVs

Reconciled by hand · no shared definitions

After
WarehouseBigQuery
Looker StudioFinance
Looker StudioCommercial

Left: figures stitched together by hand across disconnected tools, reconciled manually and prone to drift. Right: sources land in BigQuery once, then fan out to purpose-built dashboards for each team.

02 / Customer attribution

Understanding where the best customers come from.

With reporting centralised, we turned to a defining question: where are the best customers actually coming from? Working in tandem with the engineering team, we identified and resolved data discrepancies so information flowed through cleanly and was reported accurately — giving MyArtBroker clear visibility into which pieces of content were drawing users to the site and, crucially, converting them.

Attribution pipeline
Sources
Web / content
Marketing
Airtable
CRM
BigQueryCleaned & reconciled
ModelContent → conversion
OutcomeWhat actually converts

Raw signals from every channel are cleaned and reconciled once, then modelled into a consistent view of content → visit → conversion — so the team can see what genuinely resonates.

03 / CRM migration

Re-platforming the CRM without missing a beat.

In parallel, we led a significant migration from HubSpot to Attio and Zepic — moving the business from one CRM ecosystem to another — while continuing to deliver on priority data projects with no disruption to reporting.

CRM re-platform
LegacyHubSpot
migrate
CRMAttio
LifecycleZepic
Reporting stayed live throughout — zero downtime for the business.

Data and workflows migrated from a single legacy platform to a modern pairing — Attio for CRM, Zepic for lifecycle — while day-to-day reporting kept running underneath.

04 / Data foundation

A foundation that protects production.

To strengthen the underlying data we first implemented dbt to improve transformations and bring clarity to the business’s ontologies, semantic layers, and shared definitions. As the work progressed, we recognised that Dataform — native to BigQuery — was the better long-term fit, and migrated accordingly. Alongside this, we introduced engineering best practices to protect the integrity of production reporting.

Deployment safety · every change is proven before it ships
Step 01Change / model
Step 02Dev tables
Step 03CI/CD tests
gate
Step 04Production
Dataform
Transforms native to BigQuery
Semantic layer
Shared, unambiguous definitions
CI/CD
Tested before deploy, every time

No change reaches a business-critical dashboard until it has been built and tested against development tables and passed CI. If a test fails, the change never touches production.

05 / Results

Clear visibility — and a foundation to build on.

MyArtBroker’s team now has clear, dependable visibility into customer conversion and customer lifetime value. They can identify which pieces of content are genuinely resonating and driving results — with the foundation in place to move towards predictive channel modelling.

01
Conversion & LTV

Dependable visibility into how customers convert and what they're worth over time.

02
Content that works

The team can see which content genuinely resonates and drives results.

03
What's next

The foundation is set to move towards predictive channel modelling.

Want this kind of clarity in your data?

A short call is enough to know if there’s opportunity worth pursuing in yours.