Our offers · Rollout
Data governance & platform rollout
Governance, platform and delivery processes. What we put in place, the skills it calls for, and a three-year programme that gives the measure of it.
The offering
Governance, a platform, and the processes that tie them together.
Rolling out a data and AI platform engages several aspects of governance, process and technical architecture, all of which have to be put in place to deliver business value.
Governance
We support you in shaping a governance model suited to your company.
Roles held by the business
Governance is built around the business, with named product owners and data domain owners. Together they lead the data community, set a multi-year vision and prioritise deliveries day to day.
A data team in your shape
How the team supporting the business is organised varies from one company to the next: one shared team delivering data products across all domains, or a team dedicated to each of them.
The foundations
A data catalogue, documentation that is centralised, shared and maintained, and alerting on data quality levels.
Platform and delivery processes
The ability to deliver depends intrinsically on how the teams are organised and on the processes put in place.
Explicit validation gates
Sign-off on specifications, acceptance testing by the data team then by the business, code review and approval: what it takes to keep development efficient and maintenance viable over time.
A three-layer foundation
A methodology that moves data from raw layers to a unified model, through to the data products placed in the hands of the business.
Structured skills
We support you in structuring the team around three key skills, whose remits do not overlap.
The three key skills
Business analyst
Documents business needs and turns them into specifications.
Data analyst
Assesses data quality levels and tests the specifications, before and after each delivery.
Data engineer
Custodian of the data model, builds the transformations in the platform and keeps the data products durable.
The path the data takes
The technical layers are the backbone of the methodology: each has its entry rule, and nothing moves on to the next without it.
Bronze
Raw data
Sources are ingested as they are, untransformed, with their history and timestamps.
Silver
The unified model
Definitions are settled, reference data reconciled, quality rules applied and measured.
Gold
Data products
What the business actually uses: metrics, dashboards and datasets, each with an owner and monitored usage.
Our expertise within the collective covers every one of these gates and secures your investment in a technical foundation that delivers business value.
Inside an assignment
This offering, in its fourth year at a client grown through acquisitions.
The programme below is a client’s, anonymised; the figures are those recorded to date. It gives the measure of what the offering entails when it is run at group scale.
The starting point
Data fragmented by domain, with no common direction.
Each acquisition arrived with its own system, its own definitions, its own habits. The assessment surfaced six findings shared by top management, IT and the business.
Context
Growth through successive acquisitions, heterogeneous systems
Start
November 2024
Sponsorship
Top management, alongside IT
Trigger for the mandate
No consolidated view of the business, no direct business access to the data
Six findings at the outset
01
A view fragmented by the acquisitions
Each acquisition brought its own system and definitions. Nothing rolled up in a common language, neither at group level nor between regions.
02
Manual reconciliation, reporting behind the curve
Managers, finance, shared service centres: everything was reconciled by hand, from reference data to management indicators. The monthly flash landed after the 20th.
03
Client reporting produced by hand
Client reporting was produced manually by a dedicated team, with little capacity to draw insight from it without further manual work.
04
Source data locked on the IT side
The business could not reach operational data itself. From proof of concept to industrialised solution, no capacity to scale was available.
05
A multitude of tools, no common foundation
Snowflake, Redshift, Dataiku, Qlik, Power BI, Excel, SmartView, Cognos: that many technical building blocks, and no shared platform.
06
Clients with no visibility
Neither on the performance of their logistics provider, the subcontracted transport known as 3PL, nor on their environmental impact. Two expectations the market had already voiced.
The programme in figures
> 3
years
Data programme, still running
+40
FTE
Internal teams, integrators and the collective
14
domains
Launched and monitored, global to local
1
data factory
A new data platform, one single foundation
Objectives
From manual reporting to a single platform.
Five objectives set at framing, in direct answer to the findings of the assessment.
Today
Target
Data reconciled manually by every team, from reference data to management indicators.
Make data easy to reach
Monthly flash delivered after the 20th; no HR data available in one click.
Support decisions, shorten time to market
Client reporting produced by hand by a dedicated team.
Meet client expectations
From proof of concept to industrialised solution: not possible, for want of capacity to scale.
Put AI and the data assets to work
A patchwork of transformation and visualisation tools, with no common foundation.
One platform, simple to use
The approach
Four pillars of governance, one shared technical foundation.
The business owns its data again; the platform serves it rather than holding it.
Business autonomy
A business-centred operating model, rolled out domain by domain
Business teams, global to local, reach, explore and use the data according to their capacity, their skills and their appetite.
Data ownership
One certified source per data product
Every critical piece of data is owned by the business: defined, validated, documented and understandable, accessible and secured, monitored.
Data product governance
One global roadmap, prioritised collectively
Every data project and initiative, global to local, is visible to all domains, documented and accessible securely.
Data platform
One technical foundation serving the business
All certified data is made available through a single solution: an enterprise data model, shared and secured, with the right tools for each domain to stand on its own.
Three workstreams, three owners
The Data & Governance team
Data product delivery
Establishing the governance and operating model of each domain so it delivers the value expected.
The Data Factory
Technology foundation
Unifying the technical ecosystem with state-of-the-art solutions, serving every domain.
Data & Governance and Communications
Change management
Supporting the teams with the material the data transformation rollout requires.
The method
One domain at a time, a signed-off deliverable and concrete results at every step.
The programme never opened two fronts at once. Each phase ended on a deliverable signed off in the steering committee, and nothing started before it.
2 months
Framing and mandate
Data maturity assessment across all the entities brought in by the acquisitions, three-year target, mandate letter signed by top management.
Exit gate
Mandate signed, budget committed
5 months
Pilot domain
Finance first, because its definitions settle everyone else’s. Operating model, named owners, first products.
Exit gate
Monthly close produced on the new foundation
12 months
Platform and industrialisation
A single foundation sourcing operational data, catalogue, automated quality rules, direct business access to source data.
Exit gate
Release to production with no manual step
15 months
Domain-by-domain rollout
Procurement, Supply Chain, Logistics, Sales, HR then the domains that followed, up to 14 governed domains, on the same operating model.
Exit gate
Domains with an owner and a roadmap
ongoing
Handover and autonomy
Permanent hires, documented handover, shadowing, then a gradual exit by the collective, domain by domain.
Exit gate
Internal teams autonomous
The roles held by the collective
A governance model that keeps the populations apart.
Two families of profile, as within the collective: Data Governance Delivery on the business and governance side, Tech Data on the platform side. The collective held the leadership and architecture roles; internal teams did the rest, under that steering.
Data Governance Delivery
Business and Data & Governance populations.
Head of Data Governance
Data & Governance
- Defining the overall strategy with the business
- Leading the data community, global to regional
- Launching and supporting the domains
- Managing the Data Domain Managers
Data Domain Owner
Business
- Owns the domain strategy and roadmap, aligned with the business strategy
- Settles priorities and conflicts, point of contact for cross-cutting governance
- Reference point for the domain’s data expertise, coordinates the data teams involved
- Deepens the domain’s involvement in governance, training and method
Data Domain Manager
Data & Governance
- Setting up the domain operating model
- Supporting business transformation, product owners and beyond
- Implementing data governance processes
- Running the network of data correspondents
Global Data Product Owner
Data & Governance / Business
- Steers the data product roadmap and its priorities
- Manages the portfolio of business needs, writes and approves specifications
- Organises and signs off business acceptance with the users
- Owns documentation, communication and adoption of the data products
Tech Data
Data Factory population.
Head of Data Engineering
Data Factory
- Managing the data engineers
- Establishing platform best practices
- Technical reference point, cloud platform and AI development process expertise
Project teams — Delivery
Business / Data Factory / Data & Governance
- Mixed project teams mobilised per domain
- Combining business, technical and governance expertise to deliver the data products
Governed domains
14 domains launched and monitored in all, global to local.
Where the programme stands
What has changed, domain after domain.
14 governed domains, each with an owner and a roadmap, global to local.
Direct business access to source operational data: the bottleneck on the IT side is gone.
One enterprise data model, shared and secured, replacing the historical patchwork of tools.
Visibility given back to clients on the performance of their logistics provider (3PL) and on their carbon impact.
Artificial intelligence use cases delivered with a measured return, where there had been no way to start.
Governance that clearly separates the business, steering (Data & Governance) and platform (Data Factory) roles.
AI results
Two AI projects delivered in this setting, with a measured return.
Once the data foundations were in place, two AI use cases were industrialised in the Logistics & Transport domain.
Logistics & Transport · AI
AI-driven optimisation of operational resources
€70k
of cost avoided per day
Warehouse rosters optimised by cross-referencing parcel data, volume and sizes, with staff time records, both permanent and temporary.
Architecture
source system → S3 → Snowflake → Google Cloud
Logistics & Transport · AI
Automated review of client invoicing
€30k
additional monthly billing
Systematic analysis of client invoicing to spot activity carried out with no matching invoice, with business sign-off before issue.
Architecture
source system → S3 → Snowflake → Google Cloud
Assignments delivered by members of the collective; client anonymised.
What we take away
Five things that made the difference.
Bring top management in, and keep them in.
Launch iteratively, domain by domain and use case by use case.
Agree a minimum viable governance per domain, IT included.
Bring HR into the operating model and the new roles.
Set out a change management and communication plan.