A data warehouse is a single store that collects copies of data from a company's other systems and organises them for reporting and analysis. Those systems include the accounting system, CRM, point of sale and spreadsheets, and the warehouse does not replace any of them. It keeps the history they overwrite, applies one agreed definition to every term, and gives dashboards, reports and questions one place to read from.
You will recognise this
A business without a warehouse rarely describes it as a data problem. It sounds more like this.
- Every report starts with somebody exporting from three systems.
- The ERP has all our data, but getting anything out of it takes a specialist.
- Finance and sales both have a customer count, and they do not match.
- We updated the price list and lost the old prices.
- Only one person knows how the monthly numbers are put together.
- Every new question turns into a small project.
- We bought a reporting tool and it shows us five versions of the same number.
- If our supplier walked away, I am not sure we would still have our reports.
Many systems, no single version of the truth
Each of your systems is right about its own corner, and none of them can answer a question that crosses two. The accounting system knows what was invoiced, and the CRM knows what was sold. The point of sale or booking system knows what went through the till, and the spreadsheets hold everything the systems do not.
So every answer that crosses them is assembled by hand. Someone exports, pastes, reconciles the totals that disagree and rebuilds the charts. The result depends on who built it, and when that person is on leave, the number moves.
A reporting tool on its own does not fix this. Pointed straight at your operational systems, it shows you each system's version of every number, side by side.
How does a data warehouse work?
A data warehouse does three things: it loads your data, keeps its history and organises it. It loads a copy of the data from each system you run, on a schedule, usually overnight. It keeps the history, including values your systems overwrite, such as last season's prices or a customer's previous address.
It also organises the data under one agreed set of definitions. Revenue, customer and margin then mean the same thing in every report that reads from it.
Your systems keep doing their jobs. The accounting system still raises invoices and the CRM still manages the pipeline. The warehouse sits beside them, connected rather than replacing them, and reads from them overnight without changing anything inside them.
Our article, What a data warehouse actually is, and what it is not, covers the longer version. The short version: one copy of everything, one definition of every term, built for questions.
Why the definitions come first
Definitions come first because a report can only be as consistent as the words it counts. Early in the build, we agree with your team what each key term means: an active customer, recognised revenue, a region, a late order.
We call that agreed set the ontology. It covers everything your business counts, from customer and revenue to product, spend and employee, each named once and linked to the others.
It sounds academic, but it is the practical core of the work. Sales counts revenue when a deal is signed, and finance counts it when it is invoiced. Both are right by their own definition.
Connect the two systems without settling the word and you get two numbers on one screen instead of two numbers in two spreadsheets. A definition that lives in one analyst's head is a reporting outage waiting to happen.
Written down and modelled in the warehouse, a definition is inherited by every number downstream. The dashboards built on it then read the same agreed figures.
Three definitions become one
Is an ERP a data warehouse?
No, an ERP is not a data warehouse. An ERP runs the business day to day and is built for that: raising invoices, moving stock, posting journals. A warehouse is built for asking questions across the business, which is a different job.
An ERP sees its own data, not the CRM's, the booking system's or the spreadsheets'. Questions that cross those systems, or that ask how something changed over time, need a place built for them.
If you run an ERP, it becomes one of the warehouse's most important sources. We do not replace it, and we are not IT. We make the data inside it usable alongside everything else you run.
Who owns the data warehouse?
You own the data warehouse, outright, from start to finish. It is built in your environment, under your account, with your credentials. There is no version of this where leaving us costs you your data.
If we parted company tomorrow, it would keep running, and it would still be yours: the loads, the history, the definitions and every dashboard that reads from them. No lock-in, no ownership grab.
That is the difference between buying an asset and renting a report. It is the question to ask any supplier before any other.
We make your data usable; we do not take custody of it. Access is only what you agree to, and the warehouse sits inside your own account, not ours.
Your account or the supplier's?
What the warehouse feeds
Everything you see sits on the warehouse, so it all reads the same agreed figures.
Dashboards
Tiered by role, in Power BI or Looker Studio, current to yesterday. Leadership sees the whole business; each team sees its own numbers at its own depth.
Automated reports
The monthly pack builds itself from the same figures and lands in the right inboxes without anyone sending it.
Flags
Raised by rules rather than by someone noticing: a KPI marked behind, a job too long in one stage, a regular customer who has stopped ordering.
The AI layer
Ask a question in plain language and get an answer from your warehouse. It answers from your ontology and your data, not from a guess.
Your analysts
The people who used to build reports by hand get their time back for the analysis you hired them for, working from one place.
ERP, spreadsheets or a data warehouse?
The numbers behind your reporting can live in three places, and each is built for a different job.
| ERP or accounting system | Spreadsheets | Data warehouse | |
|---|---|---|---|
| Built for | Running operations | Anything, by hand | Asking questions |
| Sees | Its own data | Whatever was pasted in | Every system you run |
| History | Depends on the system | Whatever someone saved | Kept |
| Definitions | Its own set-up | The builder's | Agreed once, by you |
| Updated | Live, for its own records | When someone has time | Overnight, on its own |
| Heavy questions | Not its main job | Rebuilt by hand | What it is for |
| If one person leaves | Unaffected | Often stops | Unaffected |
What changes, in one business
A group in Malta runs several restaurants, and each outlet has its own point of sale. The group has one accounting system and a booking system for tables and private events. Staff rosters and supplier prices live in spreadsheets.
- Four people carry the numbers by hand every week, from the tills and the accounting system into one file.
- Each outlet manager counts covers slightly differently, so the outlets cannot be compared fairly.
- When a supplier changes its prices, the old ones are overwritten, and nobody can say what last season's margin really was.
- Linking private events to the revenue that followed means cross-referencing two systems by hand, so nobody asks.
- The finance lead is the only person who knows how the weekly file is built.
- Every till, the accounting system and the booking system load into one warehouse overnight.
- A cover has one definition, agreed with the outlet managers, so the outlets are compared on the same basis.
- The warehouse keeps the price history the spreadsheets used to lose.
- Outlet managers open their own outlet's numbers; the group sees all of them.
- Questions that crossed two systems take minutes instead of a week, and the AI layer answers them in plain language.
- The warehouse sits in the group's own account, so none of it depends on us.
The four people who carried the numbers now read them. The weekly file no longer stops when one person is away.
Your warehouse at the end of the first month
Implementation is the first month: three to four weeks from an accepted quote to a live dashboard on a warehouse you own.
A warehouse in your account
Built in your environment, under your credentials, on the platform we recommend before the build, based on what you already run.
Overnight loads
The systems behind the first dashboard loading on a schedule, with nobody exporting anything.
The first definitions
The terms the first dashboard depends on, agreed with your team, written in plain English and modelled once.
A live dashboard
The first view on top of the warehouse, current to yesterday, with tiered access set by role.
How it works
- A data strategy callThirty minutes with Glen, free. You leave knowing where your data is fragmented and whether 110 Analytics is the right fit.
- The data auditAbout three hours with Glen and your team, mapping every system and where each figure comes from. Then a quote for the work.
- The buildThe warehouse and the first dashboard, three to four weeks from an accepted quote. Implementation is the first month. Your team gives time to agree the definitions; we do the building.
- The fractional teamAn ongoing senior team keeps the loads running, the definitions current and the next source connected.
What we do not do
- We do not take custody of your data. It stays in your account.
- We do not build the warehouse in our own account and rent it back to you.
- We do not replace your ERP, accounting system or CRM. We are not IT.
- We do not treat the warehouse as a project we finish and leave. Systems change and definitions need revisiting, so the fractional team keeps it current.
- We do not switch the AI layer on before the warehouse can support it. It is the last thing we turn on, not the first.
Glen Sultana founded 110 Analytics after a career built where the numbers move daily. Our people have run analytics for tier-one companies, S&P 500 firms and high-growth technology businesses. Glen is the person on the first call, and he runs every data audit himself.
- Built in your environment, under your account
- No lock-in: if we part company, it keeps running and stays yours
- Based in Malta
Straight answers
What is a data warehouse?
A data warehouse is a single store that collects copies of data from a company's systems, such as accounting, CRM, point of sale and spreadsheets. It organises that data for reporting and analysis, keeps the history the source systems overwrite and applies one agreed definition to every term. It does not replace any of those systems.
Do we need a warehouse if we already have an ERP?
If your reporting needs data from outside the ERP, yes. An ERP is built to run the business day to day, and it sees its own data, not the CRM's or the spreadsheets'. Your ERP partner keeps running the ERP; the warehouse reads from it alongside everything else.
Who builds data warehouses in Malta?
110 Analytics is a data consultancy in Malta that builds data warehouses for established companies turning over roughly €1M to €10M a year. We agree the definitions with your team first and build the warehouse in your own account. Then we put the dashboards, automated reports and AI layer on top.
Where does the warehouse live?
In your own cloud account, under your credentials. We recommend the platform before the build, based on what you already run, and for a business on Google Workspace, BigQuery is the natural fit.
What happens if we stop working with you?
It keeps running, and it is still yours: the loads, the history, the definitions and the dashboards. That is the point of building it in your environment rather than ours.
Who can see our data?
Access is only what you agree to, and the warehouse sits inside your own account, not ours. We make your data usable; we do not take custody of it. Security of the account itself follows your own provider settings and IT policies.
How long does it take to build a data warehouse?
The warehouse and the first dashboard are live three to four weeks after you accept the quote; implementation is the first month. More systems and definitions are added after that by the fractional team.
What does a data warehouse cost?
The work is quoted at the end of the data audit, because the figure depends on how many systems you run and what the first dashboard needs to do. There is no price list, because a price list would be wrong for you. The cloud storage and computing sit in your own account, so the provider bills you directly.
Can we use AI on our data without a warehouse?
You can try, but an AI tool is only as good as what it reads from. If your numbers live in several systems with several definitions, it has no single agreed version to read. Our AI layer answers from your warehouse and your ontology, and it is the last thing we switch on, not the first.
Warehouse terms, in plain English
- Data warehouse
- A single store of copies of your business data, organised for questions rather than for running operations.
- Ontology
- Your business's terms, each defined once and related to the others, that every report reads from.
- Source system
- Any system the warehouse loads from: accounting, ERP, CRM, point of sale, booking, e-commerce or spreadsheets.
- Overnight load
- The scheduled job that copies each source system's data into the warehouse every night.
- Data lake
- A store for raw data in its original form. A variation on where data is kept, not a substitute for agreed definitions.
- AI layer
- Plain-language questions answered from your warehouse and your ontology, not from a guess.