Every platform you use shows you a different version of the truth. Google Ads says 212 conversions. Shopify says 180 orders. The CRM says 240 leads. So what is a data warehouse, and why does it stop that argument? It is the one place where all of those numbers are copied, cleaned and defined once, so everyone in the business reads the same figure.
What a data warehouse actually is
A data warehouse is a database built for answering questions, not for running software. On a schedule, every night or every hour, it pulls a copy of the data from each tool you use and stores it in one place, in one consistent structure, with full history.
That last part matters. Most SaaS platforms only keep detailed data for a limited window, and each one defines a "conversion" or a "customer" slightly differently. A warehouse keeps everything and lets you define each metric once, as a single source of truth that every dashboard and report reads from.
How data gets in
The modern pattern is ELT: extract the raw data from each tool with a connector, load it into the warehouse untouched, then transform it inside the warehouse into clean, joined tables. Keeping the raw copy means you can always rebuild a report when a definition changes, without going back to the source.
What comes out
Once the data is modelled, anything can read it: a live dashboard, a weekly email report, a finance export, or an AI assistant that answers questions about your business. The warehouse does not replace those tools. It feeds them all from the same numbers.
Warehouse vs database vs spreadsheet
The three get confused constantly, usually because all of them hold data in tables. The difference is what each is built to do.
| Question | Spreadsheet | Database | Warehouse |
|---|---|---|---|
| Built for | Ad-hoc working | Running an app | Analysis across sources |
| Updates itself | No | Only its own app | Yes, on a schedule |
| Keeps full history | If you remember to | Usually overwrites | Always |
| Handles millions of rows | No | Yes | Yes, cheaply |
| One definition of "revenue" | Per sheet | Per app | Once, for everyone |
A spreadsheet is still the right tool for one-off analysis. The trouble starts when a spreadsheet becomes the system: copied every Monday, edited by three people, and trusted by nobody.
Five signs your business needs one
You do not need a warehouse on day one. You need one at the point where reporting starts costing you more than the warehouse would.
Does this apply to you?
Probably not yet
With none or one of these, a well-organised spreadsheet is still the right tool.
What it costs and how long it takes
Tooling
For an SMB the sensible default is Google BigQuery. You pay for storage and for the queries you run, and at the time of writing the free tier covers the first 10 GB of storage and 1 TB of queries each month. A business with a few million rows of ad, order and CRM data typically spends very little beyond that. A connector pulls each source in on a schedule, and dbt handles the modelling for free.
The build
Connecting the first five sources, agreeing the schema so that "revenue" means one thing, and shipping one live dashboard is a two to four week project. Most of that time goes on agreeing definitions, not writing code.
The hard part was never the technology. It was getting marketing and finance to agree what a customer is.
— Every warehouse project, ever
Keeping it running
Pipelines break when a platform changes its API or someone renames a field. Budget a few hours a month of maintenance, or a support retainer if nobody in-house owns it. Good pipelines alert you when a load fails, so a broken connector is a five-minute fix rather than a month of wrong numbers.
In practice · Case study
Building a Data Warehouse
Turning five disconnected platforms into one source of truth.
Read the case studyCommon mistakes to avoid
Most warehouse projects that disappoint fail in one of three ways:
- Connecting everything first. Twenty sources and no agreed metrics produce a very expensive pile of tables. Start from the questions.
- Skipping the definitions. If "a lead" means something different to sales and marketing, the warehouse will faithfully report two different numbers.
- No owner. Someone needs to be responsible for the pipelines and the metric definitions, even if it is only a few hours a month.
If you also want to automate what happens with that data, such as sending the weekly report or routing new leads, the choice of automation tool matters too. Our Zapier vs n8n comparison covers that decision.
What to do next
List every tool that holds customer, order or spend data. Write down the three questions you cannot currently answer in under a minute. If any of those questions needs data from two of those tools, you have your business case.
You can see how this plays out in our data warehouse case study, or read more about our data and reporting service. If you would rather have someone look at your list with you, that is what the free 20-minute call below is for.
Key terms
- ELT
- Extract, Load, Transform. Copy the raw data in first, then reshape it inside the warehouse. The modern default.
- Schema
- The agreed structure of the tables: which columns exist and what each one means.
- BigQuery
- Google's pay-per-query warehouse. Cheap at SMB scale and no servers to manage.
- dbt
- An open-source tool for writing the transformations as versioned, tested SQL.
- Single source of truth
- One agreed place and definition for each number, so two reports can never disagree.
- Connector
- A pre-built integration that pulls data from a SaaS tool into the warehouse on a schedule.
Frequently asked questions
Is a data warehouse the same as a database?
How much does a data warehouse cost for a small business?
Do I need a data engineer to run one?
Can I just use Looker Studio connected to Google Ads?
Sources
- BigQuery pricing, Google Cloud
- What is dbt?, dbt Labs