I've seen the same pattern in Spanish SMEs for years: the managing director opens four spreadsheets, two tabs of the ERP and an email from the sales rep just to find out how the month went. It takes a whole morning, and by the time they're done the numbers are already out of date. That's where business intelligence for small business stops being a big-company luxury and becomes a basic tool, even for a company with 15 employees.
That said, not everyone needs Power BI up and running at nine every morning. Some companies run perfectly well on a well-built spreadsheet, while others have spent two years paying licenses for a dashboard the managing director opens once a quarter. The difference between wasting money and getting it right is knowing when to make the leap and how to approach it.
In this article I'll walk you through what to consider before you dive in, which tools work well for SMEs, what it might cost you and how to set up a first project that doesn't end up gathering dust. No marketing fluff, just what we see in real projects.
What business intelligence is (in plain English) and why it matters to an SME
Business intelligence, or BI, means taking the data your company already has, scattered across different places (ERP, CRM, ecommerce, timesheets, bank), and bringing it together to answer business questions in seconds. Nothing more, nothing less. What used to take half a morning of building a pivot table now comes from a dashboard updated in real time.
What people often miss is that BI isn't just pretty charts. It's a clean data layer (usually called a data warehouse or, for SMEs, a data mart) that any tool can query for anything. That means the day you change your invoicing software, your reports don't break. It also means you can combine data that lives in different systems: Shopify sales with costs from Holded, a Spanish accounting app, for example.
Which companies is it right for?
If your annual revenue is under €500,000 and you have two people in the office, Excel and a couple of ERP reports will probably do the job. Beyond that, once you have several sales channels, a sales team and a warehouse, and your information lives in three or four different systems, BI starts paying for itself. What we usually see at Interconecta is that the tipping point comes somewhere between 15 and 50 employees.
Clear signs you need BI (and signs you don't yet)
Before you spend a single euro, check whether your company shows any of these signs. You don't need all of them, but if you tick three or four, you're probably already late.
- You spend more than two hours a week building manual reports that always end up with the same numbers.
- When someone asks “how are we doing this month?”, the answer takes more than five minutes.
- Your data sits in more than three systems: ERP, CRM, ecommerce, marketing tool, the sales rep's spreadsheet...
- You've made an important decision (hiring, launching a product line, raising prices) on gut feeling because you didn't have the numbers in time.
- Each department has its own version of the truth, and meetings turn into arguments about which figure is right.
When you DON'T need it (yet)
If all your data sits in one system and your ERP's standard report tells you what you need, don't jump into BI because it's trendy. You're better off spending that budget on training someone on your current ERP. It also makes no sense if your basic information isn't in order: if your CRM hasn't been updated in six months, no dashboard in the world will make it look good. Get your house in order first.
BI built on dirty data is like painting the facade of a cracked building: it looks great for three weeks, then the stains come through.
Real tools: Power BI, Metabase, Looker Studio and more
The BI tools market is fairly mature, and for an SME three or four options cover almost everything. Here are their pros and cons, based on what we see in projects.
Power BI
Power BI is a particularly good fit for companies that already use Microsoft 365. The Pro version requires per-user licenses, so check which plan you need to share reports. It connects to SQL Server, Dynamics, Excel and other common sources.
Metabase
Metabase is open source and lets you choose between self-hosting and a cloud version. You need to factor in infrastructure, maintenance and access permissions. Its interface makes it easy for business users to query databases such as PostgreSQL or MySQL.
Looker Studio (formerly Google Data Studio)
Looker Studio connects to Google Analytics, Google Ads, BigQuery and Sheets. It can be a good first step for marketing reports. Check the terms of third-party connectors and the limitations when working with transactional ERP data.
Other options
Tableau is extremely powerful, but expensive and complex: for SMEs it's rarely worth it. Qlik Sense has an interesting offering for manufacturing companies. And there are newer options like Superset (open source) that we see more and more in custom projects. The choice depends less on the tool and more on where your data lives today.
Which business KPIs to set up first (and which to leave for later)
The most common mistake I see: in the first few weeks, the client wants to measure everything. Web traffic, conversion, margin per product, stock turnover, absenteeism rate, NPS... and three weeks later nobody looks at the dashboard because it's overloaded. Start with less.
For a typical SME, the KPIs that really move the business in a first dashboard usually come down to five or six:
- Monthly and year-to-date revenue, compared with the same period last year.
- Gross margin by product or service line. Selling isn't enough: you need to see what actually makes money.
- Active customer base and new customers this month.
- DSO, or average days to collect. Especially if you work B2B with long payment terms.
- Sales pipeline: open opportunities and a 30/60/90-day forecast.
- Projected cash flow 8 to 12 weeks ahead.
With those six numbers, the managing director makes 90% of the important decisions. Everything else (productivity per employee, operational metrics by department) is added in a second phase, once the team has got into the habit of checking the dashboard every morning.
A useful rule
Every KPI needs an owner and a decision attached to it. If a metric doesn't trigger any action when it turns red, it isn't a KPI, it's a curiosity. And curiosities don't belong on an executive dashboard.
What a BI project costs a Spanish SME
When budgeting for BI, it helps to distinguish three types of scope:
- Basic project (1–2 sources, one dashboard and the essential KPIs): you need to define the metrics, connect the data and validate a first report with the team.
- Mid-sized project (several sources and dashboards by department): you need a shared data model and common data quality rules.
- Advanced project (data warehouse, ETL pipelines and predictive analytics): you need an architecture sized to your data volume and level of automation.
Licenses, connectors and database hosting are priced separately. Ask for a proposal that separates implementation from recurring costs and spells out who will maintain each component.
One piece of advice we always give: budget for internal hours too. You can outsource the project, but someone in your company will need to spend between 20 and 60 hours defining KPIs, validating data and taking part in training. If nobody has those hours, you're better off not starting.
How to run a first BI project that doesn't end up gathering dust
The approach that works best for us in business intelligence dashboard projects is to work in short sprints, with weekly deliveries and a single point of contact on the client side. Here's a summary of the phases:
Phase 1: discovery (1–2 weeks)
Sit down with management and department heads to define which decisions they want to make with the data. Not what data exists, but what decisions. That's where the real KPIs come from. You also map the sources: where each piece of data lives, in what format and who updates it.
Phase 2: data model (2–4 weeks)
We set up a central database (usually PostgreSQL, or BigQuery if volumes are large) and schedule automatic loads from each source. This is where API integrations with your ERP, CRM and other systems come in. It's the least visible phase and the most important one: a bad data model drags problems along for years.
Phase 3: dashboards and training (2–3 weeks)
Building the dashboard, validating it with each manager and training the team. Training is key: if people don't know how to read the dashboard, they'll go back to calling finance and asking for the figure “by hand”.
Phase 4: evolution
After three months, a review. What gets looked at, what doesn't, what's missing. BI is a living thing that grows. The companies that get the most out of it treat the dashboard as an internal product and improve it every quarter.
A BI project isn't finished the day the first dashboard is delivered. It's finished the day the team stops asking for Excel reports.
BI + artificial intelligence: how far to go (and where to stop)
For two years now, BI tools have seen a boom in AI features. Power BI has Copilot built in, Metabase has launched natural language questions, and Looker Studio lets you query your data conversationally with Gemini. That's great, but you need to understand what it does and doesn't bring to an SME.
What already works today: asking the dashboard a question in plain language (“which customer bought the most from us this quarter?”) and getting a reasonable answer. Automatic anomaly detection (it alerts you if sales in a category drop 30% compared with the previous month). Simple time-series forecasts for cash flow or demand.
What still doesn't work well with Spanish-language data and small datasets: advanced predictive models of customer behavior, automatic segmentation and fully generated narrative reports. For that you need data volumes most SMEs don't have, and custom artificial intelligence solutions that go beyond pure BI.
My recommendation: build a solid BI foundation first and leave the AI layer for phase 2, once you have clean data and a team that knows how to read dashboards. Skipping steps here is expensive and almost always leaves everyone frustrated.
Common mistakes when implementing BI in an SME (and how to avoid them)
After a fair number of projects, some mistakes repeat with almost comical regularity. Here they are, so you can skip them.
- Wanting the perfect dashboard on day one. Better an imperfect delivery in four weeks than a perfect one in six months (which never arrives).
- Not appointing an internal project owner. Without someone on the inside who validates data and keeps the team moving, the project dies.
- Confusing BI with reporting. Reporting tells you what happened. BI helps you understand why it happened and what to do about it. If all you want is lists, you don't need BI.
- Choosing a tool because it's trendy. “I want Power BI because my brother-in-law uses it” is not a criterion. The tool has to fit your stack and your team.
- Not investing time in data quality. 60% of the effort in a BI project goes into cleaning and normalizing data. If that sounds exaggerated, you haven't done a serious project yet.
- Forgetting the GDPR. If you're going to combine customer data for analysis, review your record of processing activities. The Spanish Data Protection Agency (AEPD) has fined several companies for cross-referencing data without a clear legal basis.
Avoid these six mistakes and you'll have a BI project that really changes how decisions are made in your company. The rest is execution, and with a good team (in-house or external) it gets done.
Business intelligence for small business is no longer just for multinationals. Today a 20-person SME can have its dashboard up and running in four weeks, with a reasonable investment and tools that update themselves. The hard part is no longer the technology: it's knowing what you want to measure and why.
If you decide to take the plunge, start small, measure what really hurts and resist the temptation to put 40 charts on your first dashboard. A dashboard with five well-chosen numbers updated daily is worth more than an 80-page report nobody opens. And when it's time to choose a tool, try two or three before you sign anything: annual licenses weigh more than you'd think on a small company's P&L.
The day you stop running your company on gut feeling and start running it on data, you'll realize how much time you'd been losing. You don't need to be Amazon for that. You just need to start.