Business intelligence: a practical guide for Saudi companies

Business intelligence (BI) is the setup that turns a company's operational data into shared, regularly refreshed reports and dashboards, so managers decide from the same numbers. For a Saudi company it starts with a short list of KPIs, one trusted data model and clear access rules; the tool, such as Power BI or Data Studio, comes after.

This guide is for owners and managers who want reporting that no longer depends on someone rebuilding spreadsheets at month end. It covers the building blocks of a BI system, how to choose the first KPIs, the main tools and their prices as listed on 14 October 2026, a worked cost example, data location and the PDPL, the role of AI, and a 90-day rollout plan. O AI is not affiliated with Microsoft, Google or any other company named here.

What business intelligence is, and how it differs from one-off analysis

Saudi Arabia's National Data Management Office (NDMO) describes its business intelligence and analytics domain as analysing an organisation's data records to extract insight and draw conclusions [1]. Microsoft describes Power BI, one of the common BI platforms, as a business analytics platform for connecting, visualising and sharing data across an organisation [2]. The word that matters in both is organisation: business intelligence is analysis that has been made repeatable and shared.

A one-off analysis answers a question once. Business intelligence keeps the answers to recurring questions current for everyone who needs them, with one definition of each number and rules on who sees what. You probably need it when:

• the same monthly report is rebuilt by hand from several exports;
• two managers bring different sales figures to the same meeting;
• branch or regional managers wait days for numbers that already exist in your systems;
• decisions about stock, pricing or collections are made without current data because getting it is slow.

The five building blocks of a business intelligence system

Every BI setup, from a small company's single dashboard to a ministry's data platform, has the same five parts.

For the data model, Microsoft recommends a star schema. Dimension tables describe the things you analyse, such as products, customers, branches and dates, and are used for filtering and grouping; fact tables store the events, such as sales lines or stock balances, and are used for summarising [3]. When data volumes are large or you must keep history, Microsoft advises building a data warehouse with scheduled extract, transform and load (ETL) processes first, and connecting the BI model to it [3].

Integration is usually the hardest block. If your systems cannot export clean data on a schedule, our guide to custom software development in Saudi Arabia explains when connecting existing systems beats buying new ones.

Which KPIs belong on the first BI dashboard?

Start with five to ten KPIs that someone already acts on. A dashboard with forty charts is usually a sign that nobody chose.

Write a one-line card for each KPI before anyone builds anything:

• Name and definition in plain words.
• Formula, including what is excluded.
• Source system and field.
• Owner, the manager who answers for it.
• Refresh frequency and target.

The definition step prevents most BI arguments. In one of Microsoft's training scenarios, different reports disagreed because one counted returns in revenue and another did not; the fix was to define net revenue once, as gross sales minus returns and discounts, inside a certified model that every report uses [4].

Add a data quality check to each KPI. NDMO's standards give an example of a quality rule that raises an alert when fewer than 90% of records have a field filled in [1]; a BI equivalent is a small tile showing how many sales lines arrived without a branch or product code.

Business intelligence tools compared: Power BI, Data Studio and the alternatives

No single tool is best for every company. The practical choice usually follows the office suite you already pay for, and the cost depends less on the tool than on how many people need to view reports.

Google renamed Looker Studio, and its documentation now says plainly that "Looker Studio is now called Data Studio" [8]. For companies that must keep reports on their own servers, Microsoft offers Power BI Report Server for on-premises reporting [2].

Other BI products exist, such as Tableau and Qlik. Compare them the same way: the cost per viewer, where data is processed, Arabic support in the interface and reports, and whether your team can maintain them.

Illustrative example: BI for a building-materials retailer in the Eastern Province

Illustrative example. A building-materials retailer has five branches in Dammam, Al Khobar, Jubail, Qatif and Al Ahsa, an accounting system and a point-of-sale system. Today, two accountants spend three working days each month compiling branch reports. The company wants one dashboard for 25 managers, built and maintained by two staff.

Budget separately for the setup work: connecting the two systems, agreeing definitions and building the model. No licence price includes it, and it is where most of the value is created. The SAR 150 hourly rate and the 40 saved hours are assumptions for this example; use your own figures.

The choice between the two tools follows the suite the retailer already runs. On Microsoft 365, Power BI Pro is the natural route; on Google Workspace, Data Studio's no-cost tier may be enough to start.

Data location, the PDPL and access control for BI in Saudi Arabia

Where BI data is stored and processed matters once reports include personal data, such as customer names or employee records.

• Microsoft Fabric and Power BI regions. Microsoft's region list, updated on 1 October 2026, shows three Middle East regions: Qatar Central and UAE Central for Power BI only, and UAE North for all Fabric workloads. Saudi Arabia is not listed [9].
• Google BigQuery. Google lists Dammam (me-central2) among BigQuery's Middle East regions [10], so a company can keep its warehouse in the Kingdom and report on it with Data Studio, which connects to BigQuery [7]. Check separately where the reporting layer processes data.
• Copilot in Power BI. For tenants outside the US and EU data boundaries, Copilot is off by default unless an admin allows data to be processed outside the tenant's geographic region [11].

Saudi Arabia's Personal Data Protection Law (PDPL) limits personal data to the purpose it was collected for and to the minimum the purpose needs [12]. Under the Implementing Regulation, a written impact assessment is required when two or more sets of personal data from different sources are collected, compared or linked, which describes many BI models that join CRM, loyalty and sales data; anonymised data, from which a person cannot be re-identified, is no longer personal data [13]. The safest BI design shows aggregates by default and keeps names out of the model unless a report truly needs them.

Our guide to data residency in Saudi Arabia explains when personal data may be processed abroad. This section is general information; for a specific case, check SDAIA's guidance or ask a licensed advisor.

Access control: row-level security and workspace roles

In Power BI, row-level security filters the rows each user can see, for example one branch's data for one branch manager. It applies only to users with the Viewer role; workspace Admins, Members and Contributors see all the data [14]. Give managers the Viewer role and keep editing rights to the people who build the reports.

If your company is a public entity, or a business partner handling government data, NDMO's standards apply more directly. Their BI domain requires a BI plan, a prioritised portfolio of use cases, training, KPIs for the analytics portfolio, and a register of use cases [1].

AI in business intelligence: what Copilot and natural-language questions need

AI is now built into BI tools, but it depends on the model underneath.

• Copilot in Power BI answers questions about reports, summarises pages and helps authors build visuals and write DAX. It needs paid Fabric capacity of F2 or higher, or Power BI Premium P1 or higher; a Pro or Premium Per User licence alone is not enough [11].
• Preparation matters. Microsoft warns that without a prepared semantic model, Copilot can misinterpret the data and return generic or inaccurate results [11].
• Language. For its standalone and app-based Copilot experiences, Microsoft states that prompts in languages other than English are not officially supported yet [11], so test Arabic questions before you promise them to managers.
• Data Studio lets technical users build conversational agents on BigQuery data and share them with business users inside Data Studio [7].

The lesson for a Saudi SME is simple: clear KPI definitions and a clean star schema come first, because they are what any AI assistant reads.

A 90-day business intelligence rollout plan

NDMO's standards list what a BI use-case implementation plan should cover, from requirements and high-level design to environments, development, testing, deployment schedule, resources and acceptance criteria [1]. Scaled down for a company of 50 to 500 staff, that becomes a 90-day plan.

Common business intelligence mistakes

• Buying licences before agreeing KPI definitions.
• Building one dashboard per request until nobody knows which is current.
• Skipping the reconciliation to the accounting system, so the first wrong number destroys trust.
• Giving every manager editing rights, which bypasses row-level security.
• Measuring the project by dashboards published instead of decisions changed.

Your next step

This week, list the five numbers your management meeting argues about most, and write a KPI card for each: definition, formula, source, owner and target. That list is the specification for your first business intelligence dashboard, whatever tool you choose. Then check which tool your current Microsoft 365 or Google Workspace subscription makes easiest, and where its data will be processed. You can read about O AI, a software and AI company in Al Khobar, on our home page.