Upload Excel to Power BI (2026): Your First Semantic Model

The moment a spreadsheet becomes a data platform: one Excel workbook, one import route, and the Power BI semantic model that every dashboard in this series will stand on.

Upload Excel to Power BI, and your budget workbook stops being a spreadsheet — it becomes a semantic model: a governed, refreshable data engine that every Q&A question, report, and dashboard in this series will run on. This part performs that first ingestion deliberately: preparing the workbook, choosing between import and connect, executing the upload step by step, and understanding exactly where your data physically lands in Microsoft's cloud.

Every step and every rule here is verified against current Microsoft documentation for 2026 — including a change most tutorials still get wrong: Microsoft has deprecated the old way of uploading local Excel workbooks and retired the legacy Excel/CSV import experience, with refresh and load cut-offs in July and August 2026. If you learned this workflow years ago, this article brings it current; if you are brand new, it starts from zero.

I'm Mostafa Amaan, Senior IT Officer with over 16 years of experience running enterprise data platforms — multi-branch PACS/RIS imaging systems, WANs, Windows and Linux servers, and data-driven e-commerce operations. I have watched reporting systems rise or fall on how their data first entered the platform; ingestion-day decisions echo for years. Let's make yours correctly.

Quick Answer: How to Upload Excel to Power BI

That is the entire motion in one breath. The rest of this guide slows it down — because the difference between a model that quietly serves you for the next 46 parts and one that breaks in Part 004 lives in the details of preparation and route choice.

Why Your First Semantic Model Starts with Excel

Power BI could pull your budget data from a database, a data warehouse, or a cloud API — and by the end of this series it will. So why start with a workbook? Because Excel is where budget-vs-actual data actually lives in most organizations, and because this import teaches the platform's core vocabulary — tables, columns, data types, refresh — on data you already understand. The Excel workbook is the gentlest on-ramp to a serious discipline.

From Worksheet to Semantic Model: What Power BI Actually Takes

Here is the mental shift that separates successful imports from frustrating ones: when you upload Excel to Power BI, you are not embedding the file — you are asking Power BI to read specific structures inside it and rebuild them as a semantic model. Microsoft describes the semantic model as the engine of a car and the data as its fuel; the Excel workbook is simply the fuel canister you carry the first fill in. Power BI reads what it recognizes — tables and data models — and quietly ignores the rest. One rule is absolute, and Microsoft states it without exceptions: the data must be in a table or in the workbook's data model.

I came to this discipline from the other side of the fence — years of keeping enterprise reporting honest on hospital PACS/RIS systems, where a study cannot exist in two contradictory states, and multi-branch e-commerce where one product's stock count cannot differ between two reports. A semantic model is that same discipline, materialized in Power BI: one copy of the data, one definition of each metric, many consumers. If the underlying concept of structured, governed data is new to you, our article on why databases exist covers the foundation in depth.

What Power BI Picks Up — and What It Leaves Behind

Open your budget workbook and look at it through Power BI's eyes. The table below maps every common workbook ingredient to its fate during import — according to Microsoft's Get data from Excel workbook files documentation:

The pattern is easy to summarize: Power BI imports your data and your data model; it rebuilds your presentation layer itself. Formatting, charts, and PivotTables stay behind on the worksheet — and that is a feature, not a loss, because Part 006 of this series will rebuild them better.

Power BI Service vs. Desktop: Where Should You Ingest Your Excel Data?

When users search how to upload an Excel file to Power BI, they frequently hit an immediate architectural fork in the road: should you upload through the web browser at Power BI Service (app.powerbi.com), or should you open Power BI Desktop on your PC and click Get Data? Both routes ingest Excel workbooks, but they serve two fundamentally distinct analytical stages.

In this stage of our series, we deliberately upload via the Power BI Service web interface. This creates a lightweight, cloud-first semantic model directly inside OneLake without requiring local software installation, administrator privileges, or Windows-only hardware dependencies. It is the fastest, cleanest route for business teams who have clean, tabular spreadsheets and want to immediately query them using natural-language Q&A (which we unlock in Part 004), explore fields in the browser, and publish executive dashboards on the web.

Conversely, Power BI Desktop is an authoring workstation. As detailed in Microsoft's Power BI Desktop documentation, Desktop is where you turn when your raw data is messy: when dates must be parsed across columns, text needs splitting, multiple worksheets require appending, or you need to write custom M code in the Power Query Editor. Desktop is also where you construct multi-table star schemas and write complex DAX measures from scratch. We give that entire heavy-lifting ETL workflow its own dedicated masterclass starting in Stage 4 (Part 018 through Part 026).

Understanding this distinction upfront eliminates confusion: if your Excel data is already formatted into clean tables, the Power BI Service ingestion we execute in this guide provides immediate cloud agility. If your data demands complex pre-cleaning, you can still follow this cloud-first workflow to master semantic models, knowing that Power BI Desktop ETL will expand your toolset later.

Service web ingestion vs Power BI Desktop: web upload creates a rapid cloud-first model in OneLake, while Desktop provides deep Power Query data shaping.

Preparing the Budget Workbook: Five Minutes That Save Five Hours

Assume our case-study workbook is the classic finance shape: a Budget sheet, an Actuals sheet, and a summary tab full of formulas someone built in 2019. Before touching Power BI, run a pre-flight pass on the source — because the import copies whatever structure exists, including bad structure. Every minute of preparation here removes an hour of model cleanup later.

Step 1 — Format Ranges as Excel Tables (Ctrl+T)

Select each data range — headers included — and press Ctrl+T (or Insert > Table), then confirm My table has headers. Do this for tblBudget, tblActuals, and any other logical entity in the workbook. Formatted tables get named, typed columns in Power BI's data pane; loose ranges get nothing. Two structural habits matter while you are in there: no blank rows or columns inside the table, and one table per logical entity — resist the temptation to mix budget and actuals into a single sheet with subtotal rows scattered between.

Step 1 in action: Ctrl+T opens the Create Table dialog with the boundaries pre-filled and My table has headers checked — the same five-second move for every data block in the workbook.

Step 2 — Name for Semantics, Type for Power BI

Your column headers become your field names for the rest of the series — write them for humans and engines alike: Month, Department, Category, Budget Amount, Actual Amount — never Column1. Then fix the data types in Excel itself: dates typed as Date, amounts typed as Currency or Number, codes typed as Text. Power BI respects properly typed columns on import and re-asks fewer questions later. The Month column deserves special care — typed as a real date, it becomes the backbone of time-intelligence calculations in Stage 3.

If your budget history lives across a dozen worksheet tabs that grew one per year, consolidate before importing. Our guide to merging Excel sheets with VSTACK and HSTACK shows the modern formulas for stacking tabs into one table in minutes.

The typed result after Step 2: headers that read as field names (Department, Total Budget, Total Actuals), amounts stored as currency and a numeric percent column — the structure Power BI imports without follow-up questions.

Three Spreadsheet Traps That Silently Break Power BI Imports

In sixteen years of administering enterprise data platforms, multi-branch network systems, and relational databases, I have learned one immutable truth: reporting errors almost never originate in the business intelligence engine — they are born at the ingestion gate. Excel spreadsheets built for human eyes frequently violate the fundamental laws of tabular data. Before converting your budget ranges with Ctrl+T, inspect your sheet for these three silent killers:

1. Merged Cells in Headers or Data Rows: Merging cells across columns (such as centering a title like "Q1 Financial Plan" across columns B through F) destroys columnar schema integrity. When Power BI parses a merged block, it attributes the text value exclusively to the top-left cell; every other merged cell is evaluated as an empty null. If your header row contains merged cells, Power BI renames subsequent columns to anonymous defaults like Column1 and Column2. If data rows are merged, entire records lose their categorical association. The fix: Select the entire worksheet, click Merge & Center to unmerge all cells, and use Format Cells > Alignment > Center Across Selection if you desire visual centering without structural merging.

2. Stacked and Multi-Tier Headers: Financial accountants frequently stack three or four rows of titles, dates, and currency notations before raw figures begin (e.g., Row 1: Division, Row 2: Region, Row 3: Account). Power BI requires exactly one single header row containing unique field names. When it encounters multi-tier headers, it either assumes row 2 represents data records (corrupting data types) or fails to identify field boundaries altogether. The fix: Flatten multi-row metadata into a single clean header line (e.g., Dept_Operations_Budget) and move contextual notes outside the table boundaries. If you manage complex worksheets, our guide on cleaning spreadsheet structures and formulas outlines practical methods for normalizing tabular sheets.

3. Subtotal and Grand Total Rows Baked into the Table: In standard Excel accounting, analysts insert =SUM() subtotal rows beneath every department, plus a grand total row at the bottom. If those calculation rows fall within the formatted Excel table boundaries, Power BI ingests them as standard factual records. When you later drop the budget column into a bar visual or write a DAX calculation like SUM(tblBudget[Budget Amount]), Power BI aggregates the individual entries plus the subtotal rows — effectively doubling your numbers without raising an error. The fix: Delete all manual subtotal and total rows from the table; dynamic aggregation is Power BI's core job, executed cleanly through visuals and DAX measures in Stage 3. For more on relational data principles, see our overview on designing normalized tabular data.

The three spreadsheet traps to eliminate before Ctrl+T: unmerge cells, collapse multiple header rows to one, and remove subtotal rows.

Uploading the Workbook: The 2026 Import Flow, Step by Step

With the workbook pre-flighted, the import itself is under ten clicks. The flow below reflects the current Power BI service interface — verified against Microsoft's data sources for the Power BI service documentation — not the retired experience that older tutorials describe. Follow each step, run its check, and only then move on.

1. Step 1 — Open the target workspace: Sign in at app.powerbi.com with the account you set up in Part 002 and open My workspace in the navigation pane. On the free license, this is the one container where you can create content — and where our budget model lives for the whole of Stage 1. Check: the workspace canvas and navigation pane are visible.
   

   Step 1: signed in at app.powerbi.com with My workspace open — the free-license container where the budget semantic model will live through Stage 1.

2. Step 2 — Start the import: Select + New item, choose Semantic model from the selection pane, then pick Excel. (Interface labels shift monthly in Power BI; if you see slightly different wording, you are looking for the flow that creates a semantic model from a file — not the plain Upload button, which behaves differently, as the next section explains.) Check: a file picker opens.
   

   Step 2: + New item → Semantic model → Excel — the 2026 path that builds a model from a file. Interface labels shift monthly; avoid the plain Upload button.

3. Step 3 — Choose the file and its route: Either browse to the local .xlsx file (Power BI imports a snapshot copy) or select it from OneDrive for work or school or SharePoint (Power BI creates a live connection). For our budget workbook, OneDrive is the recommended route — the reason unfolds in the next section. Check: the correct workbook is selected.
   

   Step 3: choosing the workbook and its route — a local .xlsx snapshot copy versus a live OneDrive for work or school / SharePoint connection.

4. Step 4 — Let Power BI read the structures: The service scans the workbook for formatted tables and its data model, then builds a new semantic model with one table per Excel table — tblBudget and tblActuals arrive as first-class model tables. Check: a confirmation appears and the semantic model shows up in your workspace under semantic models.
   

   Step 4: the service reads the workbook's formatted tables, then reports the new semantic model in the workspace — no Power Query, no manual shaping.

5. Step 5 — Explore the model: Open the new semantic model and use Explore data: your budget fields appear in the data pane, ready for ad-hoc visuals and question boxes. This screen is your first proof that the ingestion worked. Check: Month, Department, and the amount columns are listed with correct type icons.
   

   Step 5: the Explore data view — the semantic model's fields ready for ad-hoc analysis and the Q&A deep-dive coming in Part 004.

6. Step 6 — (Optional) Build a first preview visual: Inside Explore, drop Department on the axis and Budget Amount on the values — one column chart, no saving required. It is a two-minute preview of the report-building craft this series proper starts in Part 006. Check: the visual renders numbers without errors.
   

   Step 6: the two-minute preview — Department by Budget Amount, rendered straight from the new semantic model before any report is saved.

Import vs Connect: Local File, OneDrive & SharePoint — Choosing the Route

The same workbook can enter Power BI through three doors, and they are not equivalent — the difference is where the copy lives and who keeps it current. Microsoft's Get data from files guide draws the line exactly; here it is as a decision table:

The OneDrive Sync Advantage: Hourly, Automatic Refresh

For the budget workbook, the choice is clear-cut. Save it to OneDrive for work or school — signed in with the same account you use for Power BI — import it from there, and Power BI connects to the file about once an hour. Change a number in Excel, and within the hour the semantic model, any reports, and any dashboards built on it update themselves. No scheduled refresh to configure, no re-import ritual; the file in the cloud is the source, and the model is its live projection. Note the boundary conditions Microsoft enforces: personal OneDrive accounts are not supported, and once a model is connected to OneDrive you can no longer update it from local files — the file in OneDrive is now the single source.

Three doors, three behaviors: snapshot import, hourly cloud sync, and file upload without a model — pick the route that matches how the workbook actually lives.

Personal OneDrive vs. OneDrive for Work or School: Why Account Type Dictates Sync

One of the most persistent frustrations among professionals learning Power BI on their own is setting up a cloud connection only to find that edits in Excel never appear in the dashboard. The culprit is almost always an account boundary: confusing personal OneDrive with OneDrive for work or school.

The distinction is structural, not cosmetic. A Personal Microsoft account (tied to consumer domains like @outlook.com, @hotmail.com, or personal Microsoft registrations) connects to consumer OneDrive storage. A Work or School account is governed by an organizational tenant managed through Microsoft Entra ID (formerly Azure Active Directory). Power BI Service is an enterprise cloud platform built entirely on Entra ID security boundaries.

When you import an Excel workbook from OneDrive and expect automated hourly synchronization, Power BI establishes an OAuth 2.0 delegation token and registers a background Microsoft Graph webhook against the cloud file. According to Microsoft's official Excel connection documentation, this automated background handshake is supported exclusively through OneDrive for work or school and SharePoint Document Libraries. Personal OneDrive does not support this synchronization protocol. If you upload from a personal OneDrive, Power BI falls back to an unlinked snapshot: the file is ingested once, but subsequent edits remain stranded on your personal drive.

To ensure your automated budget sync functions reliably, verify these three pre-flight conditions:

• Identity Alignment: The email account you sign into app.powerbi.com with must belong to the same organizational Microsoft 365 tenant where your OneDrive workbook is stored.
• URL Verification: When opening your workbook in Excel Online, examine the browser URL. A work or school repository will display an enterprise tenant address (such as https://yourcompany-my.sharepoint.com), whereas personal storage points to onedrive.live.com.
• Team Collaboration Best Practice: If multiple colleagues update actual expenditures or adjust budget allocations, do not store the workbook in an individual employee's personal OneDrive for business. Save it in a SharePoint Team Site Document Library. This guarantees that background model refreshes persist even if an individual team member changes roles or leaves the organization.

Account boundary check: automated hourly sync requires OneDrive for work or school or SharePoint, where Entra ID maintains cloud credentials.

What Changed in 2026: The Legacy Import Experience Is Retired

If you imported Excel files into Power BI before, some of what you read today contradicts your memory — and the documentation is right, your memory is not. Microsoft deprecated two things:

1. Local workbook upload with refresh. Uploading local Excel workbooks to workspaces and configuring refresh for them is no longer supported. Microsoft's migration guidance in the local Excel workbook refresh documentation is explicit: download existing local workbooks from your workspace and import them again, publish the Excel data model as a Power BI semantic model and schedule refresh, or import the workbooks from OneDrive and SharePoint Document libraries.

2. The legacy Excel/CSV import experience. The old import flow from the service's create page is retired. Microsoft has set hard dates: semantic models created with it stop refreshing after July 31, 2026 and stop loading after August 31, 2026. These deadlines are confirmed in Microsoft's data sources for the Power BI service documentation and their corresponding Microsoft Fabric release notes. Excel and CSV remain fully supported as data sources — only the old import experience is gone. If you hold models built that way, migrate them to the current flow or rebuild them in Power BI Desktop before those dates pass.

Your Three Migration Paths, Ranked

For our budget case — and for any workbook with a data model built with Power Query or Power Pivot — the ranking is straightforward. Microsoft recommends Power BI Desktop for models with data models, because the import upgrades the model to the latest version; the service-based paths are the lightweight alternatives:

Publishing from Excel: Upload vs Export, and Its 2026 Caveat

One more door exists, from inside Excel itself: File > Publish > Power BI (Excel 2016 and later), as documented in Microsoft's Publish to Power BI from Microsoft Excel guide. If you pick Upload, the workbook appears in the service like Excel Online — no semantic model is created, so nothing refreshes into the platform. If you pick Export, the tables and data model export into a new semantic model that stays synced with the workbook on OneDrive — the same hourly sync behavior as the import route. But heed the caveat in that documentation: Publish to Power BI is deprecated in Excel in Microsoft 365 and is not included in Excel 2024 perpetual. Treat it as a legacy bridge, not the future; the New item flow and Power BI Desktop are.

Where the Data Actually Goes: OneLake, Workspaces & the 1 GB Limit

Press the import button and a reasonable question follows: where did my data just go? The answer has two layers — the container it lives in, and the platform-wide lake it lands in.

OneLake: One Lake, One Copy of the Truth

Every Power BI semantic model is physically stored in OneLake — the tenant-wide, governed data lake at the heart of Microsoft Fabric. When you import the budget workbook, the tables, the data model, and every report built on them live in OneLake, under your workspace; your local file, meanwhile, simply stays where it was. If the platform-wide data lake concept is new, our primer on cloud databases across AWS, Azure, and Google Cloud maps the broader territory of where cloud data can physically live.

The Practical Limits: 1 GB Models, 16,000 Columns, Honest Expectations

Import mode is a snapshot in memory, and memory has edges. On shared capacity — which is where the free license runs — semantic models are limited to 1 GB; importing a workbook larger than that fails outright. A departmental budget with tens of thousands of rows sits comfortably inside the line; a decade of transaction-level sales data may not. For orientation at scale: import models can cache roughly 2 billion distinct values per column, with a platform ceiling of 16,000 columns across all tables — limits you will realistically never touch at spreadsheet scale, but good to know before Stage 5's aggregations conversation.

The journey of your budget data: workbook on disk or OneDrive, import flow, semantic model item stored in OneLake inside My workspace.

Troubleshooting the Import: Six Failures and Their Fixes

I have administered reporting platforms long enough to know that the first import rarely works on the first try — in my PACS/RIS world, ingestion errors were always structural, never mystical. Power BI behaves the same. When an import disappoints, it is one of a small set of structural causes; here is the diagnostic table I wish I had on day one:

The Budget-vs-Actual Workbook Becomes the Series' Living Model

Take stock of what now exists. The budget workbook — tblBudget and tblActuals — has crossed from desktop to cloud and now lives as a semantic model in My workspace, on the free license, stored in OneLake. Every artifact for the next 46 parts grows from this one item, and its location is a deliberate series decision:

My workspace, not a shared workspace. Part 002 established why: on the free license, My workspace is the container you can create in, and content there belongs to you personally. When Stage 2 brings teammates in, the model graduates to a shared workspace — a planned move, not an emergency one. Just as a small e-commerce team often starts tracking inventory in a simple database before graduating to a full system, our model starts in the simplest correct container and grows with its audience. If your team already uses automation tools to maintain budget workbooks — macros, scheduled scripts, or cloud functions — our walkthrough of Google Apps Script automation and our primer on API fundamentals map the broader automation landscape that feeds into Power BI's REST API capabilities later in Stage 8.

OneDrive, not local disk. Because the workbook is maintained monthly, it belongs in OneDrive for work or school with the import connected to it — the hourly sync keeps Power BI current with zero refresh administration. The local snapshot path remains available for one-time archives.

Tables and types, not formats. The workbook's charts and summary formulas stayed behind on the worksheets. Part 004 wires this model into Q&A so you can type "total budget by department" and get an answer; Part 006 rebuilds the visuals properly; Stage 3 adds DAX measures like Budget Variance = SUM(tblActuals[Actual Amount]) - SUM(tblBudget[Budget Amount]) and a dedicated date table. The import was never the destination — it is the foundation pour.

Quick Knowledge Check & Practical Challenge

Before the FAQ, two exercises to convert this article from reading material into muscle memory. The first tests your route-choice logic; the second finishes the actual Part 003 build.

Frequently Asked Questions

The Bottom Line: One Workbook, One Route, One Model

Upload Excel to Power BI and you did something quietly enormous: you converted a file people email each other into a semantic model a platform can reason over. You now know that data must arrive as tables or a data model; that import-from-OneDrive gives you hourly automatic refresh while local imports are snapshots; that the legacy import experience is retired with hard cut-off dates in July and August 2026; and that your model physically lives in OneLake, inside My workspace, on the free license.

Your move: run the pre-flight, import the workbook via New item > Semantic model > Excel, and open Explore data to watch your budget fields load. Then bring that model to Part 004: Power BI Q&A, where you will type questions in plain English and the model answers them. The import you refuse to rush today is the model that never embarrasses you tomorrow.