Import and DirectQuery are the two main storage modes in Power BI. Import mode, the default, loads a copy of your data into the report and holds it in memory for fast queries. DirectQuery loads no copy: it queries the source data live each time the report is used.
What a storage mode is
A storage mode is a setting on each table in a Power BI model that decides two things: where that table's data lives, and where the work happens when a report runs a query. You pick it when you connect to a data source in Power BI Desktop.
There are two modes you'll meet in everyday work: Import and DirectQuery. They are not better or worse than each other; they're a trade-off between speed and how current the data is. Microsoft Fabric adds a third mode, Direct Lake, but for most reports the choice is Import or DirectQuery.
How Import mode works
Import is the default mode and the one most reports use. When the report is refreshed, Power BI takes a copy of the data, compresses it and loads it into the semantic model, where it is held in memory. From then on, every visual queries that fast in-memory copy rather than the original source.
That design makes Import quick and capable. Visuals render fast because they query the in-memory copy. The full Power BI and DAX feature set works without restriction, and the report keeps working even if the source is offline between refreshes. The trade-off is currency: the data is only as fresh as the last refresh, and the model has a size limit that depends on your Power BI licence.
How DirectQuery mode works
DirectQuery takes the opposite approach. No copy of the data is loaded into the model: the data stays in the source. Each time someone opens the report or changes a filter, Power BI sends a live query back to the source database, which does the work and returns the result.
That keeps the report current and removes the copy-in size limit, so DirectQuery suits very large datasets and cases where the data must stay put. Microsoft's guide to using DirectQuery in Power BI Desktop is clear about the cost: it is slower than Import, some DAX functions and features are restricted, and the source must be online and fast, because the report's speed is the source's speed.
If a DirectQuery report points at a source on your own network rather than in the cloud, it also needs the Power BI gateway to reach that source. Scheduled refresh of an Import-mode model from on-premises data needs the same gateway.
Import vs DirectQuery: side by side
The two modes differ on five things that matter when you build a report:
| Import mode | DirectQuery mode | |
|---|---|---|
| Where the data lives | A compressed copy is held in the model, in memory | No copy: the data stays in the source |
| Report speed | Fast: visuals query the in-memory copy | Slower: every visual waits on the source |
| How current the data is | As fresh as the last scheduled refresh | Live: always the source's current data |
| Dataset size | Limited by your Power BI licence | No copy-in limit; suits very large data |
| Features and DAX | The full Power BI and DAX feature set | Some functions and features are restricted |
One line sums it up: Import trades currency for speed and capability; DirectQuery trades speed and capability for currency. SQLBI's explainer on Import vs DirectQuery in Power BI reaches the same conclusion, and SQLBI is the most authoritative independent voice on Power BI modelling.
Which one should you use?
For most reports, the answer is Import. It's the default for good reason: faster, fully featured and not dependent on the source staying online. Unless you have a specific reason to do otherwise, build in Import mode.
Choose DirectQuery when you genuinely need it. There are three honest reasons:
- The report must show real-time data and a scheduled refresh is not current enough.
- The dataset is too large to import within your licence's model size limit.
- A data policy requires the data to stay in the source rather than be copied.
A model can also mix the two, with some tables Import and some DirectQuery, which Power BI calls a composite model.
Choosing a storage mode is a real Power BI judgement call Picking Import or DirectQuery, and knowing when a composite model is the right answer, is one of the decisions that shapes whether a report stays fast as it grows. Our two-day, hands-on Power BI Masterclass teaches modelling and storage modes the practical way, on real data. If you are weighing up where to learn, our Power BI training buyer's guide compares the UK options.
