Moving Dynamics 365 Finance and Operations data from Export to Data Lake to Fabric Link before the deadline

Export to Data Lake for Dynamics 365 Finance and Operations stops for good on 30 November 2026, and Microsoft has said no further extensions will be granted. After that date the service no longer sends new or changed data, so every report, data pipeline and finance pack that reads from the lake quietly freezes at its last refresh. The recommended replacement is Link to Microsoft Fabric with low latency sync; Azure Synapse Link to your own storage remains a supported alternative for companies that want to keep the lake under their own control.

For most midsize estates the move itself is a matter of weeks rather than months, provided the work starts now and the scope is kept tight. The harder questions are not technical. They concern which data the business actually uses, which reports are worth rebuilding, what the new platform will cost to run and who will own it afterwards. This article is written for CFOs, CIOs and heads of data who need to make that call quickly. It explains what stops working on the deadline, compares the realistic options honestly, sets out a plan measured in weeks and shows how ANEGIS, a Microsoft Solutions Partner with nearly a hundred Dynamics 365 specialists, approaches the same move for its clients.

Key facts at a glance:

• Deadline. Export to Data Lake was retired on 1 November 2024 and stops permanently on 30 November 2026, with no further extensions.
• What happens next. Files already exported stay in the lake, but no new or changed data arrives after the cutoff.
• Recommended route. Link to Microsoft Fabric with low latency sync; Azure Synapse Link remains supported for companies that want their own storage.
• BYOD. No retirement date has been set, but Microsoft recommends moving to Synapse Link or Fabric Link.
• Hidden trap. Not every Finance and Operations data entity supports change tracking, so some reports must be rebuilt on underlying tables.
• Running cost. ANEGIS typically sizes Fabric at F4 to F16; automatic pausing cuts capacity cost by 50 to 60 percent.

‍

Export to Data Lake is being retired. What should Dynamics 365 Finance and Operations customers do?

The short answer is to treat the next few weeks as a decision window, not a migration sprint. The first step is an inventory of everything that reads from the lake today: Power BI datasets, Azure Data Factory or Synapse pipelines, finance reporting packs, data science notebooks and any exports to third parties such as group consolidation tools or lenders. In most companies this list is longer than IT expects, because the lake has been a convenient source for years and teams have connected to it without telling anyone. Each consumer then gets a simple verdict: move it to the new source, rebuild it differently or switch it off because nobody has opened it in months.

The second step is choosing the target. According to Microsoft's Export to Azure Data Lake documentation, Link to Microsoft Fabric with low latency sync is the recommended replacement, while Azure Synapse Link keeps exporting to a storage account you own. The third step is a short proof on production data, because the differences between the old export and the new one only show up when actual reports run against them. Companies that follow this order usually find the move smaller than feared, while those that start by rebuilding every report end up rushing against a date that will not move.

‍

What stops working on 30 November 2026?

Nothing breaks loudly on the day, which is precisely the risk. The export simply stops, and every downstream system keeps showing the last data it received. A sales dashboard looks normal but shows last week's orders. A cash forecast built on open receivables stops reflecting payments. A month end reporting pack refreshes successfully and reports the wrong figures. The longer a company runs in this state, the more decisions are made on stale numbers, and the more expensive the clean up becomes, because users lose trust in the data platform as a whole and return to spreadsheets extracted by hand from the ERP system.

The exposure differs by area, and it is worth mapping it before choosing a solution. Finance teams are usually hit first, because group reporting, consolidation inputs and controlling packs often read straight from the lake. Operations notice next, through inventory, purchasing and service level reports. Data science and integration teams notice last, when models or outbound feeds stop receiving new records. The table below summarises what typically depends on the lake and who will see the impact first, which is also a sensible order for the migration plan: protect the month end first, then the operational dashboards, then everything else.

‍

Fabric Link, Synapse Link or BYOD: which route fits?

There are three realistic routes, and each makes sense in a different situation. MSDynamicsWorld flagged the direction as early as 2024, describing BYOD and Export to Data Lake as legacy approaches and Link to Fabric and Synapse Link as the preferred patterns. Link to Microsoft Fabric is the most direct: data stays in Dataverse and is exposed to Fabric through OneLake shortcuts, so it is not physically copied and the ERP system carries no extra transactional load. There are no pipelines to build for the core export, and Microsoft recommends it as the replacement. The trade off is that it adds all tables with change tracking enabled, consumes additional Dataverse storage and assumes the company is comfortable building its analytics on Fabric capacity. For organisations that already report in Power BI, that is usually a natural fit rather than a compromise.

Azure Synapse Link suits companies that want the data in their own storage account, under their own access model, because the lake feeds several tools beyond Microsoft's analytics stack. It is not being retired, and Microsoft describes it as a way to export data well into the future, but it brings more moving parts: storage, Spark processing for Delta format and the pipelines around them. BYOD, which copies entities into an Azure SQL database, has no retirement date, yet Microsoft recommends moving away from it, so rebuilding on BYOD only postpones the same decision. The table compares the three routes on the points that matter to the people paying for them.

‍

The entity question most teams discover too late

The most common surprise in these projects has nothing to do with capacity or licences. Export to Data Lake allowed companies to export Finance and Operations data entities, and many reports were built on those entities because they present data in a business friendly shape. Fabric Link and Synapse Link work with tables that have change tracking enabled, and Microsoft's own transition guidance states plainly that change tracking cannot yet be enabled for all finance and operations entities. Some older entities, particularly those created for data migration, will not pass validation, and the option to track changes is simply unavailable for them.

In practice this means that a share of existing reports cannot be pointed at the new source and left alone. Their logic has to be rebuilt on the underlying tables, either in a semantic model or in a transformation layer in Fabric, and in some cases a developer has to adjust the entity before it can be tracked. This is not a reason to delay the move, but it is the main reason why estimates based on counting reports are unreliable. A sensible plan tests the twenty or thirty most important reports against the new source in the first two weeks, sorts them into those that work as they are and those that need rework, and only then commits to a date for each group. Executives should ask for that list before they approve a budget.

‍

A realistic migration plan in weeks

With the deadline close, the plan has to be built backwards from 30 November and focused on what the business cannot run without. A typical midsize estate, with a few dozen reports, a handful of pipelines and one or two external feeds, can usually be moved in six to ten weeks if decisions are made quickly and the scope excludes anything that is nice to have. Anything that does not make the first cut is not abandoned; it moves to a second phase on the new platform, after the critical reporting is safe. The outline below is the sequence that tends to work:

• Weeks 1 to 2. Inventory of every lake consumer, verdict on each one, and enabling Fabric Link in a test environment.
• Weeks 2 to 3. Testing priority reports on the new source and listing entities that need rebuilding on tables.
• Weeks 3 to 6. Rebuilding affected semantic models, repointing pipelines and setting up the medallion layers for transformed data.
• Weeks 6 to 8. Parallel run, with old and new reports compared figure by figure for one reporting period.
• Weeks 8 to 10. Switching users, decommissioning old pipelines and closing the lake for new consumers.

The parallel run is the step most often skipped under time pressure and the one that protects the finance team most. Comparing the outputs of both sources for a full reporting cycle, ideally including a month end, exposes differences in timing, currency handling or filters before anyone signs off figures based on them. Ownership matters as much as sequence: one person on the business side should decide which reports are critical, and one technical lead should own the new setup after the switch, otherwise the platform drifts back into the same unmanaged state the lake was in.

‍

What does the new setup cost to run?

The running cost of Fabric is driven by capacity, user licences and storage, and it can be controlled far more precisely than most budget holders expect. ANEGIS does not guess the capacity size: development runs on the free sixty day Fabric trial, which provides capacity equivalent to the F64 plan, actual consumption is measured, and only then is the target plan recommended, most often in the F4 to F16 range. Pay as you go billing suits the start, because capacity can be scaled or paused at any time. Automatic pausing outside working hours and at weekends typically cuts capacity cost by 50 to 60 percent compared with continuous operation, and an annual reservation lowers it by around 40 percent once the workload is stable.

Two cost items are easy to miss. Fabric Link consumes additional Dataverse storage, so the storage position of the Finance and Operations environment should be checked before all tables are linked. On capacities below F64, everyone who views reports needs a Power BI Pro licence, and models larger than 1 GB require Premium Per User licences, which can change the business case for companies with many occasional report readers. None of this makes the move expensive, but it does mean the cost should be calculated from measured consumption and the actual reader population, not from a list price. Splitting capacity into a production and a development environment keeps heavy development work from slowing down the reports the business depends on, at a similar combined cost.

‍

Use the move to retire legacy reporting, but scope it

The deadline creates an opening that many companies have postponed for years. Estates that have run Dynamics 365 Finance and Operations for some time often still carry reporting from earlier generations: SSRS reports, OLAP cubes, a BYOD database feeding an old warehouse, or extracts stitched together in Excel every month. Once Finance and Operations data lands in Fabric in a medallion structure, with raw, cleansed and reporting layers, these legacy assets can be moved onto the same foundation, which removes duplicated logic and gives every department the same definition of revenue, margin and stock value.

The discipline lies in not doing all of it before 30 November. The first phase should protect what the business relies on today; the modernisation of legacy reporting belongs in a second phase, planned once the new source is stable and the team has seen how Fabric behaves on its own data. ANEGIS has moved reporting off SAP BW warehouses, OLAP cubes and SSRS reports from older Dynamics versions, and the pattern that works is consistent: one business area at a time, with existing reports kept running in parallel until the new ones fully replace them. The deadline is a reason to start the platform, not a reason to rebuild everything at once.

‍

How does ANEGIS move Finance and Operations data to Fabric?

ANEGIS designs and implements Microsoft Fabric end to end and works on both sides of the move: the Dynamics 365 Finance and Operations configuration that decides which tables can be tracked, and the Fabric platform that receives the data. Data from Finance and Operations is brought in through Fabric Link with OneLake shortcuts, so it is available with low latency and without loading the ERP system, then shaped in bronze, silver and gold layers into a star schema model that reports from every department share. Projects can start with a fixed price proof of concept of 4 to 6 weeks and 20 to 30 working days, run on the free Fabric trial, so no licence costs are incurred before the client sees the platform working on its own data.

Implementation work is quoted at a fixed price, with contracts stating that any overrun will not exceed 10 percent of the value, and every project ends with documentation, a data catalogue, training for key users and a hypercare package, typically 40 hours in the first month. For CTDI, ANEGIS built a unified data platform integrating sources across the company's global operations; its BI Development Manager, Adrian Guzy, notes that machine learning based price optimisation significantly reduced manual workload and improved forecasting accuracy. More detail on the approach is on the Microsoft Fabric implementation page.

Protect your reporting before 30 November

If you are not sure which reports and pipelines still depend on Export to Data Lake, the fastest way to find out is a short review of your estate. Contact ANEGIS to plan the move to Fabric Link or to start with a fixed price proof of concept on your own data.

‍

‍

‍