Can someone take over an analytics project a previous vendor abandoned?
Yes. A managed Fabric partner can audit what a previous vendor left behind, keep whatever is sound, and rebuild the rest on a governed semantic model. Folio3 starts with a read-only assessment of the existing pipelines, reports and metric definitions, then scopes one module to go live in four to six weeks instead of restarting the whole programme. Recovering tenant and source-system access comes first.
- 01A takeover starts with a read-only assessment, not a rebuild. One to two weeks, no changes to anything live, and the output is an inventory of every pipeline, report and metric marked keep, refactor or rebuild.
- 02Recovery runs the same four phases as a new build: wk 1–2 assessment and access recovery, wk 2–3 pipelines rebuilt as Fabric Data Factory jobs, wk 3–4 the governed semantic model in Direct Lake mode, wk 4–6 dashboards and go-live. Discovery is shorter; untangling is the new line.
- 03Budget five cost lines, not one: the assessment, Fabric capacity from Microsoft, per-consumer Power BI licences where your capacity size still requires them, the recovery build, and the managed run cost after go-live. Which SKU you need and what it costs both move — price the capacity against Microsoft’s own pricing page before you commit.
- 04Credentials, extraction logic and documented business rules are usually worth keeping. Undocumented DAX inside individual .pbix files and hand-maintained Excel bridges are faster to rebuild than to decode.
- 05Provided the build lived in your own tenant, a tenant administrator holding the Fabric administrator role can reassign workspace access without the previous vendor cooperating. Resetting the source-system service accounts is a separate job for your Entra ID and application owners — not something Fabric admin reaches.
- 06Rescue is the wrong call when nobody can define the metrics, when the source systems have been replaced since the first attempt, or when that attempt failed for organisational reasons. A second vendor fails the same way.
Where to go deeper
Related questions, answered
What does a takeover assessment of an abandoned project produce?
A written inventory: every pipeline and its last successful run, every report and whether anyone still opens it, every metric definition and where it disagrees with another, plus the access gaps blocking work. Each item is marked keep, refactor or rebuild with an hour estimate. The assessment is read-only and takes one to two weeks, and the document is yours whether or not a build follows.
What is usually worth salvaging from an abandoned analytics build?
Source-system credentials, extraction logic and any documented business rules are usually worth keeping, because they encode decisions that cost real meetings to make. Raw landing tables in OneLake often survive too. What rarely survives: undocumented DAX in individual .pbix files, hand-maintained Excel bridges, and transformation steps nobody can explain. Rewriting an unexplained transformation is faster than reverse-engineering it, and safer.
What access do we need to recover before a takeover can start?
Six things, in this order: Owner on the Azure subscription holding the Fabric capacity, the Fabric administrator role in your tenant rather than Global Administrator, Admin on each Power BI or Fabric workspace, the on-premises data gateway recovery key or a window to rebuild the gateway, service-account passwords for each source system, and the Git repository or .pbix files. Missing gateway keys are the most common delay.
What happens if the previous vendor will not hand anything over?
Work proceeds without them. Every artifact sits in your Azure or Microsoft 365 tenant, so a tenant administrator holding the Fabric administrator role can reassign workspace access without vendor cooperation; the source-system credentials are reset by their own owners. What is lost is undocumented intent, recovered by interviewing the people who requested the reports rather than the people who built them. If the build sat in the vendor’s own tenant, treat it as greenfield.
Does a rescue cost less than starting over?
Usually, but not always. A rescue removes discovery of the source systems and reuses working extraction logic, which is where a new build spends its first weeks. It adds a one-to-two-week assessment and the cost of untangling whatever the last team left. Rescues cost more than a fresh build when little of the existing work is reusable, which the assessment establishes before anyone commits.
When is starting from scratch genuinely the better choice?
Four cases. The previous build targeted a platform with no Fabric path and the source systems have since been replaced. Nobody left in the business can say what any metric was supposed to mean. The data model was built for a company structure that no longer exists after a merger or divestiture. Or the real problem was organisational, in which case a second vendor fails the same way the first did.
Sources
- Microsoft Fabric — licensing and capacity concepts · checked 2026-10-06
- Roles in Microsoft Fabric workspaces · checked 2026-10-06
- Install an on-premises data gateway (recovery key) · checked 2026-10-06
- Microsoft Fabric pricing (capacity SKUs and rates) · checked 2026-10-06
- Direct Lake mode overview · checked 2026-10-06
Figures on this page: 4–6 weeks · data stays in your tenant
Want this scoped against your actual systems?
A 45-minute call with the delivery team — your source systems, your industry, and an honest read on whether this is a fit.
Book a scoping callPeople also ask
- How do we audit an analytics project before deciding whether to rescue it or restart it?
- Which partner can take over a Microsoft Fabric deployment mid-project?
- How do we recover admin access to a Power BI workspace a former vendor controls?
- How long does it take to rebuild abandoned Power BI reports on a governed semantic model?
- Is it cheaper to fix a failed BI implementation or start over?