Lift and shift copies the problems. Rebuild the model once, rebuild the used reports on it, and retire the rest. The rule, the exceptions, and the risks.
When migrating BI platforms you should rebuild the data model once, then rebuild the reports people actually use on top of it, and retire the rest. Lift and shift, meaning converting workbooks or apps as they stand into the new tool, is the right choice only for a small number of simple, standalone reports with a short remaining life. The reason is that the value of a report lives in its model and its definitions, not its layout, and the old platform's model is almost always the thing that was causing the problems that prompted the migration. This post sets out the rule, the three cases where lift and shift is fine, the risks of each approach, and how to avoid losing reports and logic along the way.
Why lift and shift fails
A Tableau workbook, a Qlik app or a legacy SSRS report carries three things: a data model (often a blend built inside the report), a set of calculations, and a layout. Conversion tools move the layout well, the calculations badly and the model not at all, because Power BI's engine wants a star schema and evaluates calculations in a different context. The result is a set of reports that look familiar, run slowly, disagree with each other where the old ones did, and need rework the moment anyone asks a new question. You pay for the conversion and then pay again for the rebuild, with a period in between where nobody trusts the numbers. Qlik to Power BI: lessons learned the hard way is the long version of that experience.
Why rebuild works
Rebuilding starts with the model: a star schema over the sources, with measures defined once and signed off by finance, and row-level security designed for the new platform. Once that exists, each used report is a day or two of work, because the hard part is done and shared. The rebuilt estate is smaller, faster and consistent, and the migration has fixed the problems that caused it rather than carrying them across. The cost is that the model has to be built before the first report appears, which needs a sponsor who can hold their nerve for a few weeks. Star schema explained describes the target.
The three cases where lift and shift is fine
First, simple standalone reports with a single clean source and no business logic, such as an operational list or an export, where conversion is faster than rebuilding and the result is good enough. Second, reports with a short remaining life, where a system change in six months will retire them anyway and a converted copy bridges the gap. Third, a hard deadline, typically a licence expiry, where converting the critical reports buys time for the rebuild behind them, provided the rebuild is funded and scheduled rather than hoped for. In each case the converted reports are flagged as temporary and given a retirement date.
How to decide, report by report
Start with usage logs and the report owners. Every report goes into one of four lists: rebuild on the new model, convert and retire later, retire now, or hand to its owner to rebuild themselves once the model exists. In most estates the fourth list is the largest and the first list is a few dozen reports, which is what makes the project tractable. The test for the first list is whether finance, sales or operations would notice within a week if the report disappeared. Fewer dashboards, more decisions sets out that test.
The risks, and how to avoid losing logic
The risk in rebuilding is losing a business rule that was buried in a calculated field or a load script and that nobody documented. The mitigation is to extract every calculation from the old platform into a definitions document before building anything, reconcile each rebuilt measure to the old one on a known period, and keep the old platform running read-only until the first month end has closed on the new one. The risk in lift and shift is the opposite: carrying every rule across, including the wrong ones, and never finding out which were wrong. How long does a BI migration take, and how do you keep reporting running? covers the parallel running that makes either approach safe.
What it means for the budget
Rebuilding looks more expensive on day one and is cheaper by the end of year one, because the converted estate needs a second project. If a proposal prices the migration by workbook count, it is pricing a conversion. If it prices a model, a prioritised report list and a retirement plan, it is pricing a rebuild. How much does it cost to migrate from Tableau or Qlik to Power BI? sets out the drivers, and our free audit will tell you which list each of your reports belongs in.
Related reading
How much does it cost to migrate from Tableau or Qlik to Power BI?
How do you migrate from QlikView to Power BI? Lessons learned the hard way
Shauna Duffy
Director of Professional Services
Part of the Hopton Analytics team, delivering governed analytics programmes for UK mid-market organisations.



