How do you migrate from Excel to an asset management platform?

01

Why Excel breaks at portfolio scale

The failure is rarely one big break. It is the slow accumulation of edge cases, one analyst has a slightly different capex schedule, one tab has a hard-coded exit cap, one lease abstract sits on someone's laptop. At small scale the founder holds the whole picture in their head. Past 15 to 25 assets, or past three concurrent contributors, that picture stops fitting. The spreadsheet still runs, but no one fully trusts it, and the review layer eats more hours than the underlying work.

02

Signs it is time to migrate

Most operators wait too long. The migration cost rises with every asset added, every quarter of history that has to be lifted, and every process that hardens around the spreadsheet. A useful test is the headcount test, if the team has hired an analyst whose primary job is to keep Excel running, the platform would already be cheaper. A second test is the LP test, if an incoming institutional investor would require reporting the current setup cannot produce, the timeline is set by the raise, not the team.

03

The four migration paths

Yardi and MRI are the default for institutional portfolios and carry the deepest functionality plus the highest configuration burden. Mid-market platforms suit portfolios under $500M with standard asset classes and thin AM teams. Custom platforms suit operators with unusual structures, multi-strategy portfolios, or a strong preference for a data warehouse pattern, they cost more upfront and deliver more flexibility long-term. Hybrid setups are common at the $500M to $2B range, where core ledger stays packaged and analytics move custom.

04

Data audit: what you have vs what you need

The audit is boring and non-negotiable. It typically runs 2 to 4 weeks and produces a data catalogue with source, owner, refresh cadence, and current quality score per field. The catalogue exposes the fields that live only in someone's head, the ones that live in a PDF, and the ones that exist in three versions across three tabs. Skipping the audit is the single most reliable way to make the migration overrun by 3 to 6 months.

05

Building the data model before migration

Model design is where most in-house teams under-invest. The right pattern is to sketch the target model on paper, validate it against every reporting output the fund produces (LP letter, board pack, tax pack, GRESB submission), and only then start the platform configuration. Time-series conventions matter, especially, whether NOI is stored at unit level or asset level, whether valuations are stored gross or net, whether debt is stored per tranche or per facility. Every one of these decisions has downstream consequences that are painful to unwind.

06

Historical data migration and reconciliation

Reconciliation is the step that catches years of accumulated spreadsheet drift. Small errors compound, a lease abstract that used the wrong start date, a capex line coded to the wrong asset, a valuation footnote that never made it into the tab. The reconciliation exercise typically surfaces 50 to 200 exceptions per $100M of portfolio and forces the team to decide, per exception, whether to correct in the platform or restate the history. A clean reconciliation is what makes the LP conversation defensible after cutover.

07

Running Excel and platform in parallel

The parallel period is where nerves fray. The team is doing 1.6 to 1.8 times the normal work, the platform is throwing new questions, and the temptation to shortcut the parallel is high. Discipline matters, every variance above a defined threshold (often 0.5% at asset level, 0.1% at portfolio level) gets logged, root-caused, and either fixed in the platform or accepted as a known difference. At the end of the parallel period the team should have a written sign-off from finance, AM, and IR before cutover.

08

Cutover playbook: the 30-day plan

A defensible cutover has three named owners, a data owner, a process owner, and a change owner, and a written playbook with every step timed. Typical elements include a data freeze 3 to 5 business days before go-live, a final reconciliation the day before, a formal go/no-go call, a documented rollback plan, and a two-week hypercare window where the vendor or implementation partner is on call. The first close on the platform is expected to take 1.5 to 2 times the eventual steady-state time.

09

Training the team and change management

The change side is under-invested more often than the technical side. Analysts who built the spreadsheet often feel their value was in the model, and the migration removes that value if the transition is not handled well. Redirect the role toward higher-value analysis, scenario modelling, portfolio construction, LP-facing analytics, and the team retention holds. Skip that step and the best analysts leave within 6 to 12 months, taking the institutional memory with them.

10

Common migration mistakes and how to avoid them

The recovery cost of each mistake is asymmetric. A weak audit adds 2 to 3 months. A rushed model design produces a rebuild inside 2 years. A short parallel period produces a bad first close and an LP escalation. A quarter-end cutover produces a missed reporting deadline. Weak change management produces attrition. The fix in every case is the same, slow down at the design stage, budget for the parallel, and put a named change owner on the project from week one.