Spreadsheet or Family Office Software? Follow One Number to Decide

A spreadsheet can be a capable family-office tool. It can also become a fragile unofficial system of record when a family needs several people to change the same holdings, entities and reports without losing the source or approval trail. Dedicated software may reduce that work, but it does not make imported data correct by itself.

The practical comparison is what happens after a number changes. This guide follows one fictional fund valuation through a controlled workbook, a reporting layer and an integrated system. It shows what to keep, what to test and when a migration has earned its cost.

01What the tools actually own

The key distinction is where the authoritative record lives. A spreadsheet might be the approved investment schedule while an external general ledger remains the entity book. A reporting platform might assemble views from both without becoming either record. An integrated platform might be configured to maintain one or more ledgers, but its contract and implementation must establish which ones.

Do not compare “Excel” with “software” as if one has no controls and the other always has them. Microsoft documents Excel coauthoring, workbook access restrictions and SharePoint versioning. Those features depend on storage, licensing and configuration. The office still has to define which file is approved, who may change formulas, how a source document is attached and how a released family report can be reproduced.

Ask four ownership questions before reviewing a vendor demo:

• Which system or person owns the source valuation and its effective date?
• Which record owns the entity's accounting balance, and what book basis does it use?
• Which rule produces the family ownership view, including look-through and eliminations?
• Who approves the report release, and can the prior version still be retrieved?

The answers can span several tools. Buying a dashboard without naming those owners leaves the existing control gap in place.

02One correction through three operating models

Assume an SPV holds a private-fund interest. Its approved family report carried US$8.00 million for 30 September. A revised, authorized manager statement for the same holding and effective date supports US$8.15 million. Trust A owns 60% of the SPV; Trust B owns 40%. In this simplified, wholly owned family structure, the SPV's reporting value changes by US$150,000. Trust A's look-through view changes from US$4.80 million to US$4.89 million, and Trust B's from US$3.20 million to US$3.26 million. The two trust increases sum to US$150,000; they do not add a second US$150,000 to the family total.

This is a fictional reporting-value correction, not an automatic accounting entry. The office must check the manager's document, valuation policy, entity book basis and approval authority before deciding whether a journal is required. A statement with a different valuation date would be a different event, not proof that the 30 September report was wrong.

Use the same source, date, ownership percentages and expected outputs for each route. A pass means the office can show who changed what, which document authorized it, whether accounting changed, who reviewed it, and which report was released. If the product cannot demonstrate a function, mark it unknown until the vendor and implementer prove it.

03Where the spreadsheet can still win

A controlled workbook is rational when the office can reconcile it on schedule, keep named ownership of inputs and formulas, restrict editing, preserve versions and explain every material difference. It may be particularly useful for a focused scenario model or a small, stable schedule that draws from an established accounting system. The case for keeping it is an observed control result, not a universal entity-count threshold.

Test the current process using three varied historical cases if they exist: a source-value correction, an ownership-map or formula change, and reconstruction of a previously released report. For each, retrieve the source, old and new values, effective date, formula change if any, edit authorization, review evidence, exported report and time spent. If the office lacks one case type, record that coverage gap; three examples are a diagnostic sample, not certification.

If all available cases can be reconstructed without undocumented manual intervention, the immediate answer may be tighten the workbook controls. If one cannot, hold the affected report and repair that control gap; it does not automatically prove a new platform is the right remedy.

Name a master file, separate inputs from formulas, protect calculations, keep a change log, review reconciliations and archive each approved output. A shared workbook's version history helps, but a prior file version alone does not explain why a valuation was approved.

The cost side matters. Keeping Excel is not free if two analysts spend days re-keying and reconciling. Yet software may add implementation, subscription, data-cleaning and training costs. Measure current labor and correction frequency before declaring a replacement cheaper.

04When a reporting layer is enough

A reporting layer is a plausible middle choice when official accounting and investment records are reliable, but family views require repeated manual aggregation. It can centralize feeds and present ownership, performance and documents while leaving the book owner in place. The office needs a written rule for exceptions: when the layer disagrees with the general ledger or approved valuation, which team investigates, which record changes and which report is held?

In the US$8.15m case, do not let an imported manager value silently overwrite an authorized 30 September number. Require source-date matching, duplicate detection, trust/SPV mapping, reviewer status and a way to reconcile the reported value with the retained books.

Give each difference one of three statuses: corrected in the authoritative record, accepted as a documented and approved basis/timing difference, or open—report held. An explanation without approval is not a resolution. If the product only displays a corrected dashboard but cannot return the exception to the record owner, the office may have bought another place for disagreement.

This route is not inherently cheaper or faster. Confirm integration scope, feeds, exception handling, document retention, user roles and export rights in the written proposal. The office should retain an accessible copy of the record and its reporting logic if it later changes providers.

05When to migrate the record

Consider a core-system migration when the authoritative schedule or books cannot be maintained with repeatable ownership and review. Warning signs include several competing “master” files, adjustments that never reach the entity ledger, material numbers that cannot be traced to dated documents, and close cycles that depend on one person's memory. These are symptoms to investigate, not an automatic purchase order.

First identify what will migrate: investment positions, transactions, entity general ledgers, ownership structures, documents, historical reports, or some combination. An investment ledger is not automatically a legal-entity general ledger. A platform may provide one and integrate with the other. Assign authority for each record before estimating cost.

The migration acceptance test should replay the US$8.00m-to-US$8.15m correction in a parallel period. Compare opening positions, transaction history, book-basis balances, ownership views, eliminations where relevant and released reports against the approved old process.

Keep a variance register with owner, cause, correction and sign-off. Do not switch off the prior process because two headline totals match; the underlying entity and source lineage must also pass.

06A fair cost and cutover test

Compare three options on the same scope and volume: control the existing workbook, retain the books and add a reporting layer, or move specified records to an integrated platform. For each, count initial setup, annual license, implementation, feed/data costs, training, retained staff time, controls, export/exit work and parallel running. Put uncertain prices in the quote-request column, not as zero.

Calculate a comparable first-year total as one-time setup and migration + 12 months of recurring charges + retained staff and parallel-close work + priced add-ons needed for the common scope. Do not count the same migration labor in setup and retained staff work. An unquoted or uncertain item remains unknown, never zero. Calculate the ongoing year without one-time setup, using the same workload and service levels. Track family or CFO review hours separately unless the same hourly value is assigned consistently to every option. A vendor may reduce manual entry yet increase review, exception or integration work; price the work that remains.

A practical request to each shortlisted vendor is: “Using our fictional SPV and two trusts, show a revised 30 September manager statement changing the fund interest from US$8.00m to US$8.15m. Show the source record, owner approval, Trust A at US$4.89m, Trust B at US$3.26m, SPV total at US$8.15m, any accounting decision, the superseded report and the corrected report. Then export the data and audit evidence.” The office must separately validate that its own security, privacy and US delivery requirements are met; a canned demo does not settle those terms.

Put the result through three gates:

1. Demo: pass, fail or unknown for the exact correction, roles, exception and export requirements. Do not award a pass from brochure prose.
2. Contract: name modules, data ownership, implementation responsibilities, service levels, pricing, exit files and how unresolved gaps will be closed.
3. Cutover: complete a parallel close and sign off the record-level variances before the old file or system is retired.

07The decision record

Write one sentence for the office's present choice and one condition that would reverse it. Apply the same fixture, then assign a status to each route:

• Keep — conditional: three varied actual corrections and current access/release controls pass at acceptable effort. Stop and repair a control failure before relying on the workbook for the affected report.
• Add a layer — conditional: the authoritative books/schedules pass, aggregation or distribution fails, and the configured layer demonstrates how an exception returns to the record owner. Hold the report while a material exception is open.
• Migrate — conditional: an authoritative record itself fails, and the specific replacement scope, contract and parallel close pass. Stop cutover if record-level variances or permissions remain unresolved.

For the fictional SPV, the present decision is test all three routes; no route is approved by this example. A controlled workbook could remain in use this quarter if the source, authorization, ownership calculation, reviewer and old/new reports are reproducible. The office could test a reporting layer if consolidation remains the bottleneck; it should defer ledger migration until it has proved which books it intends to move and a parallel close passes. The supplied facts do not establish that any real office or vendor passes those gates.

The useful output is a named owner and next evidence request: three historical corrections to reconstruct, one common demo script, written implementation scope and a target parallel period. A family office should change systems to solve a measured record or workflow failure, not simply because its workbook has become unfashionable.

08Frequently asked questions

1Is Excel unsuitable for family-office reporting?

No. It can serve a controlled schedule or scenario model when ownership, permissions, review and version retention are reliable. It becomes risky when nobody can prove the source, formulas, corrections and approved report across entities and people.

2Does family-office software replace the general ledger?

Sometimes a selected product and implementation may maintain a legal-entity ledger; others focus on aggregation or reporting. Ask which record is authoritative for each entity and transaction, and test accounting and report outputs separately.

3How do we know whether migration worked?

Run a parallel period using approved source documents and reconcile opening positions, transactions, entity balances, ownership views, exceptions and released reports. Record every variance and its owner. A matching dashboard total alone is insufficient.