How to prepare a portfolio performance review from scattered reports

1. Write down the decision the review makes

The written decision tells you which figures belong in the pack. A performance review answers three questions: what changed, why it changed, and what the firm does about it. For a buyout portfolio, the written decision reads like this: decide which companies go on the watch list this quarter and what intervention each one gets.

The watch list makes a strong first decision to automate because it has four traits:

• It repeats every quarter.
• A wrong call costs money.
• Its inputs sit across several sources and in people's heads.
• Rules and precedent bind it.

A one-off question, such as a single diligence memo, misses the first trait.

Then list the measures the decision reads. For a watch list, those are typically revenue against plan, EBITDA margin, cash, covenant headroom, and value creation plan milestones. Fund-level measures serve a different decision, and for those Cambridge Associates recommends reading IRR alongside multiples such as DPI and TVPI in A Framework for Benchmarking. Drop any measure that feeds no decision.

2. List every source, spreadsheets included

A recurring review needs both the company reports and your spreadsheets, because the spreadsheets hold rules the reports lack. Company reports arrive as management accounts, board decks, KPI templates, and lender compliance certificates. The spreadsheets include the deal team's model, the operating team's tracker, and the workbook the review is assembled in.

For each source, record four things:

• The company.
• The period the source covers.
• The date it arrived.
• Where each figure sits in it, down to the page and table, or the sheet and cell.

That exact locator turns a number into evidence a reviewer can open.

If your firm's reporting runs on spreadsheets, start by reading each one for the rules it holds. A formula that adds back a one-off cost is a definition of adjusted EBITDA, and a cell someone overwrites every quarter is a correction nobody wrote down. A tab that converts currencies at a fixed rate is a policy. Write each rule down in plain sentences before you change any tool, so the list becomes your firm's decision logic for the review and carries over to whatever software you pick later.

Set the review calendar against what your investors expect. Invest Europe's Investor Reporting Guidelines (2024) describe quarterly reports to investors within 60 days and annual accounts within 120 days, and the Institutional Limited Partners Association (ILPA) publishes reporting templates for limited partners. Treat both as references for timing and format, since your own fund documents set what you owe.

3. Keep data cleanup apart from the review decision

To separate data cleanup from the review decision, run them as two stages, each with its own owner and record. Cleanup extracts each figure, aligns its unit, currency, and period, and maps each company's line items to the firm's list of measures. Its output is one table of facts, each with a period, a source, and a locator.

The review decision reads from that table, applies the definitions, the source ranking, and the thresholds, and produces the watch list. A fix made by hand inside the review workbook stays hidden there, so next quarter someone has to make it again. Move that fix into the cleanup stage with its reason recorded, and it's made once and applied every quarter after. A judgment call, such as accepting a restated figure, belongs in the decision stage with a named approver.

4. Write the rules that settle each figure

Each figure in the pack needs three written rules: its definition, the source that outranks the others when two disagree, and the threshold that sends it to the decision.

Take a made-up portfolio company, Merrow, a B2B software business. Every figure here is invented for illustration.

• Merrow's Q2 management accounts report annual recurring revenue (ARR) of €14.2 million on page 6. The Q2 board deck reports €14.9 million on slide 4. The deck counts contracts signed in June that go live in Q3.
• The firm's written rule says ARR counts contracts live at period end, and management accounts outrank the board deck for any revenue figure.
• The pack shows €14.2 million with its locator, page 6 of the management accounts. The deck's €14.9 million appears beside that figure, with the reason the firm's rule set it aside, so the disagreement stays on the record.

In the same quarter, Merrow's CFO began excluding a restructuring cost from adjusted EBITDA, so Q1 and Q2 need one definition before anyone compares them. For both cases in full, see when reports disagree and changed definitions and missing facts.

5. Give every figure one of four outcomes

Every figure in a portfolio review ends in one of four ways: fulfilled with proof, refused by a rule, a named gap, or reserved for a human. Here is how the four outcomes apply to Merrow's Q2 figures and the watch-list call:

The committee sees from the pack which figures carry proof, which are waiting, and why.

The common mistake is filling a missing figure with a plausible estimate to finish the pack, and on the page an estimate looks identical to a checked figure.

6. Name who signs off, and keep each correction

For Merrow, the operating partner decides the watch-list placement and the fund's finance lead approves the restated EBITDA. Each correction the finance lead approves becomes a rule the next cycle applies. The record keeps who approved, the date, the figure before the correction, the figure after it, and the rule the correction creates. So next quarter, that rule restates Merrow's EBITDA on the agreed definition before the pack is built, and nobody has to ask again.

For approval chains and how a correction carries forward, see who signs off a figure.

7. Add new sources by mapping them to the rules you have

To add a new source without repeating cleanup each cycle, map it to the definitions and ranking you already have. An add-on acquisition, a company that switches accounting systems, and a lender that adds a report each bring a new source.

To add one:

1. Map each field to an existing definition.
2. Place the source in the ranking.
3. If it replaces an older source, run both for one cycle, so any difference surfaces as a disagreement with both figures shown.

Add a new definition when the new source reports something the review has never measured. Record each mapping once, and every later cycle reuses it.

How the method applies to a venture portfolio

A venture portfolio follows the same seven steps with a different decision and more named gaps. For a venture fund, a common recurring decision is follow-on allocation: which companies receive reserves, and how much each one gets.

Some companies report on their own templates without audited accounts. Their reports arrive monthly or at irregular intervals. So cleanup carries more of the work, and the named gaps state which company owes which figure. Tracked across cycles, those gaps show which companies report least. That pattern is itself an input to the follow-on decision.

How to tell the review is working

Over successive cycles, the work should shift from cleanup to the decision. Four counts show it:

• Figures in each of the four outcomes, per company, per cycle.
• Corrections that recur. These should fall toward zero as each one becomes a rule.
• Named gaps per company over time.
• Days from the last report arriving to the pack going to committee.

What software helps turn portfolio company reports into a reviewable pack?

Two kinds of software serve a private equity portfolio review. A portfolio monitoring platform collects, stores, and reports company data. Software that builds decision logic for enterprise agents writes the firm's rules down, and an agent applies them to each figure and records its outcome. For the platforms themselves, see what each portfolio monitoring tool does.

Keep your platform for the data it collects. The rules you wrote down in steps 2 and 4 decide what the review does with that data.

Where Saphenia fits in the portfolio review

For a portfolio review, Saphenia reads your reports and spreadsheets, then writes down the rules the review follows. Those rules say what your terms mean, which source wins when two disagree, and where the thresholds sit. The decision logic for the watch list keeps them, with the corrections your approvers sign off on. An agent then assembles each cycle's review for the committee and checks every figure against your rules before the pack goes out. Each figure ends in one of the four outcomes, with a record.

Each fact carries the period in which it was valid, and each rule is versioned. A proof holds the facts, the rule, its version, and the date. Approved corrections move the judgment in your experts' heads into your firm's own record of facts, rules, and decisions. That record runs in the environment you choose, and you take it with you if you leave. Saphenia never trains on your data.

Saphenia builds the decision logic in days, and the first decision ships in weeks. Every Saphenia engagement begins with a live agent on one real decision. To test this on your own documents, see how to evaluate a portfolio review pilot. For hosting and data ownership, see the facts buyers ask for, and for cost drivers, see how Saphenia is priced.