1. Home
  2. Insights

Why private equity firms and family offices need a live model for every portfolio asset

· Umesh Sharma

Ask a simple question in most investment firms: if interest rates stay where they are for another year, which of our companies runs short of covenant headroom first? The answer is usually “give us a week”. Someone has to find the latest model for each company, check whether it reflects the last quarter’s numbers, rebuild the debt schedule and hope the formulas still hold.

The firm owns the assets and has the data, and its people are capable. What it lacks is a way to see the portfolio as it stands today and to test what happens next. This article sets out why that gap matters for private equity firms and family offices, and how to close it.

Why it matters

Returns are made after the deal is signed

Financial engineering and rising multiples used to carry a large share of private equity returns. With debt more expensive and exit multiples under pressure, more of the return has to come from what happens inside the business: revenue growth, margin, cash conversion. Those are things you can only manage if you measure them every month against a plan.

The deal model built at acquisition is normally the most carefully built model the asset will ever have. It is also usually abandoned within a few months of completion. The investment case, the assumptions behind it and the value creation plan sit in one file, while actual performance is reported in another format by the company’s finance team. Nobody joins the two, so nobody can say with confidence whether the asset is ahead of or behind the case it was bought on.

Debt does not forgive a late surprise

Most portfolio companies carry debt, and the debt comes with covenants. A leverage or interest cover test that fails is rarely a shock to anyone who was watching the trend. The trouble is that the trend is often only visible once the quarter has closed and the compliance certificate is being prepared.

A forecast that carries the debt schedule and the covenant tests forward shows the pinch point months earlier. That is the difference between a calm conversation with a lender about an amendment and a rushed negotiation after a breach. The same forecast answers the refinancing question: how much can this business carry, on what terms, and what does the cash look like if trading is ten per cent below plan?

Investors and principals expect faster answers

Limited partners ask more detailed questions than they did five years ago, and they ask them more often. Valuations need to be supported. Quarterly reports need commentary that explains what moved and why. A firm that assembles this by hand each quarter spends its best people’s time on collecting numbers instead of interpreting them.

Family offices have the same problem with fewer people

A family office often holds a wider mix than a fund: operating businesses, property, direct stakes, fund commitments and loans, sometimes across several jurisdictions and entities. The team is small. Knowledge of how each holding is modelled tends to live with one or two people and in files only they fully understand.

That is a real risk. When the principal asks what the family’s total exposure to one sector or one lender is, or how much cash will be available for a new commitment next spring, the answer should not depend on whether a particular person is in the office that week.

What goes wrong with the spreadsheet approach

Spreadsheets are a good tool for building a model. They are a poor tool for running a portfolio. The usual problems are familiar:

  • Every company reports in its own layout, so someone re-keys figures into a master file every month.
  • Each model has its own structure, so nothing can be added up or compared without manual work.
  • Versions multiply. Nobody is sure which file went to the board or the lender.
  • A changed formula leaves no trace. Errors are found late, if at all.
  • Scenarios are built by saving a copy and overwriting inputs, so they cannot be kept in step with the base case.
  • Reporting is a separate exercise: copy the charts, write the commentary, format the pack.

None of this is unusual. It is simply what happens when a tool designed for one analyst is asked to serve a whole firm.

How to do it properly

The approach below comes from building financial modelling software for exactly this kind of work, and from years of doing it by hand before that. The tool matters less than the discipline it enforces.

1. Put every asset on one structure

Start with a common set of line items for the income statement, balance sheet and cash flow, and map each company’s accounts to it once. Take actuals straight from the accounting system where you can, or from a standard upload where you cannot. After that, each month’s numbers arrive in the same shape for every asset, and the re-keying stops.

This step is unglamorous and it is the one that makes everything else possible. Comparison, aggregation and automation all depend on it.

2. Forecast from drivers, across three linked statements

A forecast built by typing in a revenue line is hard to challenge. A forecast built from drivers, such as volumes, prices, occupancy, headcount and payment terms, can be questioned one assumption at a time. The income statement, balance sheet and cash flow should be linked so that a change in one driver flows through to cash and debt without anyone touching a formula.

Roll the forecast forward as each month of actuals lands. The model then always shows the same thing: what has happened, and what the latest view of the future is.

3. Keep scenarios alive, not archived

A base case, a downside and an upside should sit side by side in the same model, sharing the same actuals and differing only in the assumptions you chose to change. When a new month arrives, all three update. When the committee asks “what if the price rise slips by two quarters?”, the answer takes minutes.

4. Model the debt and test the covenants every month

Each facility should be in the model with its real terms: margin, amortisation, fees, maturity. The covenant tests should be calculated from the forecast for every future test date, with headroom shown in money and in percentage terms. Across the portfolio, this produces a single view of which assets are comfortable, which are tightening and when.

5. Explain the movement, not just the number

A KPI dashboard tells you EBITDA is below plan. A bridge tells you why: this much from volume, this much from price, this much from cost. Variance analysis against budget, against last year and against the investment case should be produced the same way for every asset, every month. Simple checks that flag unusual movements, such as a margin that jumps or a working capital balance that does not reconcile, catch errors and real problems early.

6. Control who can change what

A portfolio model is only trusted if people know where each number came from. That means role-based access, so that a portfolio company’s finance team can submit actuals without altering the investor’s assumptions. It means a record of every change, with a name and a date. And it means locking each reporting period once it has been approved.

7. Let reporting fall out of the model

If the first six steps are in place, the board pack, the lender package and the investor update are outputs, not projects. Charts come from the model. A first draft of the commentary can be written by AI from the variances and bridges, with every figure traceable to its source. A person then reviews, corrects and signs it off.

Where AI helps and where it does not

AI is good at the repetitive parts: reading management accounts and loan documents, mapping accounts to a standard structure, checking that numbers reconcile, drafting commentary and flagging what looks odd. It can run these steps every month without getting tired or skipping a check.

It does not replace judgement. Whether a management team’s forecast is credible, whether a “one-off” cost is really a one-off, whether to support a company through a difficult year or to sell: those decisions belong to the investor. The purpose of the system is to put the right information in front of that person sooner.

Where to start

You do not need to rebuild everything at once. A sensible order is:

  1. Pick one or two assets, ideally the ones with debt or the ones you worry about most.
  2. Agree the standard structure and map those companies to it.
  3. Build the linked forecast with the debt schedule and covenant tests.
  4. Run it alongside the existing process for a quarter and compare the results.
  5. Extend it to the rest of the portfolio, then automate data collection and reporting.

By the end of the first quarter you will know how long reporting really takes, where the data is weak and which assets need attention. That alone is usually worth the effort.

If you would like help designing the structure, building the models or automating the monthly cycle around them, get in touch.

Start with one conversation.

Tell us what you are trying to fund, fix, value or automate. In a 30-minute call we'll tell you honestly what we can do, roughly what it involves, and what a first step would look like.

Get a quote