Every PE-backed CFO eventually meets the 13-week cash flow forecast. Not because anyone enjoys it, but because somewhere between the LBO close and the first covenant test, the sponsor and the lender quietly agreed it would be the lingua franca of operational truth. Done well, the 13W is a steering instrument that compresses the distance between a working-capital wobble and a corrective action. Done badly, it is a Friday-afternoon copy-paste exercise that nobody trusts and everyone signs off on anyway.
This piece is the playbook we wish every newly minted portfolio CFO had on day one. It is written for the operator who already knows what unitranche means, who has sat through a leverage step-down conversation, and who needs the 13W to do real work rather than decorate a board pack. We will cover why sponsors care so intensely about this artefact, how to build a defensible row-by-row model, what actual really means at week W+1, how to present the output to LPs and lenders without losing the room, the five red flags that make investors reach for the phone, and the automation ladder from spreadsheet to AI copilot.
Why PE firms insist on the 13-week horizon
There is a comfortable myth that the 13-week cash flow forecast is an LBO-era relic, a tool from the era of distressed debt that got sticky. The truth is more operational. Sponsors at Eurazeo, Astorg, and the operating teams at growth-stage houses like Sequoia and Index continue to ask for it because three modern realities compound it into something close to a regulatory requirement.
Covenant cadence runs on weeks, not quarters
Modern unitranche and TLB structures are typically tested quarterly, but the leverage and DSCR ratios that drive those tests are a function of trailing twelve-month EBITDA against a point-in-time net debt position. A single working capital swing in week 9 of a quarter can move the headroom by 0.3 turns. The 13W forecast is the only artefact that gives the CFO and the deal team line of sight from today through the next testing date with weekly granularity.
Lender confidence is bought in advance
When a borrower pre-empts a covenant breach, lenders extend grace. When a borrower is surprised by one, lenders extract pricing and consents. The 13W is the document that demonstrates situational awareness. Agency banks running the credit on behalf of a club of direct lenders read the 13W as a proxy for management quality. A forecast that has reconciled cleanly to actuals for eight consecutive weeks earns the right to ask for a step-down or a covenant holiday.
Intervention speed is the sponsor's real edge
A PE operating partner who spots a 4-week receipts shortfall on a Tuesday can mobilise a working capital playbook by Thursday: tightening DSO, deferring non-essential capex, drawing on the RCF, or accelerating a factoring line. The 13W is the trigger mechanism for that intervention. Without it, problems are discovered in the monthly close, two to six weeks after the fact, by which point liquidity has compounded into a credibility issue.
Row-by-row anatomy of a defensible 13W
A defensible 13W has structure before it has numbers. The temptation to start with last year's bank statement and stretch it forward is the single biggest reason most forecasts fail their first variance test. A proper model builds from the receipt and disbursement primitives, layers in financing and investing flows, and reconciles to a closing cash position with explicit headroom against the operating facility.
Below is the canonical structure we deploy across portfolio companies. It runs to roughly twenty rows because anything fewer hides risk and anything more becomes decorative. Each row has a single owner inside the finance team and a documented source system.
| Section | Row | Description | Owner | Source |
|---|---|---|---|---|
| Opening | Opening cash position | Bank balance at start of week, net of in-flight items | Treasury | Bank feed / TMS |
| Operating Receipts | Customer collections — top 20 accounts | Named-account receipts forecast bottom-up from AR ageing | Credit Control | AR ledger + customer commitments |
| Operating Receipts | Customer collections — long tail | Statistical forecast based on DSO and invoice mix | Credit Control | AR ledger model |
| Operating Receipts | Other operating receipts | Insurance recoveries, rebates, royalty income | FP&A | Contracts register |
| Operating Receipts | VAT / sales tax refunds | Net VAT receivable from tax authority | Tax | VAT return schedule |
| Operating Disbursements | Payroll and social charges | Net pay plus employer contributions, by pay date | HR / Payroll | HRIS payroll calendar |
| Operating Disbursements | Trade payables — strategic suppliers | Named-vendor commitments with negotiated terms | Procurement | AP ledger + PO commitments |
| Operating Disbursements | Trade payables — long tail | Statistical run-rate from rolling AP ageing | AP | AP ledger model |
| Operating Disbursements | Rent and property | Lease payments, service charges, utilities | FP&A | Lease register |
| Operating Disbursements | VAT / corporate tax payments | Output VAT, withholding, instalments | Tax | Tax calendar |
| Operating Disbursements | Other operating disbursements | IT, professional fees, travel, marketing | FP&A | Budget commitments |
| Net Operating | Net operating cash flow | Operating receipts minus operating disbursements | Treasury | Calculated |
| Financing | Interest payments | Cash interest on TLB, unitranche, RCF utilisation fees | Treasury | Facility schedule |
| Financing | Scheduled debt amortisation | Mandatory principal repayments by tranche | Treasury | Facility schedule |
| Financing | RCF draws and repayments | Net movement on the revolving credit facility | Treasury | Bank feed |
| Investing | Capital expenditure | Maintenance and growth capex by approved project | Operations | Capex register |
| Investing | M&A and integration costs | Bolt-on consideration, integration spend, earn-outs | Corp Dev | SPA + integration plan |
| Net Cash | Net cash movement | Sum of operating, financing and investing flows | Treasury | Calculated |
| Closing | Closing cash position | Opening plus net cash movement | Treasury | Calculated |
| Headroom | RCF availability | Committed RCF less drawn less ancillaries | Treasury | Facility + bank feed |
| Headroom | Total liquidity headroom | Closing cash plus available RCF, the number the sponsor reads first | Treasury | Calculated |
Three details separate a model that survives sponsor scrutiny from one that does not. First, the top-20 customer and strategic-supplier rows are forecast by name, with a comment column citing the contractual basis or commercial commitment. The long tail is allowed to be statistical, but the named rows must be defensible. Second, the financing block is a literal copy of the facility schedule — no rounding, no smoothing — because a lender will tie the model directly to the loan agreement. Third, the headroom row is the one the deal partner reads first, so it must be calculated in a way that matches the covenant definition exactly, including the treatment of letters of credit, ancillaries and any clean-down requirements.
Variance analysis and what “actual” really means
Variance is where most 13W programmes quietly fail. The mechanical task is trivial: at the end of week W, replace the forecast for that week with the actual outturn and compute the difference. The hard part is deciding what actual means, because the answer is not as obvious as the bank statement.
The bank statement is necessary but insufficient
A bank balance at Friday close captures the cash position but not the working capital story. A receipt that landed on Monday but related to an invoice you had forecast for the previous Friday is a timing variance, not a value variance. A payment to a strategic supplier that was rescheduled by mutual agreement is a decision variance, not a forecasting failure. A 13W variance pack that does not decompose the gap is just noise.
The three-bucket decomposition
The discipline we run across the portfolio is a three-bucket variance model. Each line variance for the past week is classified into one of three categories and explained in a single sentence in the variance commentary.
- Timing variance. The receipt or disbursement happened, but in a different week than forecast. The cumulative variance from start of horizon should be near zero. The corrective action is to update the timing assumption for analogous future events.
- Volume or value variance. The receipt or disbursement happened on time but at a different magnitude. This is the variance that matters most for the EBITDA bridge and for covenant projections. The corrective action is to investigate the underlying commercial or operational driver.
- Decision variance. A discretionary action by management changed the cash flow. A capex deferral, a payment-run rescheduling, an early collection campaign. These should be flagged explicitly because they change the run-rate going forward.
The rolling correction logic
At the end of week W, three things happen. The week-W forecast column is replaced with actuals. Weeks W+1 through W+12 are re-forecast using the variance learnings. A new week W+13 is added at the far end of the horizon, built from scratch using the latest run-rate and any known events. The model is therefore always exactly thirteen weeks long, always anchored on today, always informed by the prior week's variance.
We do not need the 13W to be right. We need it to be wrong in explainable ways. A model with a forecast accuracy ratio of 0.78 and a clean variance commentary is more useful than a model at 0.92 with no narrative. The narrative is the product.
Presenting to LPs, lenders and the board
The 13W is a working document, but it is also a communications artefact. The version the treasury team runs on Friday is not the version that goes to the lender on Monday or to the board on the third Tuesday of the month. Each audience has a different tolerance for detail and a different question they are trying to answer. Failing to translate is a common own goal.
Lenders and agency banks
Lenders want to see two things: the headroom against the operating facility for every week of the horizon, and the projected covenant ratios at the next test date. They do not need the line-item detail, but they want to know it exists. The standard pack is a one-page summary with the weekly headroom chart, a covenant projection table, and a short variance commentary referencing the prior week. Send it on the same day every week. Predictability is a feature.
The sponsor and the operating partner
The sponsor wants the operational story. They will read the variance commentary first, the headroom chart second, and the covenant table only if either of the first two raise a flag. The cadence is typically weekly during the first six months post-close, monthly thereafter, and back to weekly in the run-up to a refinancing, an exit process, or any covenant test where headroom is below 15%.
The board and the LP advisory
The board sees the 13W in summary form, usually as a four-week-out liquidity position with a commentary on key risks and the corrective actions in flight. LPs do not see the 13W directly, but their questions in the annual meeting are informed by the GP's confidence in the artefact. A sponsor who can speak fluently about portfolio liquidity at the LP advisory has a 13W programme that works.
Five red flags PE investors look for
Sponsors who have sat through enough portfolio reviews develop a fast filter for 13W health. The signal is rarely in a single number; it is in the pattern of how the model behaves over time. These are the five flags that, in our experience, cause an operating partner to pick up the phone before the next monthly call.
1. Persistent positive timing variance on receipts
When customer collections consistently come in later than forecast — week after week, even in small amounts — it is almost never a forecasting problem. It is a DSO drift problem, and DSO drift is the leading indicator of a customer-quality deterioration or an internal collections process that has lost its edge. The 13W surfaces it three to four weeks before the monthly DSO report does.
2. RCF utilisation creeping up week-on-week
A revolver that is used tactically for working capital seasonality is healthy. A revolver that is drawn a little more each week, with no clean-down in sight, is a structural liquidity issue masquerading as a timing issue. Sponsors track the four-week trend in RCF utilisation as carefully as they track EBITDA.
3. The variance commentary that says “timing” for everything
A variance pack that classifies every line as a timing variance is a variance pack written by someone who has not actually investigated the gaps. Timing variances should net to near zero across the horizon. If they do not, they are not timing variances. Sponsors read the commentary specifically to test whether the finance team understands the business or is producing the artefact mechanically.
4. Covenant headroom narrowing with no corresponding action plan
If projected DSCR or leverage headroom narrows materially across two consecutive 13W refreshes, and the management commentary does not reference a specific intervention — a cost action, a working capital initiative, a capex deferral — the sponsor will assume the management team has not yet recognised the trend. That is a credibility event, and it is avoidable.
5. Round numbers in the long-tail rows
When the long-tail customer collections row is a perfectly round number every week, or the other operating disbursements row never varies, it tells the sponsor the model is not being refreshed at the line level. Real businesses produce ragged numbers. A 13W with too many round numbers is a 13W that has stopped doing useful work.
The automation maturity ladder
Most portfolio companies arrive at PE ownership with the 13W living in a spreadsheet. That is fine for the first quarter and increasingly painful thereafter. The maturity ladder below is the path we see across the portfolio, and the trigger points that justify each step up.
Rung 1 — Spreadsheet
A well-built Excel model with named ranges, a structured input sheet per source system, and a variance tab. Refreshed manually on Fridays from extracts of the AR and AP ledgers, the bank feed, and the payroll calendar. Works for businesses with a single ERP, fewer than five bank accounts, and a finance team with bandwidth for a four-hour weekly refresh. Falls over when the business grows through acquisition, when entity count exceeds about ten, or when the lender asks for a daily cash position.
Rung 2 — Treasury Management System
A dedicated TMS — Kyriba, Coupa Treasury, Trovata, or one of the European mid-market vendors — automates the bank-feed reconciliation, centralises the cash position across entities and currencies, and provides a forecasting module that ingests AR and AP ledger data. The 13W becomes a configured report rather than a hand-built spreadsheet. The trigger to move here is usually multi-entity complexity or a lender requirement for daily liquidity reporting.
Rung 3 — AI copilot on top of the TMS or ERP
The newest rung, and the one that genuinely changes the labour profile of the treasury function. An AI copilot ingests the same source systems as the TMS but adds three capabilities the TMS does not provide natively. It auto-classifies variance into the timing, value and decision buckets. It drafts the variance commentary in the house style, citing specific invoices and customer accounts. And it stress-tests the forecast against scenario libraries — a bolt-on, a customer loss, a covenant step-down — so the CFO walks into the sponsor call with the answers already prepared.
The economics of the third rung are notable. A treasury team that previously spent two FTE-days per week on the 13W refresh and variance pack reclaims roughly 70% of that time. The output quality is more consistent because the narrative no longer depends on which team member happened to write it. And the forecast accuracy ratio typically improves by 5 to 10 points within the first quarter, because the model is being refreshed at the line level by a system that does not get tired.
The Friday-afternoon 13W refresh checklist
The discipline of a good 13W programme lives in the weekly ritual. The list below is the checklist we have institutionalised across portfolio treasury functions. It is deliberately ordered: each step depends on the previous one, and skipping a step is the most common cause of a model that drifts.
- Pull the bank position as of Friday close across every account, every entity, every currency. Reconcile to the TMS or to the bank feed. The opening cash row for next week's model is non-negotiable.
- Replace the current week's forecast columns with actuals. Compute the line-by-line variance and tag each variance as timing, value or decision.
- Refresh the AR ageing extract and update the named-account receipts rows for the next four weeks. Anything beyond week four is statistical. Update the long-tail row using the rolling DSO.
- Refresh the AP ageing and the open PO list. Update the strategic-supplier rows by name. Confirm the next two payment runs with the AP team.
- Pull the payroll calendar and confirm the gross-to-net for the next pay date, including any bonus or commission events that fall in the horizon.
- Update the financing block from the facility schedule. Verify interest accrual calculations against the latest reference rate fixing. Reconcile RCF utilisation to the bank feed.
- Update the capex row from the approved project list. Flag any project that has slipped out of the horizon or accelerated into it. Reconcile to the integration plan if there is a live bolt-on.
- Re-compute the closing cash and total liquidity headroom rows for every week. Compare the projected covenant headroom at the next test date to last week's projection. If it has moved by more than 5%, write the explanation now.
- Add the new week-thirteen column at the far end of the horizon. Build it from run-rate and known events. Do not copy week twelve.
- Write the variance commentary in three short paragraphs: what happened this week, what it implies for the next four weeks, what action is in flight.
- Review the pack with the finance director before it leaves the building. Send to the lender and the deal team at the same time, every week, ideally before 17:00 local time on Friday.
The checklist looks long. In practice, with a mature data layer and an AI copilot sitting on top of the TMS, the entire refresh runs in under ninety minutes. Without them, it is a Friday afternoon and a chunk of Saturday morning for a treasury analyst, every week, for the life of the investment. The cumulative cost over a five-year hold is the equivalent of one full-time hire — which is exactly the business case that drives sponsors to push portfolio CFOs up the maturity ladder in the first place.
See it run on your numbers.
Connect a single bank in 4 minutes. Get your first AI-prepared cash brief tomorrow morning.
Start your free trial14 days · No credit card required · Cancel anytime
Written by the Arxa Intelligence team — finance leaders, engineers, and treasury operators sharing what we've learned in the field. We don't ghostwrite under fake names; if you want to talk to whoever wrote a piece, email us at hello@arxaintelligence.com.