Hiring guide
Sales pipeline in a cash forecast: separate evidence from probability
By CFO Index · Published
A probability-weighted sales pipeline can help summarise opportunities, but it cannot tell you which customer will pay before payroll. Ask your fractional CFO to preserve the deal-level evidence and bridge closing, delivery, invoicing and collection dates. Treat the sales forecast and the cash schedule as connected views, not interchangeable totals.
Confirm what the CRM number means
Identify the report, date filter, amount field, currency and probability setting before importing a pipeline total. HubSpot's documented weighted pipeline report multiplies deal amounts by stage probabilities. Its forecast setup also permits category mapping and manual category changes. Those are product behaviours, not proof that a particular deal will close or be collected. Keep the exported settings with the forecast and inspect unusually large deals individually. Other CRMs and account configurations can behave differently.
Use evidence layers rather than stronger-sounding labels
The following classification is a proposed management method, not a replacement for your CRM stages or accounting policy. Store each deal once under a stable ID and maintain separate dates for expected signature, delivery, billing and collection. Record the last evidence update and the person who supplied it. A signed order can still carry cancellation, acceptance or collection risk; an internal 'commit' label is not itself a signed customer agreement.
| Layer | Evidence to request | Cash-planning question |
|---|---|---|
| Signed work | Executed terms, delivery conditions and billing schedule | When can we invoice, and when is payment expected? |
| Probable but unsigned | Buyer confirmation, unresolved conditions and next dated step | Which receipt dates remain possible if signature slips? |
| Speculative opportunity | Identified need and an owner, but limited customer commitment | Should this remain outside the near-term base case? |
| Already invoiced | Invoice ID, due date, dispute status and collection evidence | Has it been removed from new-sales receipts to avoid duplication? |
A weighted €95,000 is not a €95,000 receipt
Hypothetical example, excluding tax and assuming all amounts are in euros: deal A is €40,000 of signed work, assigned 100% for this model; deal B is €60,000 at an illustrative 50%; deal C is €100,000 at an illustrative 25%. The weighted sales total is €40,000 + €30,000 + €25,000 = €95,000. These weights are invented for the example, not conversion benchmarks. The total is an expected-value calculation; no customer is obliged to pay that combined number.
Bridge the same three deals to the collection calendar
Assume A invoices €40,000 in October and is expected to pay on 15 November. B, if signed on schedule, pays a 20% deposit of €12,000 on 20 October, with the remainder after delivery in December. C has no supported collection date. For a cash check immediately before payroll on 25 October, A contributes no receipt yet, B contributes either €12,000 or nothing, and C remains outside the near-term base case. Delivery costs and any tax payments still need their own dated cash rows.
Test the actual alternatives around a large deal
Continue the hypothetical example: opening usable cash is €20,000 and all other payments through payroll total €26,000. With B's deposit received on time, the closing balance is €6,000. If it slips beyond payroll, the balance is minus €6,000. Adding a probability-weighted €6,000 deposit produces a zero balance, but neither actual outcome is zero. Keep both discrete cases visible. A weighted aggregate may be useful across a broader portfolio, but it should not conceal a single-deal timing dependency.
Freeze the pipeline before testing its accuracy
Archive the deal list as it stood at the forecast cut-off. Later, compare that same cohort with actual signatures and collections over the stated horizon. Separate lost deals, delayed signatures, delayed invoices and delayed payments; do not add newly created deals to make the original forecast look more accurate. Review stage conversion separately from payment timing. Small or changing cohorts may not support a stable probability estimate, so show the number of deals and major exceptions rather than reporting unexplained precision.
Agree how new evidence changes the forecast
Sales should own the customer evidence and next commercial step; delivery should confirm capacity and acceptance dependencies; finance should map those facts into receipts and spending. Define who approves changes to the base case and which missed milestone prompts an earlier review. Request a handover test: move one unsigned deal by a month and confirm that its deposit, delivery costs and final collection move correctly without duplicating an invoice. This is management forecasting, not a rule for recognising revenue in the accounts.