Why the Forecast, and Understanding the Business First
Before opening a single spreadsheet, there's one question that decides how the entire model gets built: what problem is this forecast actually solving? Get that wrong, and the result can be a technically correct spreadsheet that still gives the wrong answer.
"I want a cash flow forecast" can mean several different things. A lender wants to see one before renewing a credit line. An owner has been caught off guard by the bank balance once too often and wants a warning system. Growth has gotten fast enough that nobody's sure if there's enough cash to fund it. Each of those is a different job, and each changes what the model needs to emphasize — a forecast built to satisfy a bank covenant looks different from one built to give an owner a Monday-morning "what do I need to chase this week" list.
Before building anything, the real question is which of three jobs the forecast needs to do: prove liquidity to a lender, give early warning of a shortfall, or check whether growth is outrunning available cash. Each one changes what gets emphasized in the model — a lender-facing forecast leans on documented commitments; an early-warning forecast leans on weekly granularity and variance tracking.
In this case, the answer was the early-warning version: the owner wanted enough lead time to act before a shortfall, not a report that confirmed one after it had already happened. That single distinction shaped a lot of what came later in the model — in particular, the weekly (not monthly) granularity and the emphasis on tracing forecast-vs-actual gaps, covered later in this series.
Getting acquainted with the business
Next, before touching any numbers: how does this company actually make money?
This business sells specialized technical equipment to customers across many countries. That single fact already told me a lot before I'd opened a single invoice:
Multi-currency exposure, in particular, is the kind of thing worth confirming early rather than discovering three weeks into building the model. B2B equipment sales also tend to mix advance payments with credit terms — sometimes within the same customer relationship, which matters when a single customer's payment behavior doesn't fit one clean assumption.
I also noted each customer's industry. This isn't just paperwork — different industries order in different ways. A trading company might order at random times, with no pattern at all. A construction customer might order more in one season than another. A government or infrastructure customer might follow a buying cycle set by budgets, not by the calendar.
Why check this? Because it tells me whether a pattern is real or just chance. Say a customer ordered every few months for the past year. If their industry naturally runs on a cycle, that pattern is real, and I can expect it to continue. If their industry doesn't work that way, the same pattern might just be luck — three orders that happened to land a few months apart, with no real cause behind it. A forecast shouldn't be built on luck.
Here's what that mapping looked like once every customer with at least €5,000 in transactions over the trailing 12 months was sorted — no revenue figures yet, just a count of who these customers are. Smaller, occasional buyers below that line weren't part of this pass.
| Industry category | Customers | Share |
|---|---|---|
| Aerospace & Advanced Manufacturing | 32 | 73% |
| Trading & Distribution | 6 | 14% |
| Electronics & Component Manufacturing | 3 | 7% |
| Other | 3 | 7% |
44 customers in total, spread across 20 countries. Nearly three-quarters sitting in one broad category is itself a finding — it means most of this business's ordering behavior is going to be shaped by however that one sector tends to buy, not by "customers" as a generic average. That's a structural fact about the business, learned before a single dollar has been counted.
The same 44 customers also showed a clear pattern in how, and how often, they pay:
| How they pay | Customers | Share |
|---|---|---|
| Pay in advance (prepayment) | 37 | 84% |
| Pay on credit terms | 7 | 16% |
| Ordering pattern | Customers | Share |
|---|---|---|
| One payment only, so far | 6 | 14% |
| Every 2–3 months | 16 | 36% |
| Every 4–6 months | 17 | 39% |
| Irregular, no clear pattern | 1 | 2% |
| Other | 4 | 9% |
Most of these customers pay before the equipment ships, and most order every few months rather than on a fixed monthly cycle. That's exactly the kind of behavior a generic forecasting template — built around steady monthly billing — would get wrong from the start.
One more useful cut on the same 44 customers: how recently each one actually paid.
| Last payment | Customers | Share |
|---|---|---|
| Within the last month | 18 | 41% |
| 1–3 months ago | 12 | 27% |
| 4–6 months ago | 7 | 16% |
| Over 6 months ago, or no payment on record | 7 | 16% |
I also pay attention to:
- Historical days sales outstanding (DSO),
- Relationship strength,
- Seasonality.
What this phase produces
There are no forecast numbers yet. That's the point of this step. It builds the framework I'll use to make sense of the numbers (to interpret the numbers) once they arrive.
Skipping this step is the most common way a forecast ends up technically sound but practically useless — the formulas work, the totals tie out, and the model still doesn't reflect how this specific business actually moves cash. The next part covers where the real data work starts, and where the first genuine surprises showed up.
This is Part 1 of a 5-part case study. Part 2 — processing incoming money from customers — I plan to publish soon.
Why do I need all the information?
Not every expected receipt has the same probability of arriving on time. Two invoices with the same value may have completely different collection risk depending on the customer. Additional customer information helps estimate that risk and identify which assumptions deserve management attention.
Country
A customer's country can influence payment behaviour and collection risk. Payment cultures differ from one country to the next, and international transfers can take longer to clear than domestic ones. Political instability, sanctions, currency controls, or banking disruptions can delay payments outright, and legal collection procedures vary significantly between jurisdictions.
Country should not determine whether a payment is included in the forecast, but it can help explain why certain receipts deserve closer monitoring.
Industry
Industries often have predictable payment characteristics. Construction companies frequently pay after project milestones; retail businesses may experience strong seasonal cash flows; government organisations often have long but predictable approval processes; manufacturers may delay purchases during economic slowdowns.
Knowing the customer's industry helps management understand which payments are likely to follow established business cycles and which are more uncertain.
How often money is received
Historical payment frequency provides valuable context, but it should not replace information about known future business. It helps answer one question: how predictable is this customer?
| Customer type | Pattern |
|---|---|
| Recurring | Regular purchases with relatively predictable cash inflows. |
| Occasional | Purchases periodically, but irregularly. |
| One-time | No evidence of future business. |
This distinction lets management separate stable recurring revenue from unpredictable one-off transactions.
Reliance on large one-off payments
Many businesses unknowingly depend on a small number of large transactions. If several significant receipts depend on contracts that have not yet been signed, or on customers that buy only occasionally, the forecast may appear healthy while carrying substantial risk.
Reviewing payment frequency this way answers a few practical questions: how much of the forecast depends on customers who rarely buy, how much depends on contracts that have not yet been secured, and what percentage of expected cash receipts comes from unpredictable business. The greater the dependence on infrequent large deals, the greater the need for management attention and contingency planning.
Last payment date
The date of the last payment provides a simple indication of current customer activity.
| Probability | Basis |
|---|---|
| High | Paid within the last 1–2 months, or several recent payments. |
| Medium | Payments typically occur every 3–4 months. |
| Low | Only one payment in the last 5–9 months, or no recent payment history. |
This does not predict payment by itself, but it helps prioritise collection efforts and management review.
Payment terms
Payment terms reveal when cash should normally be received. Prepayment significantly reduces collection risk because cash is received before goods or services are delivered. Net 30, Net 60 and similar terms introduce credit risk and timing uncertainty. If a customer consistently ignores agreed payment terms, management may decide to require deposits or full prepayment, reduce credit limits, pause deliveries until overdue invoices are settled, or renegotiate contract terms.
Historical days sales outstanding (DSO)
DSO measures how long customers typically take to pay invoices. For businesses receiving mostly advance payments, DSO may provide limited forecasting value, because many receipts occur before invoices become due. For customers buying on credit, however, DSO remains useful because it identifies payment discipline — helping management decide which customers require closer collection monitoring, which deserve different payment terms, and where sales should negotiate deposits or milestone payments.
Relationship strength
The quality of the commercial relationship often influences future cash receipts more than accounting data alone.
This information rarely exists in accounting software. It is typically gathered from sales representatives, account managers, operations or delivery teams, and management's own experience. A simple scoring system — Strong, Good, Neutral, Weak, Poor — is usually sufficient, with supporting notes capturing context such as the key decision maker, previous disputes, price sensitivity, how quickly payments are usually approved, or a recent change in ownership.
Seasonality and known payment patterns
Many customers follow predictable annual or monthly patterns — retail customers placing large orders before holidays, construction companies slowing during winter, agricultural businesses paying after harvest, schools purchasing before the academic year, or manufacturers reducing activity during annual shutdowns. Understanding these patterns helps management distinguish between expected seasonal changes and genuine business problems.
Known payment patterns can improve forecast quality further — for example, a customer that always pays on the final day of the month, pays only after board approval, normally splits large invoices into several payments, or always delays payment by roughly a week. These behavioural patterns often produce more realistic cash forecasts than relying only on contractual due dates.