How do I forecast sales for a business with no history?

Forecast sales for a new business by starting with specific prospective customers, the steps still needed for each sale and the work you can actually deliver. Build a booked floor, a working case and a stretch case with dated assumptions. Then keep sales earned separate from deposits and later collections.

You do not need to manufacture a history to make a forecast. You need to make the uncertain parts visible enough that the next real conversation, quote or delivery can improve it.

Choose a unit you can explain

“We will capture a small share of a large market” is a distant ambition, not much help scheduling next Tuesday. Start with the thing a customer can order: a defined service package, a unit shipped, a paid appointment or another identifiable deliverable.

Describe what the price includes, how long delivery takes and what must arrive from the customer first. If those details are unresolved, forecast a range tied to the unresolved detail. A project that might need four hours or forty cannot safely be treated as one standard sale.

The SBA's business-planning guidance connects projections to the marketing and sales process and calls for specific first-year detail, including quarterly or monthly projections. That does not make an early forecast certain. It gives you a reason to explain how the numbers could happen.

Give every prospective sale a stage and a next event

Use a simple working register: organization, proposed offer, present stage, next decision, expected decision date, earliest delivery date and evidence supporting that date. Keep private identifying information in your own controlled records; a shared planning summary can use customer codes.

A person who downloaded a brochure is different from a person who approved a scope. An accepted order is different again. Avoid putting all three in a column labeled “customers” and multiplying by the same price.

Consider a fictional studio selling a twelve-slide training presentation for $600, using content already approved by the customer. Each package needs four paid production-and-checking hours at $40, plus $40 in incremental resources: $200 total job cost. The scope does not include writing the customer's training policy, unlimited revisions or sourcing missing content.

The studio's arrangement makes eight production hours available each week at that rate, paid only for hours actually used, with no minimum-hour charge. Capacity therefore reaches two packages a week. This is an explicit fictional arrangement to confirm before promising dates, not an assumption that a freelancer will reserve time for free.

Another $400 covers the month's fixed operating work and costs: five paid pipeline-and-administration hours at $40, plus $200 in other overhead. These hours do not duplicate production. For the four-week illustration, fixed payments are $100 each week. There are no equipment purchases, depreciation, financing or taxes in this finite case.

The eight organizations at the start

Scroll across to compare stages. Use the arrow keys when the table area is focused.

These are eight distinct organizations, counted once each. The accepted customers' previous inquiries do not add two more opportunities. There are no additional leads hiding in the stretch case.

Build scenarios from events, not confidence adjectives

The booked floor contains the two accepted jobs: $1,200 of sales if delivered as agreed. It describes the current order base. It is not an unconditional guarantee against cancellation, production failure or a changed agreement.

The working case delivers those two in week one, one proposal in week two and one early prospect in each of weeks three and four. That is five packages and $3,000 of sales. The three new sales are explicit assumptions, each dependent on the required steps happening in time.

The stretch case delivers all eight organizations' packages: $4,800. That requires two jobs every week, including the two already accepted for week one. The other six must become ready early enough to use the remaining slots. “They all buy eventually” would not be enough to put them in this month.

With no relevant history, I would not assign a proposal a 60% chance merely because a spreadsheet has a probability column. The number needs a basis. A scenario can state “one of these three proposals is accepted in time for week two” without claiming a measured one-in-three conversion rate.

As evidence accumulates, preserve which opportunities were actually at each stage, how long decisions took and what happened. An average drawn from unrelated offers or a different buying process can make a forecast appear more informed while making it less useful.

Let the calendar reject an impossible forecast

In this studio, approved inputs must arrive by Monday before a promised production week. Check the customer's review process, the studio's capacity and the intended delivery date together. A Friday acceptance cannot quietly become Monday's finished presentation if the work needs four hours that have already been sold.

Suppose all six unaccepted opportunities remain unresolved until week four. There may be six eager buyers, but only two packages can be delivered that week. Together with the two week-one jobs, the month can contain four completed packages and $2,400 of sales under the stated resources.

The other four would need later dates, different confirmed capacity or a declined offer. The studio must agree the actual schedule before accepting the new work; this is not permission to move already-promised deliveries unilaterally. More demand and more deliverable sales are different facts.

Do not repair the spreadsheet by assuming that the founder works all weekend. If additional production is a real option, price the paid hours, verify availability, check quality control and make a separate scenario. Otherwise the forecast has acquired an invisible worker.

Put collections beside sales without merging them

The working case assumes each new customer pays a $300 deposit on acceptance, receives the finished package that same week after timely approved inputs, and pays the remaining $300 the following week. All five jobs are delivered by the end of week four. There are no refunds or late payments in this case.

Opening cash is $1,600: $1,000 of the owner's money and the $600 of deposits already received for the two accepted jobs. Those deposits are in the bank before the forecast begins. They cannot also appear as new week-one receipts.

Week one completes the two accepted jobs, earning $1,200. Their $600 remaining balance arrives in week two. The next three jobs each bring a deposit in their delivery week and a balance the following week. That leaves the final job's $300 balance to arrive in week five.

The working case over four weeks

Scroll across to compare earned sales, receipts and ending cash. Use the arrow keys when the table area is focused.

The month earns $3,000. Its job costs are $1,000 and fixed costs $400, leaving $1,600 before tax and financing. But current-month receipts are only $2,100. After $1,400 of payments, cash rises by $700, from $1,600 to $2,300.

The difference is explainable: $600 was collected before the month, while another $300 remains receivable afterward. Every delivered package has been earned in full under the case's stated completion assumption. There is no deposit for unfinished work left at month-end.

Even the $1,100 week-one ending balance does not prove the lowest daily balance. The actual payment dates within that week still matter. Here it is a weekly planning view, not a claim that a detailed cash test has been passed.

In the booked-floor case with no further orders, the two jobs cost $400 and fixed costs still total $400. Only the $600 remaining balance is collected during the month. Ending cash is therefore $1,400, with $400 of operating profit. That is a very different plan from treating all eight inquiries as completed sales.

Update the forecast without erasing the old one

Keep an original forecast date and a separate revised version. When a proposal slips, record whether the reason was a customer decision, missing inputs, an inaccurate workload estimate or the studio's own capacity. Each explanation suggests a different response.

If the work regularly takes six hours instead of four, the unit needs repricing or rescoping. If approval takes three weeks, moving every proposal into next week's column is not a forecast improvement. If several prospects are waiting for the same purchasing meeting, they may be exposed to one shared delay rather than six independent opportunities.

Useful variance notes are short and concrete: “Customer code C has not supplied approved material; no production date accepted.” Avoid “sales underperformed” when the evidence says the delivery prerequisite never happened.

The first-paying-customer guide helps turn an untested offer into a real purchase. Once the order base grows, the break-even guide asks how much volume the business needs and whether collections arrive in time. Neither substitutes for constructing the initial sales assumptions.

Keep a forecast record someone else can challenge

1. Define the unit. Record scope, price, required customer inputs, paid work and delivery time.
2. List distinct opportunities. Separate accepted orders, outstanding proposals and early conversations. Show the next event for each.
3. State the three cases. Identify which actual opportunities supply the booked floor, working case and stretch case. Do not invent orders to fill the desired revenue number.
4. Check the sequence. Match decision dates, inputs and available paid capacity before assigning a delivery week.
5. Translate sales into receipts. Keep money already collected in opening cash and unpaid balances on their expected dates.
6. Preserve and revise. Compare actual events with the dated original. Change the assumptions that evidence has challenged.

The point is not to make uncertainty disappear. It is to discover whether an attractive month requires six customers to decide on time, a supplier to offer unavailable hours or a deposit to be counted twice. Those are questions you can act on.

Infinite Upside separates the presence of possible customers from the harder chain of winning their business, charging enough and delivering what the operation needs to provide. That makes it a useful companion when a first forecast begins to feel more certain than its assumptions. Read the opening for the wider ownership story behind those decisions.