How to Forecast Revenue When You Barely Have Any History

Forecasting feels impossible with limited history, so most small businesses either skip it or produce a number by multiplying last month by something optimistic.

Both leave you unable to answer the question forecasting exists for: can I commit to this cost, and when will I be able to pay for it?

The method below works with as little as a few months of data, because it's built from what you already know rather than from statistical history.

Three layers

Build the forecast in layers of decreasing certainty. This is the whole method.

Layer 1 — Committed

Revenue already contracted: retainers, signed projects, work in progress with agreed terms.

This is near-certain and it's the foundation. Map it to the month you'll actually be paid, not the month you'll deliver — those differ, and the difference is what makes cash planning work.

If committed revenue covers your fixed costs, the business is safe this month regardless of what else happens. That single comparison is worth the exercise on its own.

Layer 2 — Pipeline, weighted

Open opportunities, weighted by the probability of closing.

Use your own close rates by stage, not template percentages. Once you have twenty closed deals, calculate what actually happened:

Of deals that reached "proposal sent", how many closed?

That number is yours and it will differ from any template. Before you have twenty deals, don't weight at all — just list the opportunities and their values, and treat the total as a range rather than a number.

Proposal sent      £24,000  ×  0.45  =  £10,800
Qualified          £31,000  ×  0.20  =   £6,200
Engaged             £8,000  ×  0.05  =     £400
                                        ────────
Weighted pipeline                       £17,400

Apply the expected close date, not today. A deal closing in November with 30-day terms is December revenue, not November.

Layer 3 — Not yet sourced

Everything beyond the pipeline. This is where forecasts become fiction, so treat it differently.

Derive it from your own activity ratios rather than a target:

165 approaches → 50 conversations → 25 proposals → 10 customers

If you know your ratios from the pipeline, you can forecast this layer from planned activity. If you don't, mark it as a range and label it explicitly as unsourced.

Never present layer 3 with the same confidence as layer 1. Most forecasting failures are someone treating an aspiration as a projection.

Respect the sales cycle

You can forecast roughly one sales cycle ahead with confidence, and not much further.

If your average deal takes six weeks from first conversation to signature, you can see about six weeks out. Anything beyond that depends on pipeline you haven't created yet — which is a prospecting question rather than a forecasting one.

That's a useful reframe. If your quarter looks thin two months out, the answer isn't a better forecast. It's activity now, because the work you start today is what lands then.

Three cases, not one

A single number invites false precision. Produce three:

| Case | Built from | Use it for | |---|---|---| | Low | Committed only, plus the pipeline you'd bet on | Committing fixed costs | | Expected | Committed + weighted pipeline | Planning and decisions | | High | Everything closing plus some unsourced | Capacity planning only |

Commit fixed costs against the low case. That's the discipline that keeps a business solvent through a bad quarter — and it's the same logic as building a budget against conservative revenue.

Treat the gap between low and expected as decisions you make when the money arrives, not spending you've already assumed.

Check it against actuals

The step that turns forecasting from an exercise into a skill.

Every month, record what you forecast and what happened:

         FORECAST   ACTUAL   VARIANCE   %
Sep      £42,000   £38,400   −£3,600   −8.6%
Oct      £45,000   £47,200   +£2,200   +4.9%
Nov      £48,000   £39,100   −£8,900  −18.5%

After three or four months the useful thing appears: the bias. Almost everyone is consistently optimistic, usually by a fairly stable amount — and once you know yours, you can correct for it.

If you're 15% over every month, your forecast is genuinely useful once you multiply by 0.85. That correction is worth more than any improvement to the method.

Also look at where the misses come from. Consistent misses in layer 2 mean your close rates are wrong. Misses in layer 3 mean you're planning activity you don't do.

Seasonality, once you can see it

With two years of data you can spot patterns. With less, ask two questions instead:

• Do my clients have a slow period? Their seasonality becomes yours with a lag.
• Does my own capacity drop predictably? Holidays, quiet months.

Adjust the forecast for both rather than assuming a flat line. And note them as they happen, so year two has real data instead of memory.

The mistakes

1. One number instead of three cases. False precision.
2. Template close probabilities. A guess with a decimal point.
3. Forecasting revenue at delivery date, not payment date. Fine for planning, useless for cash.
4. Presenting unsourced revenue as pipeline. The main way forecasts mislead.
5. Never comparing to actuals. You never learn your own bias.
6. Forecasting further than one sales cycle. Beyond that it's a prospecting plan.

What to do next

List your committed revenue for the next three months and compare it to your fixed costs. If committed covers fixed, you know the business survives those months regardless of what else closes — and that's the single most reassuring number available.

Then start recording forecast against actual every month. Three months from now you'll know your own bias, which is where forecasting starts being useful.