The Startup Sales Forecast: A Founder's Guide to a Bottom-Up Subscription Model
A sales forecast is not a wish. It is a set of assumptions, connected by arithmetic, that shows how your business turns customers into revenue over time. For subscription businesses, the standard template is a monthly cohort model with eleven rows: ARPU, starting subscribers, new subscribers, cancellations, net additions, ending subscribers, churn rate, projected lifetime, LTV, and total MRR. This guide walks through each row, the formulas that connect them, and the traps that make founder forecasts unbelievable to investors.
Top-down forecasts start with a market size and claim a percentage. "The market is $10B; if we capture 1% we hit $100M." Investors have seen this slide ten thousand times and it persuades no one, because 1% of anything is not a plan. A bottom-up forecast starts with the smallest unit you actually control — one subscriber, paying one price, for one month — and stacks it into a revenue curve. It forces you to name the mechanism: how many new customers per month, at what price, with what churn. If any of those numbers is wrong, an investor can point at the row and tell you why. That is the entire purpose of the model.
ARPU is the monthly revenue you collect per active subscriber, blended across plans. In the template it is a flat $20. In your model it should reflect the actual mix of plans you sell today. If you have a $19 monthly plan and a $190 annual plan (which is $15.83/month amortized), blend them by expected mix — not by your favorite plan. Founders routinely inflate ARPU by assuming everyone lands on the highest tier. Assume the opposite: model the plan mix you have seen in the last ninety days, then show a second scenario where mix improves.
Hold ARPU flat unless you have a specific reason to raise it — a price increase you have already tested, a new tier launching in a named month, or a contractual annual escalator. Rising ARPU with no cause is the first place investors lose trust.
Starting subscribers in any month equals ending subscribers from the previous month. In the template this is expressed as a chain — each month reaches back to the prior month''s ending count. This is a chain, and it means an error in month 3 propagates through month 12. Lock the formulas. Do not overwrite starting subscribers with a hardcoded number "to make the chart look better."
This is the row investors interrogate hardest, because it is where founders hide their optimism. "500 new subscribers per month" is not an assumption — it is a claim about your acquisition engine. To defend it you need to show the inputs underneath: traffic × conversion × trial-to-paid, or leads × close rate, or SEO impressions × CTR × signup rate × paid rate. Whatever your channels are, the new-subscriber number must be the output of a channel model, not a guess.
If you are pre-revenue, do not project 500 net-new in month 1. Ramp in. A believable early curve starts small (25–75 in month 1), grows as content compounds or ads scale, and reaches a steady-state driven by budget or capacity. Investors trust ramps. They do not trust flat lines that begin at your target.
Cancellations are the counterpart to new subscribers and they determine whether you have a business or a bucket with a hole. In the template, cancellations grow from 10 in month 1 to 95 in month 11 — roughly tracking the ending subscriber base. That is the correct instinct: absolute cancellations rise as the base rises, even if the rate stays constant.
Model cancellations as a rate applied to starting subscribers, not as an absolute number, unless you have contractual visibility (annual contracts renewing in a specific month). Rate-based modeling forces you to state your churn assumption explicitly, and churn is the single most-scrutinized number in a subscription pitch.
Net Additions equals New Subscribers minus Cancellations. This row exists to make the trend obvious at a glance. If net additions turn negative in any month, you are shrinking, and no amount of ARPU growth will save the forecast. Investors will ask what changes in that month to reverse it. Have an answer.
Ending Subscribers equals Starting Subscribers plus Net Additions. Simple, but this is the number that flows into next month''s starting count and into the MRR calculation. Sanity-check the endpoint: does month 12 ending subscribers imply a market share you can defend? If your model shows 50,000 paying users in a market of 200,000 SMBs, expect to be asked how you took a quarter of the market in a year.
The template computes churn as Cancellations divided by Starting Subscribers. This is monthly gross customer churn. Two things matter here.
First, benchmark against your segment. SMB SaaS typically runs 3–5% monthly gross churn. Mid-market runs 1–2%. Enterprise runs below 1% and is usually measured annually. Consumer subscriptions can run 5–10% monthly and still work if payback is fast. If your model implies 1% monthly churn on an SMB product, an investor will assume the number is wrong.
Second, distinguish gross churn (customers who cancel) from net revenue churn (revenue lost after expansion). Best-in-class SaaS shows negative net revenue churn — existing customers spend more over time via upgrades and seat expansion. If you have that dynamic, model it in a separate row for expansion revenue rather than burying it in ARPU.
Lifetime in months equals 1 divided by the churn rate. A 5% monthly churn implies a 20-month lifetime. A 2% monthly churn implies a 50-month lifetime. This inverse relationship is why churn is the highest-leverage number in the model — cutting churn in half doubles lifetime, which doubles LTV.
The formula assumes churn is constant, which is a lie in the first year. New cohorts churn faster than mature cohorts. If you have cohort data, model it. If you do not, use the constant-churn approximation but flag it as conservative and revisit the number every quarter with real data.
LTV equals Lifetime times ARPU. This is gross LTV — revenue, not profit. For an investor conversation you also need contribution-margin LTV, which multiplies by gross margin. A $20 ARPU with 50-month lifetime and 80% gross margin is an $800 contribution LTV. Compare that to your fully-loaded CAC. A healthy ratio is 3:1 or better, with CAC payback under 12 months for SMB and under 24 for enterprise.
If you show LTV without CAC, investors will assume you are hiding CAC. Always pair them.
The template computes MRR as ARPU times the sum of Starting and New subscribers. This is a simplification that slightly overstates MRR because it credits full-month revenue to subscribers who joined mid-month. A cleaner formula is ARPU times Ending Subscribers, which credits revenue only to subscribers who finished the month active. Choose one convention and be consistent — the difference matters when you extend the model to ARR and valuation multiples.
Multiply the month-12 MRR by 12 to get exit-run-rate ARR. This is the number investors will anchor on for valuation, so build the model backward from a defensible ARR target rather than forward from a fantasy new-subscriber number.
Ship the forecast with three columns per month: base, upside, and downside. Base is what you actually believe. Upside is what happens if one channel outperforms. Downside is what happens if churn is 50% worse than assumed. Investors do not fund the base case — they fund founders who have already thought through the downside case and know what they will cut, delay, or renegotiate when it hits. A single-scenario forecast implies you have not thought about the downside, and that is a worse signal than the downside itself.
A one-page summary with the eleven rows above, a chart of MRR by month, a chart of ending subscribers by month, and a footnoted assumptions block explaining ARPU mix, channel-by-channel new subscriber inputs, and the churn benchmark you are targeting. Ship the underlying spreadsheet with formulas intact so the investor can flex assumptions themselves. Locked PDFs signal that you do not want the model examined. Open spreadsheets signal confidence.
A forecast is not a promise. It is a testable claim. Build it so every number can be traced to an assumption, every assumption can be traced to a benchmark or a data point, and every scenario has a plan attached. That is the model investors fund.
