Sales Forecasting: How to Build a Forecast You Can Run the Company On

If you're like the CEOs who call me, you don't have a problem closing deals. Your team can't close one without you in the meeting. The forecast is a guess. Every deal is closing next month, the same ones as last month.

The number you'd bet on isn't in the CRM. It's in your head, because the deals you're sure of are the ones you're in.

My rule: a forecast is only as good as the last thing the buyer did on each deal. Below is how to build one you can run the company on, with the data you already have: what sales forecasting is, the methods a company under $15M can run with a few reps and a year or two of CRM data, the formula, and how to replace the close date a rep typed with something the buyer did. Part of the complete guide to sales strategy consulting →

I've had a quota on me for 27 years and never missed, most of it in leadership. I've built or rebuilt the sales organization at every company that's employed me.

What Is Sales Forecasting?

Sales forecasting is estimating future sales in a set period, usually a month, a quarter or a year, from past sales data and the deals your reps are working now. The sales forecast is the number that comes out: the expected revenue, and when it books.

Most guides, written for sales leaders and the finance team, define sales forecasting by its methods and its benefits. I'd define a sales forecast by whether you'd run payroll on it.

Sales Forecast vs. Sales Plan vs. Pipeline

The sales plan sets the revenue target for the year, realistic sales targets by rep, and who chases them; the quotas in the sales comp plan add up to it. The sales pipeline is every open deal, whatever its odds. The sales forecast is the part of the pipeline you expect to close by a date, and deciding which part is the forecasting process. The plan is the target; the forecast is the call, and somebody answers for it when it misses.

Why Sales Forecasting Matters at a Company Under $15M

At your size, accurate sales forecasting is cash flow planning: it decides whether you make a hire, sign a lease or hold your nerve through a slow month, without waiting to see what lands. One founder put it to me as not having enough in the bank to stay calm through two bad weeks of sales.

Why Your Sales Forecast Is a Guess

If there's no single sales process, everyone sells their own way, and the forecast is a guess. CEOs describe the first two signs below in their own words; the rest are the common sales forecasting challenges I see on first calls:

• The reps have no idea if their deals will close. They're talking to the wrong people or pitching the wrong product, and their sales managers aren't much better.
• You missed your sales targets two quarters running, and I'd bet the forecast said you'd make both.
• The forecast changes when you ask a rep "are you sure?" Nothing in the deal says the buyer agreed to the close date. It's the one the rep typed.
• The deal doesn't move; the close date does. Count how many of this month's deals were in last month's forecast.
• The probability column was set once and never checked. Somebody put percentages on the sales stages the day the CRM went in, say 10, 25, 50, 75 and 90, and nobody has held them against the deals that closed since.
• The forecast is blind to your deals. The ones you're working aren't in the CRM, so Monday's number leaves them out.
• The CRM gets updated the night before the forecast meeting. So the forecast reflects what the reps remembered, not what happened.

Deals that still run through you get their own section below.

Sales Forecasting Methods, and Which One a Company Under $15M Can Run

With one to five sales reps and a year or two of CRM data, the sales forecasting method I'd start with is opportunity-stage forecasting, with the stages defined by what the buyer did, checked against sales cycle length. Those two use data you already have: your open deals and your closed ones.

Most guides list more sales forecasting methods than that. Opportunity-stage forecasting weights each deal by the odds of its stage. Lead-value forecasting does the same by lead source. Sales-cycle forecasting weights a deal by how old it is against your average cycle. Historical or time-series forecasting projects past performance forward and adjusts for growth, seasonality and market trends. Regression and multivariable analysis weight several factors at once, market conditions included. Rep-based, or qualitative, forecasting is each rep's judgment on their own deals. Predictive forecasting is software reading your CRM data and buyer behavior to score each deal. Top-down against bottom-up is the company target broken down, or the sales quotas added up.

Most of the other sales forecasting methods need historical data you don't have yet, or combine methods you haven't built. Regression and predictive models want years of consistent sales history. Historical forecasting assumes next year looks like last year, which at a company growing from $3M to $5M it won't, and a market shift makes last year's trend irrelevant faster still. Rep-based forecasting is the one the CEOs who call me already have.

How to Calculate a Sales Forecast

The sales forecasting formula for an opportunity-stage forecast is each open deal's value multiplied by the win rate of its stage, added up. The win rates come from your own historical sales data, the deals you closed and lost, not from a default in the software.

Say you have 20 open deals at $60,000 each: $1.2 million of pipeline. Last year, of the deals that reached a proposal, half closed. Of the deals that reached a first meeting, one in four closed. Eight deals are at proposal today: $480,000 times 50% is $240,000. Twelve are at a first meeting: $720,000 times 25% is $180,000. The forecast is $420,000.

Your reps' forecast on the same 20 deals is $1.2 million, because every one of them is closing next month. The $780,000 between those two numbers is a hire you make in the spring against revenue that was never coming.

That $420,000 holds only if "proposal" means something. If "proposal" means the rep sent a PDF, the 50% won't hold, because a rep can send a PDF to anyone, and the stage fills with deals that were never real. If "proposal" means the buyer named the budget and the date they'll decide by, I'd trust the 50%, because the buyer put the deal there, not the rep.

Then combine methods: check the forecast against your sales cycle. Say a deal takes 90 days from first meeting to close on average. A deal that's been at first meeting for 200 days isn't in this quarter's forecast whatever the stage says. Deals past twice the average cycle come out, or get a conversation.

How to Build a Sales Forecast Your Team Can Run, Step by Step

To build an accurate sales forecast, decide who owns it, define each stage by what the buyer did, pull the win rates and sales cycle from last year's sales data, tie every close date to a buyer's dated step, have the rep call each deal and the manager inspect it, build it into your CRM, run it the same way every week, and check it against what closed after a quarter.

1. Decide who owns it. Whoever runs sales owns the sales forecasting process, and you read it. In my experience, under $15M there's rarely a sales operations team to hand it to. If that's still you, start with who should own it, below.
2. Define each stage by what the buyer did. Not "demo scheduled" but "the buyer brought the person who signs." Not "proposal sent" but "the buyer named the budget and the decision date." The buyer moves the deal; the rep writes it down.
3. Pull the win rate and the sales cycle from last year's closed and lost deals. For each stage, how many deals that reached it went on to close, and how long they took. Those are your probabilities and your cycle length. If you don't have a year, use what you have and say so in the forecast.
4. Tie every close date to a buyer's dated step. A close date is the date the buyer said they'd decide, or the date of the next step the buyer agreed to, plus your average time from that step to signature. A close date with nothing behind it is the rep's hope.
5. Have the rep call it, and the manager inspect it. Each rep marks every deal commit, best case or pipeline. The manager asks the same questions of every deal: what did the buyer do last, what did they agree to next, and when. A rep who can't answer moves the deal down.
6. Build it into your CRM. The stage criteria, the probabilities and the categories live in the CRM, every rep fills in the same fields the same way so the CRM data is standardized and tracks every deal stage, and the forecast is whatever the CRM says on Monday morning. I've watched teams ignore a process that only lives in a document, and a forecast kept in a spreadsheet is a document.
7. Run it the same way every week. Whoever owns it holds one meeting with the reps and goes through every deal that changed. You read what comes out of it; you're not in it.
8. Check it after a quarter, then every month. Hold the commit from the first week of the quarter against what closed by the last. That's your forecast accuracy, and the first number to read before you change anything.

The Deals That Still Need You

If you're the one closing, half the forecast may be your calendar. Say your sales plan is 50 deals this year and 25 of them close only once you're on the call. Twenty-five deals are you, and the CRM doesn't know it, so when your week fills up the forecast slips with it, and the report blames the deals.

The rules I'd start with: every deal you're in goes in the CRM with you on the opportunity team. Its close date is the buyer's decision date, the one you and the buyer agreed on, never the date the rep typed. Whoever runs the forecast call asks the same questions of your deals as of everyone else's, and you answer them in the CRM beforehand. On those, you're the rep.

Then read two numbers every week: the deals that closed without you on the call, and the deals that have slipped once already.

CEOs tell me on first calls they thought they'd hired reps who could close. Founder-led sales → The deals that close without you are how you find out whether that's changing. Some deals will need you while the team learns, and I'd expect that count to shrink. If it's the same count a year later, the forecast isn't the problem.

A sales plan your team can't run without you on the call is a forecast with your name on it. How to work the plan back to the deals: Sales plan →

Who Should Own Your Sales Forecast

Whoever runs sales owns it, and answers for the number when it misses. Seven ways to fill the role:

You. It's right if you have the hours to inspect every deal every week and chase the ones that slipped. Most CEOs who call me don't.

Your sales manager, or a rep you promote. Pick this when someone on your team already runs the pipeline review and wants the job. Every great rep doesn't make a great leader, and every great leader isn't always your best rep.

A manager who carries deals and forecasts them with nobody looking is grading their own paper. If your manager runs deals, you're the sales manager on those: before you believe the number, ask the question the manager asks the reps. What did the buyer do last?

Sales forecasting software. Sales forecasting tools read your CRM, score each deal and streamline the reporting, and some are AI-powered, using machine learning on the reps' email and calendar to predict close rates. They work from the stages you defined, which they won't define for you, so a date the rep typed scores as a date. Add tools only after the process works.

In my experience, a common fix is a new CRM, to see what's happening without being on every call. Too often it gets botched, and you end up running two: the old one and the new one.

A sales consulting firm. Right when the stages and the CRM need rebuilding and someone inside will run the forecast call once it's handed back. In this market, a typical engagement is a 90-day sprint: the audit, the playbook and a CRM rebuild, with the stages and the forecast inside it, returned to the same team that was selling before. Nobody from the firm stays to run the forecast call. The sales strategy decisions around the forecast: The complete guide to sales strategy consulting →

A full-time VP of sales. Worth it when the sales process already works, so the person you hire runs the forecast instead of inventing it, and when you need someone in the building every day, committed for years.

In this market, at a company under $15M, that's $360,000 to $520,000 a year fully loaded, plus equity, after a search of three to six months. Fractional vs. Full-Time →

A fractional VP of sales. Right when nobody inside can run sales and you're still the one closing. A fractional VP of sales is a sales executive you bring in instead of hiring full-time. In this market, most are sold by the hours a week, at $4,000 to $15,000 a month.

Done my way, the forecast is mine: the stages a deal moves through, what has to be true before it moves, and how each deal gets reviewed, so your reps know which deals will close, and you can see it without asking. I run the weekly forecast call with your team, and you read the number without being in the room.

I'm rarely the first fix a CEO has tried. Usually it's been one train wreck after another. This time it's one person: me, with no associates and no bench. The person on your first call is the one on your forecast call every week. Your sales team reports to me, I report to you, and your head of marketing stays my peer.

The new revenue the team closes is my number, and I take no accounts of my own, so deals don't start waiting on me instead of you. When the number misses, you hear it from me first, with the reason and what changes next month.

Since 2020, averaged across my fractional clients: three times as many leads turning into opportunities, and two to six times the pipeline.

From month one it's aimed at one of two endings: someone on your team grows into leading sales, or you hire a full-time VP of sales into a process that already works. I help write the job description and sit in on the interviews.

Mine is $5,000 to $7,000 a month, remote, month to month, with no minimum term and no notice period: if it isn't working, you stop at the end of the month. That starts below the market's $6,000 to $10,000 middle, and your number is mine, no excuses: Fractional VP of Sales →

Advisory. If someone on your team can run the forecast once they know what to ask, you need advice, not a leader. In Advisory, I tell you what to fix, in what order and why, and your team builds it; nothing reports to me, and I answer to you. It's $3,000 to $4,000 a month, remote: Advisory →

How to Tell Your Sales Forecast Is Working

Accurate forecasts land where the commit said, and fewer of the deals in them need you. Read these every month with whoever runs your sales team:

• Forecast accuracy. The commit at the start of the period against what closed. Forecast accuracy is one minus the miss divided by what closed, as a percentage; a $420,000 commit that closes $400,000 missed by $20,000, or 5% of what closed, so it was 95% accurate. Bigger companies track it as absolute percentage error and forecast bias; that one number, and which direction it misses in, do the job at your size. I'd want the commit within 5 to 10% after two quarters on the new stages, and I'd read a forecast that's always high as a stage problem, not a rep problem.
• Deals that slipped. How many close dates moved this month, and how many moved for the second time. Take this count before the new stages go in.
• Deals closed without you on the call. The number from the section above, and the only one of the six that tells you whether you hired reps who can close.
• Win rate by stage against last year's. If deals that reach proposal are closing at a quarter instead of half, either the stage definition slipped or market conditions did, and the forecast has to move with it.
• Pipeline coverage. Open pipeline against the number you need to close. Three times the number is a common rule. At the one-in-four win rate from the example above, three times the number closes three-quarters of it, so I'd set the ratio from your own win rate before trusting the rule. Pipeline management is a different job from forecasting, but a thin pipeline today is next quarter's miss.
• Sales cycle length. If it's stretching, deals are sitting at a stage past the buyer's date, and next quarter's forecast is already high.

Sales teams don't improve overnight, so give the new stages a full quarter before you judge the forecast against them.

How Often Should You Update a Sales Forecast?

Every week, in the same meeting, deal by deal, run by whoever owns the forecast. You read what comes out of it. Check the forecast accuracy every month. Re-pull the win rates and the sales cycle from closed deals every quarter, so the probabilities keep up with how your team sells now.

Change a stage definition when you find one a rep can move a deal into without the buyer doing anything. Apply the new definition to new deals, never mid-quarter to the deals already in it.

The first quarter a reliable forecast holds, make the hire on it and start next year's sales plan and sales targets from it. That's running the company on the forecast instead of on your calendar.