How to Calculate Sales Close Rate and the Mistake Most Teams Make

Sales close rate is deals won divided by the opportunities that reached a decision — won plus lost — measured on the cohort those deals came from. Get the denominator wrong and the same 76 wins report as 6.3%, 15%, 25.3%, 38% or 40%. Same team, same deals, five answers.

How do you calculate sales close rate?

The formula has three parts, and only the first is the one people argue about:

Close rate = deals won ÷ (deals won + deals lost), counted on a defined population of opportunities, measured after that population has had time to decide.

• Numerator: opportunities marked closed-won in the period.
• Denominator: opportunities from the same population that reached a terminal outcome — closed-won or closed-lost. Not opportunities still open.
• Population: a stated stage. “Enquiries”, “qualified opportunities” and “sales conversations attended” are three different populations and give three different rates.
• Clock: the population is defined by when it was created, and reported only once it is old enough to have resolved.

This matches how the major CRMs define it. Salesforce’s own reporting documentation gives the win-rate summary formula as WON:SUM / CLOSED:SUM — won opportunities over closed opportunities, with open pipeline excluded from both sides (Calculate Win Rate on Closed Opportunities in a Report, Salesforce Help). Most spreadsheets built by hand do not do this.

A close rate without a stated denominator, a stated stage and a stated window is not a metric. It is a number someone chose.

The mistake most teams make: open deals in the denominator

The single most common miscount is dividing wins by every opportunity created, including the ones still sitting in the pipeline undecided. This is the open-pipeline error, and it does two bad things at once.

First, it always understates. An open deal is not a loss. Putting it in the denominator books it as one before the prospect has said anything.

Second — and this is the part that causes internal arguments — the number moves when the pipeline grows, even if nothing about selling changes. Add 100 fresh opportunities today and your close rate falls this afternoon. A metric that gets worse when you generate more demand is not measuring sales performance.

Statisticians have a name for the underlying problem: the open deals are censored observations. Their outcome exists but has not been observed yet. The correct handling is to exclude them from the ratio, not to score them as failures.

The Decided-Deals Rule: an open opportunity is neither a win nor a loss. It stays out of the denominator until it decides.

Worked example: how the same 76 deals produce five close rates

Illustrative inputs, not a client account, chosen so you can substitute your own numbers.

Take one month’s demand for a services business. March generated 1,200 enquiries. Of those, 620 were contacted, 300 met the qualification criteria, and 200 attended a sales conversation. By the end of June the March cohort had fully resolved: 76 won, 124 lost, 0 open.

Seventy-six deals. Here is every close rate that can honestly be computed from them:

Note the last two rows against the fourth. At day 30 a little over a third of the cohort had decided, which is what a 45-day median looks like at that point. The decided-only method reported 40.0% when the cohort’s true rate was 38.0%: the 75 deals that decided first won at 40.0%, and the 125 that took longer won at 36.8%. Decided-only is the right method and it is still slightly optimistic early. The all-opportunities method was out by 23 percentage points — it reported less than half the true rate.

Nobody in this example got better or worse at selling. The 6.3% and the 38.0% describe the same 76 deals.

The second mistake: measuring a period instead of a cohort

The other error is subtler and survives longer because it looks reasonable: dividing deals closed this month by opportunities created this month. Those are two different populations, because they are two different months. If your deals take about six weeks to close, this month’s wins came out of a cohort created six weeks ago, and you are dividing one month’s output by a different month’s input.

HubSpot documents the underlying property plainly: Days to close is “the time between Create date and Close date” (HubSpot’s default deal properties). If that median is 45 days, a monthly close rate is comparing wins that came out of one month’s opportunities to the opportunities created six weeks later.

The distortion is exactly calculable, assuming deals close about n months after creation. If the true cohort close rate is r, opportunity creation is growing at g per month, and the lag is n months, then:

Reported period close rate = r ÷ (1 + g)ⁿ

Two conditions, and the second is the whole point of this page. First, deals have to close roughly n months after they are created; where closing times are spread out, the reported rate is r × E[(1 + g)^−L] over the lag distribution L, and the direction of the error is the same. Second, r and the period denominator have to be measured on the same population. If r is your close rate on attended conversations, the period denominator has to be opportunities created that month — attended conversations too. Divide this month’s wins by this month’s enquiries instead and you have stacked the population error on top of the timing one, and the number falls further again: on the cohort above that method reports about 4.5% at +25% monthly growth, against a true enquiry-to-customer rate of 6.3%.

With a true close rate of 38% on attended conversations and a six-week lag (n = 1.5, matching the 45-day median in the worked example), here is what the period method reports while the sales team’s performance is held perfectly constant:

Read the last row. A team whose opportunity creation halves reports 107.5% — an impossible value for wins divided by decided deals, and the tell that the period method was never computing a rate at all — it measures your growth rate wearing a sales metric’s name. This is why scaling businesses conclude their closers have gone soft in the exact quarter they turned up the ad spend, and why a business in decline gets a flattering number on the way down.

Which close rate should you use for which decision?

There is no single correct denominator. There is a correct denominator per decision, and the discipline is stating which one you used every time you say the number out loud.

That last row is not a throwaway. Most published close-rate benchmarks do not disclose their denominator, which makes them unusable for comparison. If you are weighting open deals to build a number for the board, the stage-level decided-outcome rate is the input worth having — more on how that feeds a model in our note on AI sales forecasting and pipeline accuracy.

When is the number old enough to trust?

A cohort reports a truthful close rate only once nearly all of it has decided. The practical test we use:

The Two-Cycle Rule: do not report a cohort’s close rate until the cohort is older than twice your median days-to-close. Before that, report it as provisional with the open count printed next to it.

In the worked example the median was around 45 days, so March’s cohort was reportable from about the end of June — not on 30 April, when it would have read 15% or 40% depending on which mistake you made. The cost of the rule is that your newest and most interesting cohorts are the ones you may not conclude anything from. Inconvenient, and still correct.

The rule also exposes a data-hygiene problem most CRMs have: opportunities that are dead but never marked closed-lost. They sit open forever, they never age out, and they quietly corrupt every close-rate calculation on both methods. Auto-closing anything with no activity for three times your median cycle fixes more measurement problems than any dashboard.

What close rate tells you about your reps, and what it does not

Close rate is the metric most often used to judge a salesperson and most often caused by something the salesperson never touched. If the denominator is enquiries or contacted leads, you are grading a closer on response times, follow-up persistence and show-up rates — three things that belong to lead flow and process. In the example, the entire distance between 6.3% and 38.0% happened before anyone sold anything: 580 enquiries were never contacted, 320 contacted people never qualified, and 100 qualified people never made it into a conversation.

That is why the diagnostic order matters: compute both ends first, and only investigate call technique if the attended-conversation rate is the weak one. Response speed is the best-documented upstream driver — we keep the numbers in our lead response time benchmarks for Australia — and attendance is the other, covered in how to improve sales appointment show rates. Both move the number a manager gets blamed for.

Being explicit about our own claims, since this is a page about measurement honesty: in our own client work at LeadsNow we typically see roughly a 300% lift in conversion for a business moving from 2020-era manual operations to 2026 AI-driven operations, and most of that lift lands upstream of the close — in contact rate and set rate, not in closing skill. Our component figures are the same kind of operator claim: speed to lead alone around 3x, doubling contact rate around 2x, doubling set rate around 2x. Those do not multiply. 3 × 2 × 2 is 12x and we do not see 12x, because the levers overlap — fixing speed to lead is part of how contact rate improves, and contact rate is part of how set rate improves. These are observations from running campaigns, not a study with a sample size and a window. Where we do publish a figure with a stated method, we publish the method: our 7x average sales lift methodology defines the window and discloses that the median is closer to 4x.

What it costs to measure this properly

The method above is complete and you can run it yourself. Here is honestly what it takes, because the effort is the reason most teams do not.

• Five fields, populated every time: opportunity created date, current stage, closed date, outcome (won/lost), and a disqualification reason. Four of these are CRM defaults. The fifth is the one that decays.
• A cohort report, not a period report. HubSpot’s funnel reports count “the number of contacts or deals in each stage that moved through all the selected stages during the specified time range” (HubSpot funnel report types) — useful, but you still have to choose the window deliberately rather than defaulting to “last month”.
• Setup: roughly two to four hours to build the cohort report and the auto-close rule, once. Then about 30 minutes a month.
• The part that breaks: reps marking deals closed-lost. Wins record themselves because they are attached to money. Losses do not, and a pipeline where losses go unrecorded reports a close rate that drifts upward for months.
• The skill: whoever owns the report has to be willing to say “this cohort is not old enough” to a founder who wants the number today.

The alternative to policing this is removing the ambiguity at the source: if the only thing that enters the pipeline is a qualified, attended conversation booked against fixed criteria, the denominator defines itself. That is the model we run — 50,769+ AI-booked sales appointments since 2017 and 1M+ leads generated, two aggregates that we print separately and that nobody should divide by each other, since they span nine years, different clients and different populations (many of those appointments came from clients’ own databases rather than from leads we generated) — on AI appointment setting priced per booked qualified appointment rather than on a retainer, which only works if both sides agree what counts. This page sits in our pipeline-stage series covering each stage from enquiry to close; the close stage is the one where the definition argument is worth having first, because every downstream forecast inherits it.

Frequently asked questions

What is the formula for sales close rate?

Close rate = deals won ÷ (deals won + deals lost), for a stated population of opportunities, measured after that population has had time to resolve. Salesforce’s reporting documentation uses the same denominator, giving the win rate summary formula as WON:SUM / CLOSED:SUM — won over closed, with open opportunities excluded (Salesforce Help).

Should open deals be included in the close rate denominator?

No. An open deal has no outcome yet, so scoring it as a loss understates the rate and makes the number fall every time you add pipeline. Exclude open opportunities from both the numerator and the denominator, and report the open count alongside the rate so nobody mistakes a young cohort for a finished one.

Is close rate the same as win rate?

In most CRMs they are the same calculation, and Salesforce and HubSpot both label it “win rate”. In practice teams use “close rate” more loosely — often meaning enquiry-to-customer across the whole funnel, which in the worked example above is 6.3% rather than 38.0%. If the two words are being used in the same meeting, make someone state the denominator.

Why did my close rate drop when we increased lead volume?

Almost always because you are dividing this month’s wins by the opportunities created this month — two populations separated by a month or more, because this month’s wins came out of an earlier cohort. With a true close rate of 38% on attended conversations, a six-week sales cycle and opportunity creation growing 25% a month, that method reports 27.2% indefinitely while nothing about the sales team has changed. Divide by this month’s enquiries instead and it reads lower again, about 4.5% against a true 6.3%. The formula only works when the rate and the denominator are measured on the same population; switch to a cohort measurement and the effect disappears.

How long should I wait before measuring a cohort’s close rate?

Twice your median days-to-close. HubSpot defines that property as “the time between Create date and Close date” (HubSpot default deal properties), so pull your own median rather than assuming an industry figure. A 45-day median means a cohort is reportable at about 90 days.

Can a close rate be over 100%?

Only if you are computing it wrongly. Dividing deals closed this period by opportunities created this period will exceed 100% when opportunity creation falls sharply — with a 38% true rate, a six-week lag and creation halving each month, the period method reports 107.5%. An impossible number is the clearest available proof that the denominator is not the population the numerator came from.