How to calculate your win rate honestly.

What is win rate?

Win rate is the percentage of closed sales opportunities that end in a won deal. Salesforce's own definition states it as "the percentage of qualified opportunities your team converts into closed business," and the formula it publishes is Won Opportunities divided by Total Opportunities, multiplied by 100. HubSpot's sales team defines it the same way from the other direction: "the percentage of final stage prospects that closed and became customers divided by the total number of deals in a given period."

Both point at the same idea. Win rate isn't asking how many leads you started with. It's asking, of the opportunities that reached a real decision, how many did you convert. That's what separates it from a broader top-of-funnel conversion rate, which Salesforce calculates as won deals divided by total leads, a much larger and more forgiving denominator.

The formula, and the choice hiding inside it

The standard version:

Win rate = (Won opportunities / Total closed opportunities) x 100

Gong's own documentation for its win/loss analytics uses the identical formula, written as won deals divided by (won deals plus lost deals). That "plus lost deals" is doing more work than it looks like. Total closed opportunities means won deals plus lost deals, not won deals plus everything still open in the pipeline. Open, in-progress deals are not part of either side of the equation, because they haven't reached an outcome yet. A pipeline full of stalled deals that never get marked lost will quietly inflate win rate for as long as those deals sit open and uncounted, which is exactly why CRM hygiene and win rate accuracy are the same problem wearing two names.

Where the real choice sits is in how "lost" gets defined, and that's the next question, because most teams have never written the answer down.

Does a "no decision" count as a loss?

Yes, if you want the number to mean what people assume it means. HubSpot's own guidance names this directly as a decision a company has to make, not a default: some companies count only prospects that chose a competitor as a loss, and treat a prospect who evaluated, saw pricing, and then did nothing as excluded from the calculation entirely. Others count that same outcome as a loss. HubSpot's stated view: "the key takeaway here is to be consistent in which accounts are and are not included in your win rate calculation."

The size of that choice is bigger than most sales leaders assume. Matthew Dixon and Ted McKenna, in a study spanning more than 2.5 million recorded sales conversations across transactional and complex sales, found that "anywhere between 40% and 60% of deals today end up lost to customers who express their intent to purchase, but ultimately fail to act," rather than lost to a named competitor. If your calculation quietly excludes every no-decision outcome from the denominator, you are not measuring how often you beat the field. You're measuring how often you beat the field among the smaller group of prospects who happened to also make a decision, and that group can be less than half the deals you actually ran.

Neither choice is dishonest on its own. What's dishonest is switching between them depending on which quarter needs a better story.

Which opportunities count as the denominator?

This is where win rate gets flattered most often, and it doesn't require touching a single outcome. It only requires moving the starting line.

Compare two ways of drawing the denominator on the same pipeline:

• Every opportunity that was ever created, from first qualification through to close. This counts every deal that fell out early, which is most of them for most teams.
• Only opportunities that reached a late stage (proposal sent, verbal commitment, whatever your last stage before closing is called). This quietly deletes every deal that died before it got that far.

The second version will almost always report a higher win rate, because the deals most likely to lose are usually the ones that die earliest and never reach a late stage. That's not a coincidence. Deals stall early precisely because something is already wrong: no budget, no real problem, no economic buyer. A denominator that only starts counting once a deal has survived past those failure points is a denominator built to flatter.

A worked example on one set of numbers

Take 200 opportunities that reached full qualification in a quarter. Of those, 50 closed won, 90 closed lost to a named competitor or on price, and 60 ended with the prospect going quiet after a proposal, no purchase from anyone.

Neither 35.7% nor 25% is the "real" win rate. They're both true, honest answers to two different questions. The problem starts when a board pack reports whichever one is higher that quarter without saying which question it's answering, and a leadership team ends up comparing this quarter's 35.7% (no-decision excluded) against last quarter's 25% (no-decision included) and concluding performance improved when nothing about the underlying deals changed at all.

New logo and expansion are not the same number

Selling to a brand-new account and selling more into an existing customer are different motions with different win rates, and reporting one blended figure hides whichever one is actually struggling. Ebsta and Pavilion's 2025 GTM Benchmarks Report, built from an analysis of 655,000 opportunities worth $48 billion in pipeline value across 387 companies, found a new-logo win rate of 19% against an expansion win rate of 45% for the same period, a gap of 26 points.

A single blended win rate across both motions can sit anywhere between those two figures depending on the mix of new-logo versus expansion deals a company happens to run, which means the same blended number can describe two teams in completely different health. Reporting the two separately is what actually tells a board where to look.

How many deals do you need before the number means anything?

Win rate calculated on a handful of deals is mostly noise. Gong's own documentation for its win/loss analytics, a feature available on its Forecast plans, states that "at least 100 qualifying closed deals are needed within the selected time range to present insights," and that the more deals available, the better the insights become. Below that volume, a single unusual deal, a big enterprise account that happened to close or a stalled deal that happened to get marked lost, can swing the percentage by several points on its own, and a swing driven by one deal isn't a trend.

A team closing 15 deals a month that wants a stable monthly win rate is really asking for a number built on too few outcomes to be reliable. The honest fix is widening the window (a rolling quarter instead of a rolling month) rather than reporting a monthly figure that bounces around on small-sample noise and reacting to every bounce as if it were a real signal.

Common mistakes that flatter the number

• Counting only late-stage deals in the denominator, which deletes the early failures and reports a win rate that describes survivors, not the full pipeline. See the denominator section above.
• Silently excluding no-decision outcomes, when 40% to 60% of deals in Dixon and McKenna's research ended exactly that way. If the exclusion isn't stated next to the number, the number is unclear about what it measures.
• Blending new-business and expansion win rates into one figure that can sit anywhere between two very different underlying rates, as shown above.
• Reporting a period too short to be stable. Below Gong's own 100-deal threshold, month-to-month swings are closer to noise than signal.
• Changing the definition mid-quarter. A quarter that started counting no-decision as excluded and finished counting it as a loss (or the reverse) doesn't have one win rate for the quarter. It has two incompatible ones stitched together and reported as if they were the same measurement.
• Leaving stalled deals open instead of marking them lost. An opportunity that should have closed months ago but is still sitting "open" in the CRM is invisible to both sides of the win rate formula until someone closes it out, which is a CRM hygiene problem wearing a metrics costume.

Which version belongs in a board pack

State three things next to the number, every time it's reported: the time period, whether no-decision counts as a loss, and whether new-logo and expansion are shown separately or blended. A win rate with those three facts attached is a number a board can actually act on. A win rate without them is a number that changed for reasons nobody in the room can name.

If the underlying discipline is missing (deals sitting open past their close date, no consistent stage where "opportunity" starts, no written rule for no-decision), the win rate itself is the wrong thing to fix first. Fixing what feeds it, starting with exit criteria on every pipeline stage and a CRM decay rate worth trusting, is what makes the number stable enough to report at all.

Winning back the honesty in this one number usually surfaces a handful of pipeline and CRM problems that were hiding behind it.

If you want a second pair of eyes on your own setup, Salestruct runs a free diagnostic.