Most companies still do not track win rate by source

Ask a head of sales for their win rate and you get one number. Roughly one in five. The reported average B2B win rate sits near 21% of deals, rising to about 29% once you count only qualified opportunities.¹ That figure goes on the board pack, gets compared to last quarter, and quietly governs how the pipeline is judged.

The blend is an average of things that should never have been averaged. A referral and a cold-list lead are not two samples of one population. They are different products with different physics. Pool them and you get a number that describes none of them and misleads about all of them.

Most companies still do not break their win rate down by where the lead came from. We deliberately avoid a precise figure for how many, because the percentages that circulate do not trace back to a primary survey we would stake a claim on. The measurement gap itself is real and documented: most marketers cannot attribute revenue cleanly across channels, and the default reporting view stays a single blended win rate. What is not in doubt is the size of the gap the blend conceals.

How big is the conversion spread by source?

Separate leads by origin and the conversion rates do not vary by a few points. They vary by an order of magnitude. In one widely cited dataset of lead-to-qualified conversion by channel, website inbound converted at 31.3% and customer or employee referrals at 24.7%, while cold outbound lists converted at 1.7% and bulk email at 0.9%.²

The same pattern shows up at the deal stage, not just the top of the funnel. A 2026 benchmark study put partner and referral opportunities at a 35% to 55% win rate and self-qualifying inbound demo requests at 30% to 45%, while pure cold outbound landed at 8% to 15%.³ Trust transfers from the referrer. Intent is already present in the inbound request. The cold prospect has neither.

Why is the cheap lead often the expensive one?

Here is where the missing breakdown costs real money. Cost-per-lead tables make outbound and bought lists look like a bargain. But the number that matters is not what a lead costs. It is what an acquired customer costs. Divide the cost per lead by the close rate for that source and you get the true cost per customer.

Work it through. Suppose an Australian services business buys leads from five sources at very different prices, and the close rate for each is the channel benchmark above. The cost-per-lead column looks like a clean ranking. The cost-per-customer column reorders it completely. The lead prices below sit inside published Australian cost-per-lead ranges for services verticals (trades roughly A$40 to A$120, professional services A$80 to A$250).⁴

The A$12 cold lead, the cheapest in the building, costs A$706 to turn into a customer, because you have to buy and process roughly fifty-nine of them to close one. The A$90 inbound lead, more than seven times dearer on the sticker, costs A$288 per customer. The team chasing the lowest cost-per-lead is optimising for the wrong denominator, and a win-rate-by-source view is the only thing that surfaces it. Close rates here are the cited channel figures;² the lead prices are illustrative and sit inside the published Australian ranges.⁴

This is the same idea as Profit Velocity, the rate at which sales effort converts into durable contribution margin, and the metric we build everything around. It rises when you move budget toward the source with the lowest cost per profitable customer. It falls every time you reward a source for being cheap per lead while it quietly burns rep hours on deals that never close.

How much does response speed change the odds?

One reason inbound and referral convert so much better is that they are usually worked faster, and response time has an outsized, well-evidenced effect. The MIT and InsideSales.com Lead Response Management study analysed three years of US data covering more than 15,000 web-generated leads and over 100,000 call attempts. Firms responding within five minutes were about 100 times more likely to connect with a lead and 21 times more likely to qualify it than firms that waited thirty minutes.⁵ Wait an hour and the odds collapse.

If you only track a blended win rate, a slow-response problem on your best inbound source looks identical to a quality problem on a cheap one. The fix and the cost are completely different, and you cannot tell them apart without the breakdown.

How do you start tracking it this quarter?

You do not need a new system to fix this. You need three columns next to every closed deal.

• Stamp the source on every opportunity, once. Lead origin captured at creation and never overwritten by the last-touch channel. Most CRMs already have the field; the discipline is filling it honestly and freezing it.
• Compute win rate and cost per customer per source, not per lead. Cost per lead divided by close rate. Rank sources by cost per profitable customer, then by total margin contributed. The order will not match your cost-per-lead report, and the difference is the point.
• Set source-specific response and effort rules. If inbound converts at 31% and closes faster when worked in minutes, route it to your fastest responder and protect that path. Stop spending equal rep time on a source that closes at under 2%.

The blended win rate will keep telling you the pipeline did roughly one in five, quarter after quarter, while the real story moves underneath it. Split the number by source, read it next to cost per customer and response time, and you stop managing an average and start managing the decision. That is the whole difference between a metric you report and a metric you act on.