Lead Conversion Rate: Formula, Denominators and a Funnel Sheet

Lead conversion rate = customers won ÷ all leads received in the same period × 100. A lead is anyone who gave you contact details and could plausibly buy. A customer won is a lead that has paid within your closing window, the time you allow a lead to buy before you stop counting.

Group leads by the month they arrived (a cohort) because buying takes weeks. If you divide this month's customers won by this month's leads, you mix buyers from old enquiries with leads still in progress. Set a closing window that matches your sales cycle, and wait for it to pass before you report the cohort. For illustration (our assumption, not an industry figure): 30 days for a clinic consultation, 60 to 90 days for property.

Say what counts as converted, because every funnel step has its own rate. First Page Sage's 2026 B2B software data gives 39% from lead to MQL, 38% from MQL to SQL (a sales qualified lead, one that sales has accepted), 42% from SQL to opportunity (a qualified deal sales is actively pursuing) and 37% from opportunity to customer won. Multiplying them, 0.39 × 0.38 × 0.42 × 0.37 = 0.023, so about 2.3% of leads become customers won (our calculation from those four published rates, which cover B2B software only).

Salesforce's State of Sales 2026 report shows how the label changes the answer. Its AI sales agents contacted 1,30,000 leads and created 3,200 opportunities in four months. That is 3,200 ÷ 1,30,000 = 2.46% (our calculation), a lead-to-opportunity rate, not a lead-to-customer-won rate. Write yours the same way: "2% of all leads became customers won within 60 days."

The same leads give different rates because each rate divides by a different group, called the denominator. In this illustration (our model, not market data), a business received 1,000 leads in a month, reached 450 of them, qualified 150, and won 20 customers.

All three are arithmetically correct, and they answer different questions. The first tells you what your lead spend and calling effort produce. The second and third tell you how well you close once a conversation happens. Quote the first one externally, since it is the hardest to flatter, and use the others inside the team to find problems.

Judge your rate against your own earlier cohorts, split by lead source, because published figures count different things. Two examples show the gap: about 2.3% of B2B software leads become customers won (calculated above), while Invoca's 2025 benchmark found 37% of phone leads converting during the call, across more than 60 million calls in nine industries. Neither is wrong, and they are more than ten times apart.

Before using any outside number, ask what it divided by, which lead source it covers and which country and product it describes. This is our recommendation. Referral, portal and ad leads behave differently, so a blended rate moves whenever your source mix moves, even if your team has not changed. Keep one line per source and review every four to six weeks.

Treat small cohorts with care. With 100 leads, one extra customer won moves the rate by a full percentage point.

Find the leaking stage by dividing each step by the one before it. The headline rate equals the product of these stage rates, so the step that has fallen furthest against your earlier results is where to work first.

Use this sheet for one cohort. The example column continues the 1,000-lead illustration above.

Check the working: 0.45 × 0.333 × 0.40 × 0.333 = 0.02, which is 2.0%. The headline must equal the product of the four stage rates. A mismatch means a lead was counted twice or skipped.

Read the sheet with these tests:

• Fail: B ÷ A has fallen against your last cohort from the same source. Fix reaching leads before anything else, because a better pitch cannot help a lead you never spoke to.
• Rule of thumb (ours, not a benchmark): if B ÷ A is below 50% on a source you phone, start with reaching. The 45% in the example would fail this test.
• Fail: C ÷ B is falling while B ÷ A holds steady. Your lead source has got worse, or your qualified test has loosened.
• Fail: D ÷ C has fallen. The next step is too hard to book, so offer a specific slot on the same call.
• Pass: a step matches or beats your last cohort from the same source.

Record each call's result in the same words every time. Our guide to call disposition codes lists outcome labels you can reuse, so steps B and C come from the call log and not from memory.

Call within an hour of the enquiry, and sooner if you can. In a Harvard Business Review study of 2,241 US companies that received a test web enquiry, firms that responded within an hour were nearly seven times as likely to qualify the lead as firms that responded an hour later, and more than 60 times as likely as firms that took 24 hours or longer (Oldroyd, McElheran and Elkington, HBR, March 2011). The scope is US web enquiries in 2011, so read it as evidence that delay costs leads, not as a promise for Indian phone leads.

Our recommendation for retries: make a second attempt the same day at a different hour, and a third the next day, then stop. Place the retries in the hours when most of your calls are answered, which your peak calling hours report shows. Our pages on lead response time and follow-up call scripts cover the first call and the retries in detail.

The arithmetic of reaching leads is simple, and you can run it on your own numbers:

• Dials needed = conversations wanted ÷ answer rate. At a 50% answer rate, 100 conversations need 200 dials (100 ÷ 0.50). At 40%, they need 250 dials (100 ÷ 0.40).
• One attempt per lead caps step B at your answer rate. A team whose calls are answered half the time cannot reach more than about half its leads on a single try.
• If each attempt connected independently 50% of the time, three attempts would reach 87.5% of leads (1 − 0.5³ = 0.875). Real attempts are not independent, so this is an optimistic illustration (our model), but it shows why repeat calls matter.

Our Sales Call Analytics gives you answer rate, missed calls, talk time and peak hours, which feed the reached step. Call notes and AI Call Insights help you judge the qualified step. Visits, demos and customers won you record yourself from your CRM (customer relationship management software) or invoices.

Reports run daily and monthly, and you can download them. The answer rate is the input for the dials formula above, and peak hours tell you when to schedule retries. Our Missed Call Automation is designed to call the customer back automatically after a missed call, so a retry no longer depends on someone remembering.

Every call made through our Business Dialer is logged with the contact, time, duration and outcome, and you can add a note to the call. AI Call Insights flags buying intent, objections, pricing discussions and follow-up commitments in recorded calls. Recordings begin with an announcement to the caller.

As of October 2026, our Pro plan costs ₹999 + GST (goods and services tax) a month with three hours of talktime a day, and extra talktime is ₹49 + GST an hour. Paid plans include unlimited users and contacts, and you can cancel anytime. To see how your team's calls would map onto this sheet, talk to our expert.

Run it in seven days by fixing definitions first, then counting a past cohort, then changing one thing. Do not change two steps at once, or you cannot tell which change worked.