Portal invoices say 1,000 leads, the CRM says 970, and nobody can say how many ended with no outcome. Here is the definition, the Lead Leakage Rate formula and an illustrative month, stage by stage.
Written by The IndiaCalling.ai team · We build AI voice sales agents for Indian real estate developers, working directly with presales teams and portal leads.
(illustrative) Month-end at a Hyderabad developer's presales office. The portal invoices add up to 1,000 enquiries for the month. The CRM shows 970. The marketing head says presales is not calling. The presales head says the leads were poor. Both are describing something real, yet nobody can answer a simpler question: how many of those 1,000 leads ended the month with no recorded outcome, neither a booked visit nor a logged reason to stop? Without that count, every review becomes an argument about blame. This post gives the count a name, a formula and a worked month.
Key takeaways
Lead Leakage Rate (LLR) = leaked leads / leads received x 100, for one enquiry-month cohort read at day 30.
Leaked is not lost: a lead closed with a logged reason is a decision, a lead with no outcome is leakage.
Stage leakage = exits with no outcome / leads entering that stage, so each of six stages gets its own rate.
One Indian case study (LeadSquared) reports leakage of 80% without saying how it was measured, so vendor figures are not comparable.
No Indian primary benchmark for leakage was found, so the 34.5% in the worked example is illustrative.
What is lead leakage in real estate?
Lead leakage is the share of leads a developer has paid for that leave the funnel with no recorded outcome. IndiaCalling.ai builds AI voice agents (software that phones a lead, holds the conversation and logs the result) for Indian real estate developers. A CRM (customer relationship management system) can only report what someone entered into it.
LeanData, a lead management software vendor, defines leakage as "the progressive and cumulative loss of Leads" as they progress through the sales funnel. LeadSquared's explainer likewise describes prospects who fail to reach the end of the pipeline, and gives no formula. A definition you cannot compute cannot be compared month to month, so this post adds a test.
Call it the three-outcome test. At every stage, each lead ends as Progressed (moved to the next stage), Closed with reason (declined, disqualified, invalid number or cadence completed) or Leaked (exited with nothing recorded). A cadence is the planned sequence of call attempts. Only the third outcome counts as leakage.
How is lead leakage different from lead loss?
Lead loss is a decision. Lead leakage is the absence of one. A buyer who says the budget is too low and is logged as declined is lost, and the desk learned something. A buyer never dialled has leaked.
Sell.Do, an Indian real estate CRM vendor, draws a similar line in its broker article. A lost deal is a conversation that failed and gets a post-mortem, while a lead that was never called leaves no record. This post widens that idea: a lead called twice and then dropped before the planned cadence ended has also leaked, because nobody decided to stop.
What is the Lead Leakage Rate formula?
Lead Leakage Rate (LLR) is leaked leads divided by leads received, times 100, for one enquiry-month cohort read at day 30. The denominator is leads received from the sources, taken from portal invoices or dashboards, not the leads that reached the CRM.
That choice matters, because leads that never reach the CRM are the first leak. Divide by CRM leads and you delete them from the problem. Sell.Do's support documentation says 99acres pushes leads to its CRM using a base URL and API key, so a broken key or mapping is one way a lead can stop at the door unseen. That is our inference, not the page's claim.
Fix the cohort by enquiry month and read every status at day 30, so the figure stays comparable. For one stage, the stage leakage rate is leads exiting with no outcome divided by leads entering it.
Where do portal leads leak?
A portal lead can leak at six stages: capture, ownership, first attempt, contact persistence, next step and visit. The table counts each one for a single illustrative cohort of 1,000 portal enquiries. The causes are covered in the five-leak diagnosis of why real estate leads are not converting; this section only measures.
Lead Leakage Rate by stage: one illustrative monthly cohort
Stage
Entered
Leaked (definition)
Stage leakage rate
Capture
1,000 received
30 not in CRM
3.0%
Ownership
970
20 unassigned
2.1%
First attempt
950
60 never dialled
6.3%
Contact persistence
890 attempted
120 abandoned before cadence end
13.5%
Next step
520 connected
90 with no dated next step
17.3%
Visit
130 booked
25 no-shows never recalled
19.2%
Total
1,000 received
345 leaked
34.5% LLR
Representative figures based on typical real estate presales operations. Arithmetic only: all counts and the Rs 600 CPL are assumptions, not measured results; cohort is one enquiry month read at day 30.
The case study attributes the 80% to lead disposal not being on time. It does not say how leakage was measured, or whether its later 70% reduction is relative or in points. Treat any figure without a stated denominator the same way.
How do you calculate it for a month? A worked example
Assumptions: the cohort in the table, an assumed cost per lead of Rs 600, every status read at day 30. Representative figures based on typical real estate presales operations.
Leaked leads are 30 + 20 + 60 + 120 + 90 + 25 = 345, so LLR = 345 / 1,000 = 34.5%. Of 890 attempted, 520 connected, 50 were invalid numbers, 200 completed the cadence without a connect and 120 were abandoned (520 + 50 + 200 + 120 = 890). Of the 520 connected, 130 booked a visit, 300 closed with a reason and 90 had no dated next step (130 + 300 + 90 = 520). Of the 130 visits booked, 75 were held, 30 no-shows were closed with a reason and 25 were never recalled (75 + 30 + 25 = 130).
Spend is 1,000 x Rs 600 = Rs 6,00,000. The 345 leaked leads carry Rs 2,07,000 of it. With 75 visits held, cost per held visit is Rs 6,00,000 / 75 = Rs 8,000.
The largest count (120) sits in contact persistence, while the highest rate (19.2%) sits at the visit stage, so report both per stage and in total.
Now suppose the desk re-attempts the 60 never-dialled and 120 abandoned leads, 180 in all, and assume 40% connect at 2.5 minutes each. That is 72 connected calls and 180 connected minutes. IndiaCalling.ai's pricing page lists Rs 6 per connected minute with zero monthly fee, so the calling cost is 180 x Rs 6 = Rs 1,080. Check your consent and DND (Do Not Disturb registry) obligations with counsel before re-contacting older leads.
A next-step leak is the easiest to picture. In this illustrative exchange the call ends with a dated callback and not a vague promise:
Buyer: Abhi busy hoon, shaam ko call kijiye.
Agent: Bilkul. Aaj shaam 7 baje theek rahega?
Buyer: Haan, 7 baje chalega.
Agent: Theek hai, main aaj 7 baje call karunga aur callback note kar leta hoon.
Translation: the buyer says they are busy and asks for a call in the evening. The agent proposes 7 pm today, the buyer agrees, and the agent notes the callback. The lead now has a dated next step, so it is Progressed and not Leaked. Confirm in a demo whether a deployment writes the callback to your CRM. (illustrative)
When this does not work: limits of the measure
Small cohorts. In a cohort of 100 leads, each lead moves the rate by a full point. Use a rolling three-month figure.
Day 30 is a convention. Long sales cycles may need day 45 or 60. Pick one and keep it.
Closed with reason can be gamed. A desk judged on LLR can tag a never-dialled lead as an invalid number. Audit reason codes against recordings monthly.
Channel partner leads and walk-ins sit outside the denominator unless they pass through the same intake. This measures portal and campaign leads only.
Duplicates. Decide whether a buyer on two portals counts once or twice, in numerator and denominator alike.
A low LLR with few bookings points to lead quality or pricing. Leakage measures process, not demand.
Baseline three monthly cohorts before setting a target. The speed-to-lead post covers first response, and a cold lead reactivation campaign is one way to work abandoned leads once you can count them.
No Indian primary source we found publishes a lead leakage benchmark, and vendor figures rarely state a denominator, so a good rate is the one you beat from your own baseline. Measure three monthly cohorts with the same day-30 reading, report the rate per stage as well as in total, and only then set a target. The 34.5% in this post is illustrative, not a benchmark.
What is the difference between lead leakage and lead loss?
Lead loss is a decision recorded in the CRM: the buyer declined, was disqualified or had an invalid number, and someone logged why. Lead leakage is a lead that exited with no recorded outcome, such as a number never dialled or a cadence dropped halfway. Losses teach the desk something. Leaks only cost money, which is why the two are counted separately.
Is an unreachable lead a leaked lead?
Only if the desk stopped before its own cadence ended. If the plan was six attempts over seven days and all six were made without a connect, the lead is closed with a reason: cadence completed. If the desk dialled twice and moved on, it leaked. The test is whether a person or a rule decided to stop, and whether the record shows it.
Should I use leads received or leads in the CRM as the denominator?
Use leads received from the sources, taken from portal invoices or dashboards. Leads that never reach the CRM are the first leak, and dividing by CRM leads hides them. Compare the two counts each month. In the worked example, 1,000 were received and 970 reached the CRM, so capture alone leaked 30 leads, or 3.0%.
How often should a developer measure lead leakage?
Report it monthly, by enquiry cohort, read at day 30, so months stay comparable. Between reports, check the open cohort weekly for early warning, for example how many leads have no attempt logged after two days. Do not change the day-30 convention midyear, because a new reading point breaks the comparison with earlier cohorts.
Does a CRM stop lead leakage?
Not by itself. A CRM records the leads that reach it and the activity someone logs, so it cannot count leads that never arrived or calls that were never made. It helps once you add required fields, such as owner, attempt count, outcome and next-step date, and report leakage from them. Compare CRM leads with portal-reported leads every month to see the capture gap.
Can an AI voice agent reduce lead leakage?
It can remove two leaks: leads never dialled and cadences abandoned early, because IndiaCalling.ai's pricing page says the agent calls every lead within 30 seconds, qualifies buyers and books site visits. It cannot fix a capture gap or a junk lead. Test it on your own cohort and compare stage leakage rates before and after.
See it work on your own leads
Book a 30 minute call and we will show you an AI sales agent built on your projects and your numbers.