What is revenue leakage? A working definition for B2B teams

Revenue leakage is the commercial value of records your company already owns (leads, deals, and quotes) that still have a plausible buyer attached but no next step and nobody working them. That's the entire definition. Owned demand, still alive, going unworked.

This guide is the full treatment: what qualifies and what doesn't, where leakage concentrates, how to measure it in a way that survives a skeptical CFO, what causes it, how to stop the flow, and how to recover what already leaked. It's long because the subject deserves rigor, not because it needs padding. Use the section headings to jump.

The definition, taken seriously

Every word in the definition is doing work, and the edge cases are decided by reading it literally.

"Records your company already owns" scopes it to demand that has been captured: a form filled, a conversation logged, a quote sent. A market segment you haven't touched isn't leaking; it's unaddressed. Leakage is loss after capture.

"A plausible buyer attached" is the honesty clause. A contact whose email bounces, a company that folded, a student doing research: these aren't leaks, they're noise, and counting them inflates the figure into fiction. Plausibility is checkable (the domain resolves, the person is still employed there, the company still fits your profile), and it gets checked before anything is counted.

"No next step and nobody working it" is the operative failure. Not slow, not deprioritized with a date attached, but silent: no scheduled task, no owner acting, no recorded decision to stop. A deal a rep deliberately parked with a follow-up task for March isn't leaking. The same deal with no task is.

The unit of leakage is the record, but the substance is a relationship mid-conversation. Behind every dormant record is a buyer who asked a question and stopped getting answers.

What revenue leakage is not

The definition earns its keep through its exclusions, because each one points at a different problem with a different fix.

It is not a lost deal. A deal worked to a real no (wrong fit, chose a competitor, no budget authority) is a conclusion. The system did its job; the answer was no. Leakage is the absence of a conclusion: the deal that never got to no because it never got another touch.

It is not churn. Churn is revenue you had and lost after the sale, and it belongs to retention. Leakage happens earlier: revenue you earned the right to pursue and then stopped pursuing.

It is not a data-quality problem, though it hides behind one. A dormant record is usually accurate: right company, right contact, right deal size. It isn't wrong. It's abandoned. Cleansing tools can't fix it because there's nothing to cleanse, only something to work.

And it is not a demand problem. Demand problems are solved at the top of the funnel with more spend. Leakage is a middle-of-funnel problem, and pouring more demand into it makes it worse, because attention is the constraint and new leads compete with old ones for it.

Why it matters: the economics

Three properties make leaked pipeline economically unusual, and all three favor recovery over replacement.

First, the acquisition cost is already sunk. A dormant record cost the same marketing dollars and rep hours as the leads that converted; recovering it costs only the outreach. When a worked lead costs $400 to $700 to generate in a typical sales-led motion, a pile of several hundred dormant records represents six figures of spend with no return yet booked against it.

Second, the clock runs faster. Re-engaged buyers skip the education phase; they already know the problem and evaluated you once. Across the 214 audits behind our medians, the first closed-won from a re-engaged record lands at a median of day 26, which is a different planning horizon than a cold pipeline build.

Third, the pile compounds silently. Leakage is a flow, not a stock: the median audited team sees about 1.2 records per rep go quiet every month. The pile in the portal today is that flow accumulated over every quarter since the CRM went live. Left unmeasured, next quarter's pile is this quarter's plus one more layer. For a $10M company at ordinary assumptions, the single-year floor alone runs six figures; the arithmetic is walked through in what a modest leak rate costs a $10M company.

The four places it concentrates

Audited portals leak in the same four places, in different proportions. Reading them in order is reading the lifecycle of a leak.

Unworked inbound

Leads that arrived, were logged, and never received a first touch. Usually the largest bucket by count, and the one paid for most directly. Detection: contacts created more than 90 days ago with no owner or no logged activity. The upstream cause is almost always the first-touch gap: routing built for an old org chart, handoffs where each side assumes the other is watching, and triage that answers demo requests while content leads rot. The decay curve and the standard that stops it are covered in someone else is calling the lead you paid for.

Stalled deals

Opportunities that stopped moving mid-stage and were never closed out either way. The costliest bucket per record, because the buyer was already deep in the conversation; a proposal-stage deal with 30 days of silence has all the expensive work behind it and only the follow-through missing. Detection: open deals past their stage's 75th percentile age, or in proposal stages with no scheduled next activity.

Expired "not now"s

Deals closed-lost on timing or budget, where the stated reason has since lapsed and nobody returned. A list of buyers who already evaluated you, qualified themselves, and gave a reason with an expiry date. In audit after audit this bucket carries the highest close rate per record worked, because the qualification already happened. The re-approach playbook is in "not now" is not "no", and its close cousin, the deal that drifted through three close dates before anyone called it, is treated in a pushed close date is a decision nobody made.

Ownerless records

Leads, deals, and renewals stranded by rep departures, territory changes, or imports that never got assigned. Revenue with no name on it can't appear in anyone's pipeline review, which makes this the only bucket that is invisible by construction. Every departure since the portal went live left a layer. Detection: records owned by deactivated users, plus the subtler variant, records bulk-transferred to an active user who never made a first touch. Prevention is procedural: the rep offboarding checklist.

All four detection filters, with HubSpot-specific recipes, are collected in the four places revenue leaks out of a HubSpot portal.

How to measure a leak rate

Leak rate is the share of eligible records, or of the value attached to them, that meets defined leak conditions. The number is only as good as its discipline, and the discipline has three parts.

Define eligibility before counting

Exclude records with no plausible buyer: dead companies, bounced emails, students, job seekers, competitors, vendors. A leak figure padded with ghosts is a marketing number. In practice this pass removes a third to a half of superficially dormant records in a typical portal, and doing it first is what separates a defensible figure from an impressive one.

State the conditions

"No logged activity in 90 days and no scheduled next step" is a condition someone can check. "Feels neglected" is not. Conditions should fit the record type: 90 days of silence is the right window for a lead, far too loose for a proposal-stage deal, where 30 days already means trouble. Whatever the conditions are, write them down with the finding, so anyone can rerun the count and get the same answer.

Discount by your own history

Dormant records that get worked again close at a lower rate than fresh ones. The median across our audits is 4.4%, but the honest multiplier is the one from your own portal's history on re-worked records, or a deliberately conservative default if you don't have one. Keep recorded value separate from derived value while you do it: a deal amount a rep entered is evidence, a value inferred from company size is an estimate, and a total that mixes them is neither. The full counting doctrine, including what gets thrown out before a dollar is claimed, is in why we count conservatively.

The formulas

Record leak rate: eligible records meeting leak conditions, divided by all eligible records. Value leak rate: pipeline dollars attached to leaking records, divided by dollars attached to all eligible records. Expected recoverable value: leaking records × your re-engagement close rate × average deal size, plus stalled-deal amounts at a stated discount.

Measure both rates, because they diagnose different diseases. A high record rate with a proportional value rate means follow-up is failing broadly: a coverage problem, fixed with SLAs and next-step rules. A low record rate with a high value rate means the team follows up on everything except, somehow, the big ones, which usually points at deal complexity or avoidance of a hard conversation rather than volume. The fixes are deal-level: stage-age alerts and honest close dates.

Cadence and trend

Measure quarterly, same conditions every time, and date the results. The trend is the real report: a leak rate falling quarter over quarter means the process fixes are holding; a flat or rising one means the leak is outrunning them. A single snapshot tells you the size of the problem. Only the trend tells you whether you're solving it.

For a first pass that needs no tooling, the 15-minute dormant-pipeline self-check produces a conservative floor figure from four filters and one multiplication.

Why leakage happens

The root causes are structural, which is why exhortation never fixes it.

• Incentives point at motion. Comp plans pay for closing and never punish letting a record age. Every individually rational choice of the fresh lead over the aging one accumulates into a pile nobody chose.
• Attention is finite and triaged by urgency. New records arrive loud; old ones age silently. Under load, the queue sorts itself by recency, not value.
• Shame compounds with age. The longer a record sits, the more embarrassing it is to touch, so the records most in need of a touch are the least likely to get one.
• There is no road back in. Pipelines model forward motion; almost none have a stage for "went quiet and came back," so re-entry depends on individual memory and courage.
• Transitions strand books. Departures, territory changes, and reorgs convert whole books to silence at once, wholesale rather than retail.

The full argument that this is a systems problem rather than a rep-discipline problem is in why good reps let leads go quiet.

Stopping the flow

Prevention is a short list of mechanical standards. None require new software; all require an owner.

• A first-touch SLA set where intent lives: same business day for hand-raisers, 48 hours for everything else, with the miss rate on a dashboard someone reads.
• A next-step rule: every open deal carries a scheduled activity a buyer would recognize, or it gets closed honestly.
• The two-push rule: a close date moved twice triggers a closing question, and twice-pushed deals are forecast at their real historical rate.
• A revisit schedule for timing losses: closed-lost-on-timing records get a dated follow-up task at creation, so the pile can't form.
• An offboarding procedure that reassigns by record, not in bulk, and audits the reassignment at 30 days.
• A weekly review that inspects silence, not just motion. The five saved filters that do this are in five questions that make a pipeline review tell the truth.

Teams that hold these standards stop accumulating dormant records at the old rate. What the standards can't do is recover the backlog that formed before they existed.

Recovering what already leaked

Recovery is its own discipline, and it fails in predictable ways when treated as a campaign.

Work the pile in order. Rank by expected value and recency, not alphabetically: stalled proposals and expiring "not now"s first (highest close rate, shortest clock), unworked inbound by source quality, orphaned records newest-departure-first. A recovery effort that starts with the easiest records instead of the most valuable ones runs out of attention before it reaches the money.

Re-approach with context, not a campaign. The message that works references the real prior conversation in the rep's own voice; the message that fails is a drip sequence pretending the silence didn't happen. Specificity is the signal that a person remembered.

Route revived buyers back to sales with history intact. A re-engaged opportunity that re-enters as a "new lead" forces the buyer to start over, which is the second abandonment. The record's history is the asset; keep it attached.

Count outcomes, not activity. A delivered message is activity. A reply is engagement. A reopened opportunity is pipeline. Revenue is recovered only when the commercial outcome lands in the CRM. Any recovery effort that reports sends instead of outcomes is measuring the wrong end of the pipe.

Expect honest numbers. At the 4.4% median re-engagement close rate, a 500-record recovery pool yields roughly 22 reopened-and-closed deals over time, not a transformed company. What makes recovery attractive isn't a miracle rate; it's that the records are already paid for and the clock is short.

This is the motion Regather runs as a managed system: the audit names and ranks the pool, recovery plays work it from your reps' own inboxes with human approval on every send, and daily monitoring catches the next leak while it's one record instead of a layer.

When leakage is not your problem

Candor clause. Some portals are tight: disciplined follow-up, clean ownership, honest close dates, and a leak floor under 2% of revenue. If the self-check says that's you, the correct spend on this problem is nothing beyond a quarterly re-measure. Leakage is also rarely the priority for very young CRMs (nothing has had time to leak), purely product-led motions with no human follow-up step, and businesses whose deals close in a single call. The concept earns attention exactly where the fit list says: sales-led teams, multi-touch deals, and a CRM with history.

The vocabulary

Working definitions for the terms used across these guides. Revenue leak: an owned lead, deal, or quote with remaining commercial potential but no active next step and no recorded conclusion. Leak rate: the share of eligible records or value meeting stated leak conditions. Pooled pipeline: the total discounted value of a portal's leaking records, the number an audit exists to establish. Recovery play: a structured re-engagement of one leak bucket with an owner, a list, and a date. Dormant record: any record meeting leak conditions; dormancy describes state, leakage describes the value passing through it.

Where to start

Measure before you buy anything, including from us. Run the 15-minute self-check for a floor figure this afternoon, or request the free Revenue Leak Audit for the record-level version: every dormant record named with its owner and last touch, exclusions listed, value discounted against your own history, and the recovery order ranked. If the number is small, you'll know your portal is tight, and that answer is free. If it isn't, you'll know exactly which records to put back in front of a rep first.

A leak rate turns silence into a number, and numbers get owners.