More leads aren't always better: what a healthy pipeline actually looks like

There is a familiar sequence in businesses that are missing revenue targets. Sales says the leads are poor. Marketing points to the volume it delivered and the cost per enquiry it improved. The board asks for more of the thing that is easiest to buy, which is usually leads. A demand generation programme gets approved, activity goes up, and two quarters later the shortfall is roughly the same size as before.

Everyone in that sequence is behaving rationally. Lead volume is often the only variable a leadership team can see clearly. Pipeline health is not on the dashboard, because most pipeline reporting describes what the pipeline contains rather than how it behaves.

That distinction is the whole problem. Value and coverage tell you what people have entered into a system. Health tells you whether those entries mean anything.

The quickest way to tell whether you have a lead volume problem

Three questions, answered from your own data rather than from memory.

What percentage of the opportunities created last quarter reached the next stage, and the one after that? How old is the average open deal, and how does that compare with the deals you actually won? Why did you lose the last ten deals you expected to win, in terms specific enough that someone could act on it?

These are not unusual measures. Salesforce's own pipeline analytics explicitly surface stage-to-stage progression, time spent in each stage, stalled opportunities and opportunities that have gone untouched, because aggregate pipeline value on its own does not tell you whether deals are progressing.

If those answers are not readily available, the constraint is unlikely to be lead volume. Adding demand to a pipeline nobody can read does not change the rate at which deals close. It increases the amount of work required to produce the same revenue, and it buries the evidence you would need to work out why.

This is not an argument against generating demand. It is an argument about sequence. Lead generation is an amplifier. It multiplies whatever your commercial process already does, including the parts that lose money.

What a healthy pipeline actually looks like

A healthy pipeline is not a large one. It is one where four things are true.

Every open deal sits at its stage for a reason you could show someone. Stages should be defined by evidence from the buyer, not by activity from the seller. "Proposal sent" is something you did. "Buyer has confirmed budget, timing and who signs" is something you learned. Only the second kind of definition tells you anything about the likelihood of closing.

Deals move at a rate you can describe. Health is visible in velocity and ageing, not in the total. Two pipelines of identical value behave completely differently if one has an average open deal age of five weeks and the other has deals that have been open since last summer. The second is not a pipeline. It is a list.

There is useful external evidence for treating ageing as a warning signal rather than an administrative metric. Ebsta and Pavilion's 2024 B2B Sales Benchmarks analysed 4.2 million opportunities across 530 companies. They found that 44 per cent of deals were pushed back, and that opportunities spending 50 per cent longer than average in qualification were 120 per cent more likely to slip. More than seven days of inactivity with no future activity was associated with a 65 per cent reduction in win rate.

Conversion between stages is stable enough to plan against. Stability matters more than the absolute number. A business that converts 18 per cent of qualified opportunities every quarter can make investment decisions. A business that converts 30 per cent and then 9 per cent cannot, and probably has a definitional or mix problem worth investigating before it has a volume problem.

That is also why a single universal "good" conversion rate is not particularly useful. Salesforce defines win rate around qualified opportunities reaching closed-won, while its stage-analysis tools calculate conversion from the actual history of the pipeline. The useful comparison is therefore your own rate by stage, segment and source over time, with an external benchmark used as a sense-check rather than a target.

Losses are recorded in a way that changes a decision. If most of your loss reasons say "price", you almost certainly do not have enough information yet to conclude that you have a pricing problem. You may simply have a reporting field that nobody has been given a reason to complete honestly.

This matters because headline loss reasons can conceal very different commercial problems. In Ebsta and Pavilion's 2024 analysis, 61 per cent of lost deals were recorded as being lost to "indecision", while the analysis of seller behaviour and conversations showed substantial differences between average and top performers in qualification, objection handling and deal management.

Coverage ratios depend on all four of those things being true, so treat them with care. A three-times coverage rule is a planning heuristic, not a law of nature. Salesforce currently recommends roughly three times target as a starting point for forecasting, while also warning that poor data quality, inconsistent stage definitions and disconnected systems weaken the forecast underneath it. Your required coverage should therefore be calibrated against your actual win rate and sales cycle rather than copied from somebody else's dashboard.

Symptoms that get misread as a shortage of leads

The real story behind common revenue complaints

What leadership sees in the dashboard often triggers a familiar blame game, but the underlying issue is usually a gap in shared definitions, process design, or measurement scope rather than a simple performance failure.

When sales says lead quality is poor

Leadership typically hears “marketing targeting or channel choice” and assumes the problem sits with acquisition. In reality, there is often no agreed, recorded definition of a good-fit account, so “quality” remains an opinion that cannot be queried, tested, or improved consistently.

When the forecast slips by a quarter, repeatedly

The surface explanation points to optimistic sellers overstating their chances. More often, the pipeline stages are defined by seller activity rather than buyer progress, so deals can advance in the system even though the buyer has not actually moved.

When there are plenty of open opportunities but few closings

The instinctive diagnosis is pricing pressure or aggressive competitors. What is actually happening is that deals are stalling rather than dying, and nothing in the process forces a decision one way or the other, leaving a growing graveyard of “open but inactive” opportunities.

When conversion rates look fine but revenue misses

The usual conclusion is “not enough at the top of the funnel.” The hidden dynamic is a shift in deal mix: more small deals are closing faster, while fewer large deals are progressing, so the average deal size and total revenue fall even though the conversion percentage looks healthy.

When marketing, sales and web numbers disagree

The first assumption is an analytics configuration error. The deeper issue is that tracking was built to report on channels, then never extended to the commercial stages that leadership decisions depend on, so each team is optimising to a different slice of the journey.

When sales and marketing report different totals

The immediate reaction is that someone is wrong. In many cases, both teams are technically right: they are simply counting different objects, typically enquiries against opportunities, without a shared taxonomy that aligns the two views.

The pattern is consistent. In each row, the visible symptom points at the top of the process, while the likely cause sits in the definitions, the handover or the record.

And the fit question is particularly worth quantifying. In Ebsta's H1 2024 benchmark update, opportunities matching the company's ideal customer profile recorded 3.1 times higher win rates, yet only 18 per cent of pipeline matched the ICP. That is an unusually clear example of why "more pipeline" and "more useful pipeline" are not the same instruction.

Why pipelines stop being readable

Most pipelines are not badly designed. They are designed once, for a business that has since changed, and then never revisited.

Stages get written from the seller's point of view because that is who fills them in. Fields get added for reporting requests and never retired. Lifecycle definitions get agreed verbally between a marketing lead and a sales lead who have both since left. Meanwhile the business adds a market, a product line or a partner channel, and the model quietly stops describing how revenue is actually won.

Systems make it worse when they fragment. When marketing, sales and customer teams work in separate tools, no one has a reliable single version of the customer journey, and manual reconciliation becomes a permanent tax on the people you least want doing admin. Salesforce similarly identifies disconnected systems, inconsistent stage definitions and poor-quality CRM data as causes of forecasting blind spots rather than simply reporting inconveniences.

FutureGroup's work with Kefron, a B2B document storage and information management business, is useful evidence of what changes when that is addressed: teams that had been operating in disconnected systems with spreadsheets and no reliable single source of truth moved to a unified platform with rebuilt pipelines, lifecycle stages and role-specific dashboards, giving them visibility of the full customer journey and, importantly, of where the conversion bottlenecks sat. The value was not the software. It was that bottlenecks became locatable.

Which is worth saying plainly: you may not need a new CRM. Replacing the system rarely fixes definitions, ownership or discipline, and a migration performed on top of unresolved definitions simply reproduces them in a more expensive tool. The question to ask first is whether your current system is genuinely incapable of representing how you sell, or whether nobody has ever configured it to try.

A diagnosis you can complete in a fortnight

None of this requires a new platform, a research programme or an agency. It requires four quarters of closed deals and someone with the authority to ask awkward questions.

Start with closed deals, not open ones. Open pipeline reflects what people currently believe. Closed deals, won and lost, are the only record of what actually happened.

Rebuild the stage conversion table by hand. Opportunities created, then the percentage reaching each subsequent stage, split by source and by segment. Do it in a spreadsheet before you build a dashboard. You are looking for the stage where the drop is steepest and least explained.

Measure age and movement, not just count. How long do won deals take? How many open deals have exceeded that by 50 per cent or more, and what is anyone doing about them? Introduce a stall threshold and a rule for what happens when a deal crosses it.

The 50 per cent threshold is not arbitrary as a diagnostic starting point: Ebsta's 2024 data found that opportunities sitting in qualification for 50 per cent longer than the norm were 120 per cent more likely to slip. Salesforce takes a similarly relative approach in its own tooling, classifying opportunities as stalled when they remain in a stage longer than the historical average for comparable opportunities.

Score won and lost deals against fit retrospectively. Take the profile you claim to target, apply it to the last two quarters, and compare hit rates. If good-fit accounts convert materially better than the rest, you have quantified the cost of an undefined ideal customer profile, and you have the basis for a scoring model rather than an argument.

External benchmarks suggest the difference can be substantial. Ebsta's H1 analysis found a 3.1-times increase in win rate for ICP-matched opportunities, while its 2025 qualification study, based on more than 655,000 B2B opportunities worth $48 billion, found well-qualified deals were 6.3 times more likely to close and closed 21.6 per cent faster than poorly qualified ones. Those are not targets to import into your business, but they make the case for measuring fit and qualification before simply increasing volume.

Read the last twenty loss reasons as prose. Not the dropdown values, the notes. Then ask the sellers involved. The gap between the field and the conversation tells you how much your loss data is worth.

Reconcile three numbers for one month. Enquiries in your web analytics, contacts created in your CRM, opportunities accepted by sales. Where they disagree, find out why once, and document it. Numbers nobody trusts get replaced in meetings by anecdotes, and anecdotes are always about the most recent deal rather than the most representative one.

What to fix before you buy more demand

Sequence matters more than scope here.

Definitions come first, because everything else measures against them: what a good-fit account is, what evidence moves a deal to each stage, what a stalled deal is and who owns it. Then the handover, including what sales receives with a lead and what marketing gets back. Then reporting built around the decisions you actually make, which is usually where to put the next pound and which deals need help this week, rather than reporting built around what the tool happens to display.

FutureGroup's work with Crowdcube is a useful illustration of what that ordering looks like in practice, because volume was never the constraint. The business had high inbound lead volumes alongside marketing-supported outbound activity, and the requirement was for marketing to become a reliable engine rather than a campaign function. The work implemented lifecycle design, advanced lead scoring and prioritisation, and reporting that connected marketing engagement to pipeline impact, so that high-intent leads were clearly prioritised across inbound and outbound, and progression was driven by recorded behaviour rather than instinct. This is work-led evidence rather than a performance claim, and that is the honest way to read it: the improvement was in what the business could see and prioritise, which is the thing more leads cannot buy you.

Third-party research points in the same direction. Ebsta and Pavilion found that despite a 23 per cent increase in pipeline generation in their 2024 benchmark population, declining win rates meant additional pipeline was not translating automatically into better revenue performance. The same study found that top performers were substantially more disciplined about qualification and disqualifying weak opportunities early.

That does not tell us what FutureGroup's RevOps team most commonly finds in the first fortnight of an engagement; that remains an internal question worth answering. But it gives the eventual SME observation a stronger role: it can show whether FutureGroup sees the same pattern in practice, or a different one.

When more leads genuinely is the answer

Sometimes it is, and treating every shortfall as an operational problem is its own form of avoidance. Buying more demand is a reasonable decision when four conditions hold at once.

Conversion between stages is understood and reasonably stable. Fit is defined and enforced, so additional volume arrives with a filter rather than replacing one. Sales has genuine capacity, meaning current opportunities are receiving proper attention rather than being triaged. And your losses point outwards rather than inwards: you are losing to competitors you were never shortlisted against, or to buyers who had never heard of you, rather than to your own follow-up and process.

That last condition is the useful one. If you are losing because you were unknown, that is a demand and visibility problem and more investment is justified. If you are losing because you were unconvincing, slow or inconsistent, additional leads will simply give you more opportunities to lose in the same way.

For a useful external sense-check, Ebsta's 2025 work puts mid-market opportunity win rate at about 21.2 per cent. More important for this diagnosis, opportunities that ran beyond normal cycle lengths fell to win rates as low as 14 per cent. The benchmark should not be treated as a universal target — deal size, category, route to market and qualification rules all change the denominator — but the direction is useful: ageing and conversion need to be read together.

The broader 2025 Ebsta/Pavilion benchmark analysed 655,000 opportunities alongside input from more than 2,000 CROs and sales leaders. It found that delayed deals materially reduced win rates and that top-performing teams differentiated themselves through efficiency and buyer engagement rather than sheer activity.

So benchmark your pipeline in this order: first against its own historical behaviour, then against comparable segments of your business, and only then against somebody else's published number. An external average can tell you that something is unusual. It cannot tell you why.

The question worth asking at your next pipeline review

Most pipeline reviews ask how much is in the pipeline. A more useful question is what would have to be true for each significant deal to close, and whether anyone has evidence that it is.

A pipeline is a set of claims about the future. Its health is not the size of those claims but the speed at which the wrong ones surface. Businesses that grow predictably are rarely the ones with the most in the pipeline. They are the ones that find out fastest when a deal is not real, and spend the recovered time on the deals that are.

If the pattern in your business is plenty of interest but unpredictable revenue, the place to start is the pipeline you already have rather than the one you were about to buy. You can read more about how FutureGroup approaches building a stronger pipeline, or, if the issue sits specifically in what happens after a lead arrives, where good leads quietly disappear.