How to Increase Sales Funnel Conversion Rate: The Conversion Chain

Your sales funnel conversion rate is closed deals divided by funnel entries, and it is the product of every link in between. In a worked eight-link funnel converting 0.21% of ad clicks, a 20% gain on each link compounds to 4.3x: 2.1 closed deals per 1,000 clicks becomes 9.1. Fix the cheapest link, not the worst one.

The short version

• The formula. Funnel conversion rate = closed deals ÷ entries into the funnel over the same cohort. Everything else is a link rate.
• The structure. Eight sequential link rates multiply. Five more pipeline stages are recovery or repeat loops that re-enter the chain rather than sit inside it.
• The arithmetic. +20% on all eight links = 4.30x end to end. +10% on all eight = 2.14x. +50% on the landing page alone = 1.50x.
• The decision rule. In a multiplicative chain a 20% relative gain is worth exactly the same at every link, so attack the link that is cheapest to move 20%, not the one with the lowest absolute rate.
• The benchmark you can compare against. Unbounce puts the median landing page at 6.6% across all industries. That is link one of eight, and it is the only link with a large public benchmark.

What does “sales funnel conversion rate” actually measure?

It measures closed deals against entries into the funnel, for the same cohort of people, over a window long enough for that cohort to finish converting. Two things break it in practice. First, the denominator is undefined: click-to-closed-deal, lead-to-closed-deal and booked-call-to-closed-deal are three different numbers and teams quote whichever is flattering. Second, the numerator and denominator are usually taken from the same calendar month, which mixes deals that entered in March with leads that entered in June.

A funnel conversion rate that is not cohort-matched is not a conversion rate; it is a ratio of two unrelated months. Pick one denominator, state it in the metric name (“click-to-closed-deal, 90-day cohort”), and never compare it to a number whose denominator you do not know.

THE CONVERSION CHAIN: the eight links between a click and a closed deal

THE CONVERSION CHAIN is our name for the structure underneath the single number. A funnel is not a leaky bucket, it is a chain of eight conditional probabilities, and the end-to-end rate is their product. The worked example below is a mid-ticket B2B funnel with 1,000 paid clicks a month; substitute your own rates, the structure holds regardless.

Multiply the example column: 0.08 × 0.60 × 0.50 × 0.60 × 0.70 × 0.60 × 0.70 × 0.50 = 0.21%, or 2.1 closed deals per 1,000 clicks. Multiply the +20% column and you get 0.91%, or 9.1 deals. Nothing heroic happened at any single link.

Five of the seventeen pipeline stages are not links in this chain at all. No-show recovery, cancellation recovery and long-term follow-up are re-entry loops: they put people back into link 5 or link 7 rather than multiplying through. Database reactivation and referral or repeat business start a new chain with a shorter front end, which is why their economics look nothing like paid traffic — across our own Colliers-era reactivation work we recorded 4.4% average and 8.9% peak conversion on dormant records, which is our record on our own campaigns rather than an industry benchmark. The remaining twelve stages are sequential, and they compress into the eight measurable links above, because pairs of them share a single timestamp: ad click and form capture are link 1, speed to lead and contact are link 2, the discovery call and the proposal it produces are link 6, and between-call follow-up and the second close call are link 7. Modelling a loop as a link is the most common structural error we see: it double-counts people and makes the chain reconcile to a number your CRM never produced.

Why 20% on every link beats 50% on the best-studied link

This is the arithmetic that answers “we have already optimised the landing page”. The landing page is link one of eight, and improving it changes the end-to-end rate by exactly the proportion you improved it — no more.

A 10% gain on all eight links beats a 50% gain on the one link the entire conversion-optimisation industry studies. The 55% row is not hypothetical: the 2025 Ebsta x Pavilion GTM Benchmarks, built on 655,000 opportunities and input from 2,000+ CROs and sales leaders, report that early decision-maker involvement “boosts win rates by 55%”. That is a genuinely large result — and win rate spans links 6 to 8 of this chain rather than sitting in any one of them — yet on its own it is worth less than a boring 10% across the board.

How to increase sales funnel conversion rate, in order: the cheapest-20% rule

Because the links multiply, a 20% relative gain is worth an identical amount at every link. Link 1 and link 8 have the same leverage. That collapses prioritisation to a single question, and it is not the question most teams ask.

The cheapest-20% rule: rank the links by what it costs you in hours, headcount and calendar time to move each one 20%, and start at the top of that list — not at the link with the lowest absolute rate. A 50% show rate looks like the emergency; it is only the emergency if moving it 20% is cheaper than moving something else 20%.

Run it in this order:

1. Instrument all eight links before changing any of them. You need eight timestamps per record in your CRM and one query. If a link has no timestamp, its rate is currently an opinion.
2. Price each 20%. For each link, write down the concrete change and its real cost. “Reply to every enquiry within five minutes including nights and weekends” is a roster and tooling change; “improve the discovery call” is a coaching programme measured in months.
3. Do the two cheapest first, together. Two links at +20% is 1.44x. Ship both before you touch anything expensive.
4. Re-measure the cohort, not the month. Wait one full sales cycle. Improvements to link 7 do not show up in a 30-day window if your cycle is 60 days.
5. Repeat. The links you skipped because they were expensive get cheaper once the ones above them are automated and producing volume.

In practice the cheapest 20% is almost always in links 2, 5 and 7 — contact rate, show rate and the follow-up between calls — because they are mechanical rather than creative. Getting an enquiry answered in minutes rather than the next business day is a speed-to-lead automation problem, and holding a booked call in the calendar is a reminder and reschedule problem. Neither needs a creative breakthrough. Both are what an AI sales agent handling first response and follow-up is actually for.

What good looks like on the only link with public benchmarks

Link 1 is measured to death and links 2 through 8 are almost not measured publicly at all. That asymmetry is why so much funnel work concentrates where the data is rather than where the money is.

• Landing pages. Unbounce’s Conversion Benchmark Report puts the median landing page at 6.6% across all industries, from 41,000 landing pages, 464 million visitors and 57 million conversions as at Q4 2024. Median, not mean — the mean is distorted by a long tail.
• Paid search. The LocaliQ and WordStream search advertising benchmarks report an average search advertising conversion rate — Google Ads and Microsoft Ads together — of 8.18% across industries for 2026, drawn from over 13,000 campaigns across 23 industries.
• Links 2 to 8. There is no comparably sized public dataset. Every number you will find for set rate, show rate or proposal-to-close is either a vendor’s own book of business or an unsourced round figure copied between blog posts.

If your click-to-lead rate is inside the 6–8% band, link one is not your problem, and no further landing page test will change your funnel conversion rate by more than a rounding error. The absence of benchmarks for the middle of the chain is not evidence that the middle is fine. It is evidence that nobody publishes it.

What we see in our own client work — and where the numbers stop reconciling

This section is our operator experience, not research. In our own client work we typically see a business still running 2020-era operations — enquiries answered next business day, two follow-up attempts, no reactivation, no recovery of no-shows — roughly triple its paid-ad conversion rate over the first months of an engagement. We typically see speed to lead alone worth about 3x, doubling contact rate through more outbound worth about 2x, and doubling set rate worth about 2x. Those are our observations across the campaigns we run. They are not a study: there is no published n, no fixed window and no dataset behind them, and we will not present them as one.

Here is the part most agencies leave out: those component numbers do not multiply. 3x times 2x times 2x is 12x, and we do not see 12x. They overlap heavily — fixing speed to lead is a large part of how contact rate improves, and a higher contact rate is a large part of how set rate improves, so you cannot bank all three. The honest headline is roughly 3x, not the product of the parts. We publish that because a set of numbers that do not reconcile is worse than a smaller number that does.

The one lift figure we publish with a stated method is different: our methodology page defines the 7x average sales lift as trailing 3-month closed-deal revenue at month 6 of engagement over trailing 3-month revenue immediately before launch, averaged across clients who supplied both numbers — and the same page discloses that the median is closer to 4x. That is an average pulled up by high performers, not a typical outcome. The compounding table further up this page is the more useful artefact anyway, because it is arithmetic you can check rather than a result you have to trust.

Do it yourself, or hand the chain over? The thresholds we use

You can run all eight links yourself, and below a certain volume you should. The measurement is a spreadsheet job: eight timestamps and one query. The ongoing cost is not the tooling, it is keeping the stage definitions honest — roughly an hour a week of someone senior enough to say “that is not a qualified lead” — and that is the first thing to decay when the team gets busy. The thresholds below are the ones we use when we scope work ourselves; they are judgement, not a measured finding.

Below roughly 50 new leads a month, hire nobody and buy nothing — fix the two cheapest links by hand and re-measure in a quarter. We run this on a pay-per-result basis, which means we are paid on booked qualified appointments rather than a retainer or a seat count, and that only makes sense on the volumes above. This page is one of a set covering every stage of the pipeline individually — as those pages go live, the pipeline stages hub will link down to each of the seventeen stage pages, and each stage page back up to this one.

Frequently asked questions

What is a good sales funnel conversion rate?

There is no single good number, because the answer depends entirely on your denominator. For click-to-lead there is a real benchmark: Unbounce’s Conversion Benchmark Report puts the median landing page at 6.6% across all industries, from 41,000 landing pages and 57 million conversions as at Q4 2024, and LocaliQ and WordStream report an average search advertising conversion rate — Google Ads and Microsoft Ads together — of 8.18% for 2026 across 13,000+ campaigns. For click-to-closed-deal there is no equivalent: the illustrative chain on this page multiplies out to about 0.2%, but that is arithmetic from example rates rather than a benchmark, and no public dataset covers the full chain.

How do you calculate sales funnel conversion rate?

Take a cohort of entries — every click, or every lead, in a fixed period — and count how many of those specific records closed, waiting at least one full sales cycle before you read the result. Divide, and name the denominator in the metric. Dividing this month’s closed deals by this month’s new leads is the standard mistake: it mixes cohorts and reports a number that rises when lead volume falls.

Which funnel stage should I fix first?

The one that is cheapest to improve by 20%, not the one with the lowest rate. Because the link rates multiply, a 20% relative gain produces an identical 1.2x end-to-end wherever you win it, so cost of improvement is the only thing that varies. In most funnels that means contact rate, show rate and follow-up between calls before creative or copy.

Does improving one stage really change the whole funnel?

Yes, proportionally and no more. Lift one link by 20% and the funnel rate rises 20%. That is why single-lever wins disappoint even when they are large: the 2025 Ebsta x Pavilion GTM Benchmarks, based on 655,000 opportunities, found early decision-maker involvement “boosts win rates by 55%” — a 1.55x end-to-end effect, which a 10% gain on eight links (2.14x) still beats.

Why do my stage conversion rates not multiply to my reported funnel rate?

Three usual causes, in order of frequency. Each stage is measured over a different window, so the cohorts do not match. Records re-enter the chain through recovery loops — a recovered no-show is counted once at link 5 and again on the re-book — which inflates the product. And at least one stage has no timestamp, so its rate is an estimate someone typed in. Fix the timestamps first; the reconciliation usually follows.

How long before a funnel improvement shows up in the numbers?

One full sales cycle plus the length of your measurement window, minimum. Changes to the front of the chain (links 1 and 2) surface fastest because the affected records are still early; changes to links 6 to 8 take a complete cycle before a single cohort has finished converting. Reading a link-8 change after 30 days on a 60-day cycle measures noise.