SaaS demand generation

Most demand generation programs in SaaS are capture programs with a broader title on the org chart. They bid on category terms, publish comparison pages, pay for G2 category placement, and retarget whoever those channels produce. That works right up to the point where you own most of the intent your category already generates, and then every extra dollar buys a worse click. Breaking through that ceiling means doing a second, slower job: manufacturing the demand you later capture, and measuring the two halves with completely different instruments.

Why capture-only programs stall at the same predictable point

Paid search, review marketplaces and bottom-funnel content all harvest intent that existed before you spent a dollar. That intent is a fixed pool in any given quarter. Once your impression share on the terms that matter passes roughly 70 percent, the next increment of budget buys broader match types, worse queries and people who were never going to buy.

The symptom is specific and easy to spot: cost per lead climbs 40 to 80 percent over three or four quarters while lead volume stays flat or drifts down. Teams usually respond by blaming creative, swapping agencies, or rebuilding the landing page. None of that fixes it, because the problem is not efficiency. The pool is empty.

The LinkedIn B2B Institute popularised the framing that explains it, drawn from Professor John Dawes at the Ehrenberg-Bass Institute. At any moment roughly 5 percent of business buyers are actively in-market and 95 percent are not. Capture channels compete for the 5 percent against every other vendor in the category, in a live auction, with price set by the most desperate bidder. Creation channels talk to the 95 percent so that when they enter the market, your name is one of the two or three they type into Google unprompted.

There is a genuine cost to acting on this. Creation spend does not return a measurable opportunity inside a monthly reporting cycle, and if your board reviews marketing on a 30 day cadence you will be defending a line item that looks like it does nothing for two quarters. That is the real reason most teams stay capture-only. Not ignorance, cowardice about the reporting conversation.

What actually counts as creation and what counts as capture

Take $50,000 and spend it both ways. Spent on capture, it buys perhaps 18 to 40 sales-accepted opportunities from Google Search, G2 and retargeting in a mid-market SaaS category at $25K ACV. You will see most of them inside six weeks, they will be traceable, and roughly a fifth will close within two quarters.

Spent on creation, the same $50,000 buys about four months of a serious LinkedIn reach program, a research report with real sample sizes, and a podcast tour. In the same six weeks it returns almost nothing you can put in a pipeline report. What it returns over nine months is a measurable lift in branded search volume, a higher share of demo requests that say ‘a colleague recommended you’, and capture channels that start converting better because the name is no longer cold.

The line between the two is not the channel, it is the intent of the person on the other end. A LinkedIn ad pointed at a demo request from a lookalike audience is capture with poor targeting. A comparison page that ranks for a competitor’s name is capture. A Gong-style report on what separates winning sales calls from losing ones is creation, even though it sits on the same blog. We go deeper on the distinction and the budget mechanics in demand creation versus demand capture.

How to split the budget between creation and capture by stage

The rule is simple and unpopular: do not fund creation until capture is saturated and measured. A seed-stage company with 15 percent impression share on its own category terms has no business running a brand campaign. Go buy the demand that already exists, because it is cheaper than the demand you would have to manufacture.

Read the percentages as program spend, not total marketing budget. Headcount, product marketing and customer marketing sit outside this split. If you want to model the split against your own numbers rather than accept a table, the demand generation budget allocator takes ACV, target pipeline and current impression share and returns a recommended allocation.

One warning about the transition. Moving from 80/20 to 65/35 usually means cutting a capture line that currently reports attributable opportunities in order to fund a line that reports none. Somebody senior has to sponsor that, in writing, with an agreed measurement window of at least two quarters. Without that sponsorship the creation budget gets clawed back in month four, which is exactly when it is at its least defensible and about one quarter from working.

What each channel costs per opportunity, by ACV band

Cost per lead is a vanity number in SaaS because lead quality varies by an order of magnitude between channels. The comparison that matters is cost per sales-accepted opportunity, the payback window it implies, and the volume ceiling before the channel degrades.

Three things fall out of that table that most channel guides skip.

Branded search is the cheapest line in every account and the one most often argued about. Someone will propose cutting it because ‘they would have found us anyway’. Run the test before you agree: pause brand bidding in half your geographies for four weeks and measure total demo requests, not clicks. In contested categories where competitors bid on your name, the loss is usually real and large. In categories where nobody bids on you, it is close to zero and you should cut it.

Outbound looks expensive per opportunity and often is, but it is the only capture channel where you choose the account. At $60K ACV with a defined target list of 400 companies, a cost per opportunity of $6,000 against a payback ceiling of 24 months is a perfectly sane trade. At $8K ACV the same number is ruinous. This is the whole reason we keep demand generation playbooks split by ACV band rather than publishing one channel mix.

Review marketplaces have a hard ceiling that nobody warns you about. G2 category traffic is finite, and once you hold the top sponsored position the incremental spend buys adjacent categories where fit is worse. Expect the channel to cap out somewhere between 15 and 40 opportunities a month for most mid-market categories, then flatten no matter what you pay. Our full breakdown of how to sequence these sits in the SaaS demand generation channel strategy guide.

The three layer measurement stack that survives a board meeting

Any attribution system that reports one confident number for a nine month buying process involving eight people is lying to you. Not approximating. Lying. The buyer heard a podcast in March, saw three LinkedIn posts in May, asked two peers in a Slack community in July, then typed your brand name into Google in September and clicked a paid ad. Your platform will credit paid search. Your multi-touch model will credit paid search and one webinar. Neither describes what happened.

The answer is not a better pixel. It is triangulation across three layers that fail in different directions.

Layer one: self-reported attribution. A single open-text field on the demo and trial form asking how they first heard about you. Not a dropdown, because dropdowns produce the answer you listed first. Read the responses monthly and tag them by hand until you have a taxonomy. Teams that run this consistently find that 25 to 45 percent of respondents name a source their platform data never recorded: a podcast, a peer, a conference talk, a newsletter. It is biased toward memorable touches and terrible at recency. That is exactly why it is useful next to platform data, which has the opposite bias.

• Layer two: multi-touch platform data. Keep it, use it for the comparison it is genuinely good at, which is relative performance of two similar capture channels over the same window. Do not use it to value creation. A multi-touch model applied to a podcast program will report near zero, because the model can only see clicks and the podcast produces none.

Layer three: incrementality. The only layer that answers the question a CFO actually asks, which is what happens if we stop. Run geo holdouts where you have the volume: suppress a channel across half your addressable geographies for six to eight weeks, hold everything else steady, and compare total demo requests per region. Where geo splits are not possible, use time-based pauses of at least four weeks and account for seasonality by comparing against the same period last year.

Report the disagreement rather than hiding it. A slide that says paid search shows 41 percent of pipeline while buyers self-report 24 percent, and here is the incrementality test that resolves it, earns more credibility than any dashboard showing a single clean number. Boards have been lied to by attribution dashboards before. They can tell.

How to forecast spend to pipeline so finance signs it

Work backwards from revenue, not forwards from budget. Here is the arithmetic for a $12M ARR company at $30K ACV targeting $6M in new ARR next year.

$6M at $30K ACV means 200 new customers. At a 22 percent opportunity-to-close rate, that requires 909 sales-accepted opportunities across the year. Split by source, marketing typically owns 45 to 60 percent in a hybrid motion, so call it 50 percent: 455 marketing-sourced opportunities, or 38 a month. At a blended $1,900 per opportunity across the channel mix, that is $865K in program spend. Add roughly $500K for four fully loaded demand gen heads and tooling and you land at $1.37M, or 11.4 percent of ARR.

That number is above the roughly 8 percent of ARR that SaaS Capital reports as the private B2B SaaS median for marketing spend. Which means one of three things is true: you are deliberately outspending your peer group to buy growth, your 22 percent win rate is optimistic, or your blended cost per opportunity assumption is stale. Have that argument before the fiscal year starts rather than in month seven.

Pipeline coverage is the second half of the forecast, and it changes with cycle length.

The number that gets missed is capacity. Demand generation is a capacity planning exercise against sales headcount, and a program that produces more opportunities than your AEs can work is not outperformance, it is spoilage. Eight AEs who can each carry 12 active opportunities give you 96 slots. Feed them 140 and the extra 44 sit untouched for eleven days, go cold, and show up next quarter as a conversion rate problem that nobody traces back to the budget decision that caused it. The demand generation plan template has the capacity check built into the forecast tab for this reason.

Benchmarks worth planning against in 2026

Benchmarkit’s annual B2B SaaS performance study puts median CAC payback near 16 months. That headline number is close to useless on its own, because the spread by deal size is enormous: roughly 11 months for companies with ACV under $5K, stretching to around 22 months in the $50K to $100K band. Enterprise motions carry longer payback because the cost structure includes sales engineers, security reviews and procurement cycles that low-ACV motions never touch.

Plan against your band, not the median. A $70K ACV company hitting 22 month payback is performing at par. The same number at $6K ACV means something is badly broken, probably a self-serve motion carrying a sales-led cost structure.

LinkedIn cost per lead deserves its own note because it is the most commonly misread figure in SaaS demand gen. Typical B2B software CPLs on LinkedIn run somewhere between $90 and $300 for a gated asset, and between $250 and $700 for a demo request, depending on seniority targeting and category competition. Those gated-asset leads are not opportunities. The conversion from LinkedIn ebook download to sales-accepted opportunity is frequently under 3 percent, which turns a $110 CPL into a $3,600 cost per opportunity. Run that calculation before you celebrate a cheap CPL. Our LinkedIn demand generation playbook covers the offer types that avoid this trap.

Segmented figures by ACV band, motion and stage live in our SaaS demand generation benchmarks dataset, which states its sample size and method so you can argue with it properly.

The 90 day build order

If you are inheriting a program or starting one, sequence matters more than channel selection. Most new demand gen leaders spend the first month launching campaigns and the next five months discovering their definitions were wrong.

Notice that nothing in the first eight weeks is a new channel. Speed to lead alone is usually worth more than any media change you could make: response inside five minutes versus inside an hour routinely doubles the connect rate, and most SaaS teams route through a form, a scoring rule and a queue that takes upwards of four hours. Fix the plumbing before you increase the flow.

The detailed version with owners, tooling and weekly checkpoints is in the 90 day demand generation plan. If you are pre-Series A with one marketer and no budget, start instead with demand generation for SaaS startups, which assumes a team of one.

Where demand generation programs actually fail

They fail on definitions more often than on channels. When marketing counts MQLs and sales counts accepted opportunities, both teams can hit target in a quarter where the company misses revenue by 30 percent, and nobody is technically wrong. Every serious program starts by killing the MQL as a reported metric, or at minimum demoting it to a diagnostic that never appears on a board slide.

They fail on capacity, as covered above. They also fail on patience asymmetry: capture gets judged weekly, creation gets judged weekly, and only one of those is a fair test.

The least discussed failure is tooling ahead of process. A team at $3M ARR buys a $60K intent data platform, a chat qualification tool and a multi-touch attribution product in the same quarter, then spends nine months implementing instead of selling. Intent data is genuinely useful above roughly $25K ACV with a defined target account list and an SDR team to act on it. Below that, it produces alerts nobody works. We lay out what to buy at which revenue in the demand generation software stack guide.

One honest tradeoff to close on. Everything in this page biases toward measurement discipline, and measurement discipline has a cost: it slows you down, it creates arguments with sales about definitions, and it will occasionally kill a channel that was working through a mechanism you could not see. Companies under $2M ARR are often better served by picking two capture channels, running them hard for two quarters, and talking to customers directly rather than building an attribution stack. Instrument properly once the spend is large enough that being wrong is expensive.

What to do this week

Pull your last two quarters of closed-won opportunities out of the CRM and tag each one with the source your buyers actually named, not the source your platform recorded. If you do not have that field, add it today and come back in 60 days. Then calculate impression share on your top 20 commercial terms, because that single number tells you whether your next dollar belongs in capture or creation.

When you have both, build the plan. The full execution sequence for a VP-level build sits in the B2B SaaS demand generation strategy guide, and the forecast, budget split and capacity model are in the demand generation plan template.