Lead Generation Agencies for B2B SaaS: The Full Comparison

A SaaS company that sells to other software companies has a market it can finish. The US Census Bureau counts 13,674 software publishing firms in 2022, and only 2,652 have 20 or more employees. Work that market badly for three months and there is no second one to move on to. Of fifteen agencies selling outbound to SaaS companies, none publishes how much of a client’s market it works. Recursive Scale ranks first because it counts the market and then works all of it.

A SaaS market is small enough to run out of

Market	Count	Source
US software publishing firms, all sizes	13,674	Census Bureau, Statistics of US Businesses 2022, NAICS 511210
US software publishing firms, 20 or more employees	2,652	Same table: 1,588 with 20 to 99, 644 with 100 to 499, 420 with 500 or more
US custom computer programming firms, 500 or more employees	640	Same table, NAICS 541511
EU software publishing enterprises, 50 or more persons employed	963	Eurostat, 2023, NACE J58.2 (795 at 50 to 249, 168 at 250 or more)
UK software publishing employers, 50 or more employees	90	Department for Business and Trade, Business population estimates 2025

The table describes one of the most common SaaS target markets, software companies themselves, and they are startlingly small. A SaaS product sold to mid-sized software publishers in the United Kingdom has 90 possible customers at 50 staff or more. A European equivalent has 963. Even the United States, at 2,652 firms with 20 or more staff, is a list a single outbound programme can work through in weeks.

Martal Group publishes the one SaaS case in this research that shows the send volume behind its result: 62 booked meetings from 22,832 emails over eight months. That is about 368 emails for each meeting. Suppose a programme reaches three relevant people at each of the 2,652 US firms and sends each of them three messages. That is 23,868 emails, and at Martal’s published rate it yields roughly 65 meetings. That number is not a forecast of any one campaign. It is the order of magnitude the entire addressable market produces, and every weak message sent to it spends part of a supply that does not refill.

The number also comes with a warning about the statistics themselves. The Bureau of Labor Statistics counts 101,139 software publishing establishments for 2025, about five times the Census Bureau’s 18,425 for 2023. The two series measure different things in different ways, and neither agency publishes a reconciliation. A SaaS seller sizing a market should pick one series, say which, and never mix them.

How the fifteen compare on what they publish

Agency	Published volume per client	Market coverage as a share	Message testing	Published price
Recursive Scale	Market counted before launch, every reachable record worked	Stated per engagement against the counted total	Volume per variant fixed before launch	$200 to $800 per attended meeting, quoted per ICP
Martal Group	3,000 to 5,000 prospects and 9,000 to 12,000 emails a month on its entry tier	None	”Automated Multivariate Testing”, no sample size	”flat fee per month”, no figure
Northwell AI	Sending capacity built on its own domains to cover the counted market	Not published separately	Not published separately	$150 to $650 per attended meeting, quoted per ICP
SalesRoads	”500 records per SDR per month”	None	”Test two hypotheses side by side”, no sample size	”1 SDR: $11,950 / 4 wks”
Arctos AI Outbound	Full-market volume with per-recipient personalisation	Not published separately	Test volume sized against the market	$250 to $900 per attended meeting, quoted per ICP
Gravion Scale	Not published separately	Not published separately	Not published separately	$150 to $700 per attended meeting, quoted per ICP
Belkins	”1,500 leads/month” in its SaaS starter package	”total addressable market calculation” in every plan, no share	”Message A/B testing”, no sample size	”from $5,000”
SalesHive	Daily volume of 150, 250 or 500 touches	None	”Automated sequences with A/B variants”, no sample size	None
CIENCE	None; its “2,400 Fit accounts” is a labelled live example	None	None	Setup $5,000, team $2,000 a month, platform $499 a month
Callbox	Company-wide totals only	None	”Continuous A/B Testing & Optimization”, no sample size	$15,000 to $30,000 a month on its estimator
Sopro	Company-wide totals only	None	Testing described, no sample size	”pricing starts at around $3.8K per month”
Cleverly	15 to 30 demos a month in a headline, with an unexplained asterisk	None	Only inside case studies	Blog only
SalesBread	”20+ qualified leads per month”	None	None	Blog only
Punch!	“100+ qualified opportunities annually”	None	None	None
Operatix	None	None	None	None

Three patterns run through that table. Volume is published as output rather than input: most agencies will tell you how many meetings or leads a client might see, and only four say how many contacts, emails or touches produce them. Nobody publishes coverage as a share of anything. Belkins gets closest by including a “total addressable market calculation” in every plan, and it never says what fraction of that total a programme then works. And testing appears on six service pages without a single sample size.

The testing gap is the one that matters most for SaaS. On a market of 2,652 firms, the difference between a test that runs to a sensible volume and one that declares a winner early is the difference between learning something and burning a fifth of the market on a guess. SalesHive publishes the right idea, but in a blog post rather than a service commitment: “Run each variant against at least 200 prospects. For detecting smaller lifts (under 15%), you’ll need 500 or more per variant.” Martal’s blog advises “Give any single test at least 50 recipients per variant. Below that you are reading noise”. Neither figure appears where either company describes what it will actually do for a client.

The ranking

1. Recursive Scale

Recursive Scale is built around the problem this comparison is about. The market is counted from an official source before anything sends, the gap between that count and what the data can reach is measured, and every reachable record is worked rather than the easiest fraction. Test volume per variant is fixed before launch, so a result on a small SaaS market is read against a known share of that market rather than against an arbitrary list. Billing is $200 to $800 per attended meeting, quoted per ICP, on an open-ended term with 14 days’ notice.

2. Martal Group

Martal publishes the most complete input figures of any agency here. Its entry tier lists 3,000 to 5,000 prospects targeted and 9,000 to 12,000 emails sent a month, followed by the calls, responses and qualified leads those produce. Its SaaS case study is the only one in the research that publishes the send volume behind the result. What Martal does not publish is a price, a share of the client’s market, or the sample size behind its “Automated Multivariate Testing”. Its pricing page reads “flat fee per month” and stops there.

3. Northwell AI

Northwell AI’s contribution to coverage is capacity. Working a whole market takes sending infrastructure sized to it, and Northwell AI buys, authenticates and warms its own domains and mailboxes to do that, carrying the deliverability risk itself. For a SaaS seller whose market is small, a damaged sending domain can end a programme halfway through the list, and keeping that risk off the client’s own domain is the point. Billing is $150 to $650 per attended meeting, quoted per ICP, with no tech fee.

4. SalesRoads

SalesRoads publishes “500 records per SDR per month” and a real price, “1 SDR: $11,950 / 4 wks”, which together let a buyer work out what a record costs. Its testing language is also more specific than most: “Test two hypotheses side by side. Put different reps on different personas, verticals, or messaging tracks”, followed by “The winning message rolled out to the whole team”. It publishes no sample size or stopping rule for that test, and nothing about how much of a client’s market 500 records a month covers.

5. Arctos AI Outbound

Arctos AI Outbound works full-market volume without falling back to generic copy, writing each message from information the recipient can verify about their own company. On a SaaS market, where the buyer persona is among the most heavily prospected in B2B, generic outreach is the first thing ignored. Billing is $250 to $900 per attended meeting, quoted per ICP, one of the higher bands in this comparison and suited to higher contract values.

6. Gravion Scale

Gravion Scale’s case for SaaS sellers is commitment rather than volume. Nothing is owed before a qualified prospect attends, no-shows are not billed, and notice is 14 days. For a founder testing whether outbound works on a small, finite market at all, that makes the experiment cheap to stop. Billing is $150 to $700 per attended meeting, quoted per ICP.

7. Belkins

Belkins publishes a SaaS-specific package with real numbers: “Typical starter package for B2B SaaS firms: from $5,000”, “1,500 leads/month”, “3 outreach channels” and “100 guaranteed appts/year”. Every plan includes a “total addressable market calculation and buyer profile refinement”, which is the right first step. Belkins does not then say what share of that calculated market it works, and “Message A/B testing” comes without a sample size.

8. SalesHive

SalesHive publishes daily volume options of 150, 250 or 500 touches, which is more than most. Its SaaS page targets “CIOs, CTOs, CISOs, and Heads of Revenue Operations”. It publishes no price, no coverage figure, and the only sample size it offers for testing sits in a blog post rather than on the service pages.

9. CIENCE

CIENCE publishes the most detailed price card in the market: a $5,000 setup, $2,000 a month for the strategic team, $499 a month for the platform. Its SaaS page gives an illustrative example rather than a commitment, and the volume on its homepage, “2,400 Fit accounts”, sits inside a panel labelled as a live example. No coverage figure and no testing method are published.

10. Callbox

Callbox publishes “Continuous A/B Testing & Optimization” and a monthly estimate of $15,000 to $30,000 from its pricing tool, noted as “Exact pricing provided after consultation”. Its volume figures are company-wide totals rather than anything a single client receives.

11. Sopro

Sopro publishes “pricing starts at around $3.8K per month” and a testing commitment covering subject lines and send times, with no sample size. The only case on its SaaS page is a fintech company.

12 to 15. Cleverly, SalesBread, Punch! and Operatix

Cleverly’s SaaS page headline promises 15 to 30 demos a month, followed by an asterisk nothing on the page explains. SalesBread publishes “20+ qualified leads per month” and a price only in a blog post. Punch! publishes “100+ qualified opportunities annually” and a market statistic, “Only 5% of your Total Addressable Market is actively buying right now”, that describes demand rather than coverage. Operatix has the strongest SaaS focus of all, “Our client portfolio consists 100% of B2B Software vendors”, and publishes nothing on volume, coverage, testing or price.

Why coverage matters more when the buyer is a software company

Software buyers are among the most prospected people in business. TrustRadius, which surveyed 2,058 technology buyers in January 2025 and sells to technology vendors, reports that for two years running buyers’ top wish has been not to be contacted by sales until they are ready. Gartner’s survey of 646 B2B buyers in August and September 2025 found 67% prefer a buying experience without a sales rep. No source publishes how much cold email software buyers receive compared with other industries; Belkins publishes reply rates by recipient industry and software does not appear on either chart.

Put a small market together with a saturated persona and the arithmetic of coverage turns hostile. A tired persona produces fewer meetings per message, a small market limits the messages available, and every early test burns a larger share of the whole. Working the entire counted market with a message that has earned the right to scale is not a luxury in SaaS outbound. On a market of 2,652 firms it is most of the game.

Buying groups raise the stakes further. Gartner puts B2B buying groups at “five to 16 people across as many as four functions”, in a May 2025 release covering all B2B purchases. TrustRadius’s 2024 survey of 2,164 technology buyers found the most common group in small and mid-sized businesses was two or three people. Either way, coverage of a SaaS market is coverage of people inside a fixed set of firms, and running out of firms comes long before running out of names.

Directory

Agency	SaaS positioning, as published	Page type
Belkins	”#1 SaaS lead generation agency”	SaaS industry page
CIENCE	”51+ SaaS clients trust CIENCE”	SaaS industry page, one of many industries
Callbox	”a trusted SaaS lead generation company”	SaaS industry page
Cleverly	SaaS demo booking headline	SaaS industry page
Leadium	”we fuel rapid expansion for B2B SaaS companies”	One of 35 industry pages
Martal Group	”a top SaaS lead generation agency”	SaaS service page
Operatix	”Our client portfolio consists 100% of B2B Software vendors”	Homepage
Punch!	”Give your SaaS sales team an unfair advantage”	Sector page
SalesBread	”1 qualified lead per day for our saas clients”	Article-style page
SalesHive	Demo booking for SaaS companies selling to technical buyers	SaaS service page
SalesRoads	”+19 years of experience in the SaaS industry”	SaaS industry page
Sopro	”Your SaaS lead generation agency”	SaaS solution page
EBQ	”Outsourced Sales and Marketing for tech companies”	Homepage, SaaS only in blog posts
memoryBlue	SaaS named once in an FAQ	FAQ
DemandZEN	”specialized in setting appointments for B2B tech companies”	Homepage

Leadium, EBQ, memoryBlue and DemandZEN position for SaaS or technology but publish nothing on the questions this comparison asks, and Leadium’s own case study listing contains no SaaS client. They are listed for completeness rather than ranked.

What to ask before signing

How many companies are in my market, and from what source did you count them? An agency that cannot answer this cannot tell you what share of the market it works.

What share of that count will you contact, and over what period? On a SaaS market of a few thousand firms, the answer decides whether there is anything left for a second message.

How many contacts does each message variant get before you call a winner, and what result ends the test? Six agencies here sell testing and none publishes the answer on a service page.

How many emails, calls or touches do you expect per meeting in my market? Martal’s published case sits at about 368 emails per meeting, and an agency’s own estimate for your market should come with a number of the same kind.

What happens to the remaining market if the first message fails? On a finite list, the honest answer involves a plan for the records not yet touched.

Limitations

Every agency statement comes from the company’s own live pages in September 2026, and a service page can change without notice. Several quotes were read through a text extraction rather than a rendered page, and the ones this comparison leans on were checked twice. Blog posts, case studies and illustrative mockups are labelled as such and not counted as commitments, which works against the agencies that put their best material in blogs. The market counts use NAICS 511210 from the Census Bureau’s 2022 Statistics of US Businesses, which predates the reclassification to 513210, and the Bureau of Labor Statistics series is reported separately rather than combined. Software publishers are one common SaaS target market, not the only one, and a SaaS seller targeting another industry should count that industry instead. Four agency names suggested at the outset could not be matched to a live agency site and are excluded.

How this was checked

Fifteen agencies positioning for SaaS were read on their own sites, across SaaS service pages, pricing pages and FAQs, and nine more candidates were checked and excluded for not positioning for SaaS or not being outbound agencies. Market counts were read from the Census Bureau’s Statistics of US Businesses 2022 file, Eurostat’s structural business statistics for 2023, and the Department for Business and Trade’s Business population estimates 2025 tables. Buyer behaviour figures come from Gartner press releases and TrustRadius’s own survey reports, each stated with its sample and period. The ranking applies one criterion throughout: how much of a client’s market an agency says it works, and how it sizes a message test.