Sixty percent of sales development reps hit quota last year. That is the lowest reading on record in The Bridge Group's 2025 SDR Models, Motions and Metrics report, which surveyed 351 B2B companies.
Now set a second number from the same research beside it. Pipeline generated per SDR reached $3.78 million, up sharply from $2.83 million in the 2022 edition.
Fewer reps clearing quota, more pipeline per rep. Those two facts only coexist if the job changed underneath the people doing it, and it did. Median monthly quota per SDR has fallen 40% since 2018. SaaS companies stopped buying activity and started buying precision. The discipline that emerged to deliver it now has a name: GTM engineering.
The inbox became a licensed channel
On February 1, 2024, Google began requiring anyone sending more than 5,000 messages a day to Gmail accounts to authenticate with SPF and DKIM, publish a DMARC record, support one-click unsubscribe, and keep spam complaint rates below 0.30% in Postmaster Tools, per Google's own sender guidelines. Microsoft matched it on May 5, 2025 with equivalent requirements for high-volume senders hitting Outlook.com, Hotmail, and Live addresses at 5,000 messages a day or more. Microsoft went further than filtering: as dmarcian documented, non-compliant mail is now rejected outright with a 550 5.7.515 error rather than quietly filed in a junk folder.
Small technical change, large commercial consequence. For roughly a decade the marginal cost of an outbound email was effectively zero, which made volume the cheapest growth lever available to any SaaS company with a list and a sequencer. It is not free anymore. Sender reputation has become a gated asset with a maintenance cost, and a 0.30% complaint ceiling puts a hard limit on how loosely a team can target before the channel closes on it.
The operator caveat that matters here: adding sending domains does not fix this. It spreads the same complaint behavior across more assets, which is why teams running 30 or 40 rotating domains tend to discover the problem all at once rather than gradually. Domain sprawl buys a quarter. Relevance buys the channel.
Buyers show up later, and they show up informed
6sense's 2025 B2B Buyer Experience Report, built on nearly 4,000 responses, found buyers were 61% through their buying journey before contacting a seller, down from 69% the year before. That reads like good news for sales teams, and partly it is. The same study found 79% of vendor engagements are still buyer-initiated, and that the winning vendor came from the day-one shortlist 95% of the time.
Ninety-five percent reframes the whole prospecting question. If the deal is nearly always won by a vendor the buying group already had in mind on day one, then the job of outbound is not to interrupt an active evaluation. It is to be resident in memory before the evaluation starts. Volume programs are built for the first job. Almost nothing in a standard sequencer is built for the second.
Buyers are also arriving better armed. 6sense found 94% of them used large language models somewhere in the purchase process, while time spent evaluating each vendor compressed from 2.6 months to 2.0 months. Writing in Harvard Business Review in June 2026, Amit Joshi, Ivy Buche, and Caroline Schwaer argue that generative AI is moving discovery and evaluation into environments vendors neither own nor fully understand. That puts a premium on machine-readable credibility rather than sequencer throughput, and it rewards companies whose category position is legible to a model summarizing options for a buyer.
What a GTM engineer actually builds
Clay, which says it coined the GTM engineering title in 2023, raised a $100 million Series C led by CapitalG in August 2025 at a $3.1 billion valuation, more than double the $1.25 billion mark it carried six months earlier, according to Crunchbase News. Co-founder Varun Anand described the role as building scaled systems within set parameters, adding that instead of coding software, GTM engineers are “coding revenue.”
Category funding is a weak signal on its own. The stronger evidence is what the work produces. Vercel took ten inbound SDRs down to one in six weeks by deploying an AI agent, using roughly a quarter to a third of a single GTM engineer's time, as Vercel's Jeanne Grosser told Theory Ventures' Tomasz Tunguz in November 2025. Lead-to-opportunity conversion held flat. The gain came from response speed, which cut the number of touches needed per conversion, and the nine remaining reps moved to outbound rather than out the door.
That is the shape of the shift. Not fewer people in go-to-market, but a different allocation: systems absorbing the deterministic work, humans concentrated where judgment changes the outcome. SaaS Mag covered the earlier phase of this transition in its analysis of how product-led growth evolved into full-stack GTM.
The headcount math is a mix shift, not a cut
Emergence Capital's Beyond Benchmarks survey of more than 560 venture-backed B2B software companies, fielded in April 2025, found 36% had cut SDR and BDR headcount over the prior year while only 19% increased it, as reported by SaaStr. Account executives split differently: 25% decreased, 28% increased. Professional services grew at 34%. Sales development was the only go-to-market function where cuts outpaced additions.
Spending data tells the same story from the other side. SaaS Capital's March 2026 spending benchmarks, drawn from more than 1,000 private B2B SaaS companies, put median sales spend at 15% of revenue, up from 13% the year before, while marketing held flat at 8%. R&D was unchanged at 22%. Companies are not spending less on going to market. They are spending more, on fewer and more technical people.
The honest counterweight comes from Emergence's own 2026 edition, which reports that across every segment, non-AI companies still generate more revenue per employee, concluding that AI remains an investment story more than a productivity one. Any 2027 plan built on realized AI efficiency gains is running ahead of the evidence. Budget for the capability, not for the savings.
Where the automation stops working
Gartner surveyed 645 B2B buyers between August and September 2025 and found sales reps were 28 percentage points more likely than generative AI to advance a buyer to the next purchase step, and 39 points more likely to demonstrate that they understood the buyer's need, according to its research. In a separate release, Gartner predicts that by 2030, 75% of B2B buyers will prefer sales experiences that prioritize human interaction over AI, a reversal of the long drift toward rep-free digital buying.
SaaStr's Jason Lemkin put the operational version more bluntly in December 2025, reporting that AI SDRs delivered geometrically more volume for the same results. That is the trap worth naming. An AI sequencer pointed at a loosely defined ICP is not a productivity tool. It is a faster way to burn sender reputation, and with a 0.30% complaint ceiling the bill arrives inside a single quarter.
The teams getting a return are using automation for research and routing, not for persuasion. Gartner's parallel survey of 227 chief sales officers found organizations giving sellers AI-enabled next best actions were 2.6 times more likely to achieve commercial growth, and those prioritizing seller AI upskilling were 2.4 times more likely to post strong revenue growth. Gartner also expects 95% of seller research workflows to begin with AI by 2027, up from under 20% in 2024. Research is the part machines are good at. Convincing a skeptical CFO is not.
The metrics replacing dials and emails sent
Activity metrics survived as long as activity was free. With sender reputation now a constrained resource and buyers self-serving most of the journey, four measures carry more weight.
- Inbox placement rate and spam complaint rate, per sending domain. Track both against Google's 0.30% threshold and treat a rising complaint rate as a targeting defect rather than a copywriting problem.
- Day-one shortlist inclusion. If 95% of wins come from the initial consideration set, the number that matters is what share of your ICP's evaluations you enter on day one. Win rate measures the deals you were invited into. Shortlist inclusion measures the ones you never saw.
- Pipeline per go-to-market FTE. The Bridge Group's $3.78 million per SDR is the single-role version. The portfolio version, pipeline divided by all revenue-generating headcount plus tooling spend, is the one that survives a reorg.
- Fully loaded cost per qualified meeting. Include data, enrichment credits, domain infrastructure, and the GTM engineer's salary. Sequencer seats are cheap. A senior technical hire and a real data budget are not.
Two caveats before anyone puts these on a board slide. Pipeline per FTE flatters companies with long sales cycles and generous opportunity-creation rules, so pair it with win rate by source or it will drift upward while conversion rots. And day-one shortlist inclusion is only measurable through win-loss interviews. It does not fall out of the CRM, which is exactly why most companies do not track the number that predicts their revenue best.
What to change before 2027 planning
Three moves are doing the most work for SaaS teams heading into next year's plan.
Audit the sending infrastructure before you rewrite the sequences. SPF, DKIM, and DMARC alignment are table stakes at both Google and Microsoft now. A team that discovers a 550 5.7.515 rejection mid-quarter has already lost the channel for the weeks it takes to rebuild reputation.
Fund the role before the tooling. Vercel's result came from one engineer at partial allocation, not a platform migration. A GTM engineer with real access to the data warehouse and the CRM will find more pipeline in a quarter than another sequencer license will in a year.
Move pipeline sources off cold email entirely. Partner-sourced and marketplace-sourced pipeline both route around the deliverability constraint, which is part of why ecosystem-led growth and cloud marketplaces have moved from experiments to budget lines. A channel you do not control the gate on is a channel worth diversifying away from.
Frequently asked questions
What is a GTM engineer and how is the role different from sales ops?
A GTM engineer builds the automated systems that generate and route pipeline: enrichment workflows, signal detection, AI agents for inbound response, and the data plumbing underneath. Clay, which says it coined the term in 2023, frames the work as coding revenue rather than coding software. Sales ops maintains the process and reporting that an existing team runs on. A GTM engineer builds new revenue-generating machinery. The practical difference shows up in the skill set: SQL, APIs, and workflow design rather than forecast hygiene and territory planning.
Is cold outbound still worth doing for B2B SaaS in 2026?
Yes, but not at the volumes that worked in 2021. Google and Microsoft now enforce authentication and complaint-rate thresholds on anyone sending more than 5,000 messages a day, and 6sense found the winning vendor came from the buyer's day-one shortlist 95% of the time. That combination penalizes broad sequencing and rewards narrow, well-timed outreach to accounts showing real signal. The Bridge Group's data supports the shift: pipeline per SDR rose even as quota per rep fell 40% from 2018 levels.
Should we hire a GTM engineer or add more SDRs?
For most SaaS companies past roughly $5 million in ARR with an existing SDR team, the engineer is the better marginal hire. Vercel replaced ten inbound SDRs with one agent plus a QA rep in six weeks using a fraction of one engineer's time, and conversion held. The exception is a company that has not yet found repeatable messaging. Automation scales whatever you point it at, so a team still testing its ICP should buy human iteration first and automation second, once the pattern is clear.
What email authentication rules do SaaS companies need to meet now?
Senders pushing more than 5,000 messages a day to Gmail must have SPF and DKIM in place, a published DMARC record, one-click unsubscribe on marketing mail, and spam complaint rates below 0.30% in Postmaster Tools. Microsoft applies comparable requirements to Outlook.com, Hotmail, and Live at the same 5,000-per-day threshold, and rejects non-compliant mail outright with a 550 5.7.515 error instead of routing it to junk. Both thresholds count per mailbox provider, not per sending domain, so domain rotation does not exempt a program.
How should we measure outbound performance if activity metrics no longer work?
Replace dials and emails sent with four measures: inbox placement and complaint rate per sending domain, day-one shortlist inclusion from win-loss interviews, pipeline per go-to-market FTE, and fully loaded cost per qualified meeting including data and infrastructure. The first two are constraint metrics that tell you whether the channel is still open and whether buyers know you exist. The last two are efficiency metrics. Pair pipeline per FTE with win rate by source, or a long sales cycle will make it look better than it is.
The bottom line
Outbound did not break. It got priced. When mailbox providers set a quality floor and buyers do most of their homework before a first conversation, the cheapest input in the SaaS growth model stops being cheap, and the advantage moves to whoever can build systems that earn attention rather than buy it at scale.
That is a better trade for the industry than it first looks. Median sales spend rising to 15% of revenue while marketing holds at 8% is not a story about retrenchment. It is capital moving toward a function that is finally being engineered instead of staffed. The SaaS companies that treat pipeline generation as a systems problem, with a named owner, a real data budget, and metrics that measure whether buyers can find them, will compound that advantage for several years before it becomes table stakes.
