Google Ads ROAS for non-branded B2B SaaS campaigns is 78%, below breakeven. In contrast, LinkedIn delivers 121%. This gap should prompt a reevaluation of budget allocation. However, the deeper issue lies in the metrics teams optimize against, which bear little relation to meaningful outcomes. The correlation of CTR with pipeline is r=0.09, cost per lead is r=0.23, and cost per SQL is r=0.71—the only metric that predicts downstream revenue. Despite this, many paid programs still prioritize the first two metrics. This is the playbook problem.

The Structural Shift Underneath the Numbers

Paid acquisition's share of B2B SaaS pipeline dropped from 34% in 2023 to a projected 26% by 2026. Platform costs are rising: LinkedIn ad costs increased 24% year-over-year, while Google saw a 19% rise. CAC payback has stretched from 12–15 months to 18. These trends indicate that while paid isn't dead, the tolerance for sloppy measurement has diminished. CFOs have taken notice. The pressure to report on contribution margin and pipeline coverage ratio, rather than MQLs or form fills, has intensified. Many teams lack the infrastructure to comply, creating a disconnect: marketing organizations run sophisticated campaigns but report as if it’s still 2019. The intent gap exacerbates this issue. Paid social traffic converts at 0.9% on identical landing pages where direct traffic converts at 3.3%. This disparity isn't a creative or bidding problem; it's a signal problem, and treating every visitor the same compounds it.

What the Operators Are Saying

Emily Kramer's recent MKT1 newsletter highlighted perspectives from four paid practitioners, revealing instructive tensions. Keith Putnam-Delaney, Co-Founder and CEO of Primer, stated, "The audience is one of the last meaningful levers you can control when AI features will literally redo all your creative for you." Conversely, Lizzy Marano, Director of Growth Marketing at Justworks, cautioned against hyper-segmentation: "Don't get too hyper-segmented. You're going to miss out on a lot of people in your target audience if you add too many filters." These statements may seem contradictory but aren't. Audience is a lever, yet the instinct to slice it into thinner segments backfires when platforms require volume to learn. The resolution lies in defining "audience"—tight enough to exclude waste but broad enough for algorithms to find signal. Max Hogan, Co-Founder of Closing Media, emphasized speed: "Paid fundamentals across targeting and messaging still work...but the 20% of teams that can execute and test faster, at higher volume and quality, with AI are crushing the other 80%." Kamil Rextin, Founder at 42 Agency, noted, "AI can help you scale creative once you have a good baseline, but it can't replace taste."

Where the Playbook Actually Breaks

Three specific failures account for much of the waste in B2B paid programs today: Optimizing to the wrong conversion event. Google's value-based bidding is practical only if teams pipe first-party CRM data back into the platform to optimize for pipeline creation rather than form fills. Most teams haven't built that infrastructure. The fix requires RevOps and demand gen to agree on definitions, but this handoff often stalls. Treating Google non-brand and LinkedIn as interchangeable line items. With ROAS at 78% for Google and 121% for LinkedIn, these channels need different goals, creative, and measurement windows. Running both against the same CPL target guarantees over-investment in whichever appears cheaper at the top of the funnel and under-investment in the one producing better pipeline. Ignoring AI search dynamics. AI Overviews have compressed paid CTR to about 16.2%, compared to 21.8% without them. Earned media is cited by AI systems 84% of the time; paid content, only 0.3%. Some paid budget could yield more durable returns if redirected toward placements that influence how LLMs reference your brand, such as sponsored content or YouTube sponsorships, rather than chasing increasingly expensive clicks.

The Experiment Worth Running This Week

Select one campaign and shift the optimization target from form fill to a qualified pipeline stage. Run a two-week holdout and measure lift against SQL creation, not lead volume. The hypothesis: tightening the conversion definition will raise CPA and lower volume, but improve cost per SQL. Set a stop-loss at a 30% CPA increase. If quality improves while volume contracts, that’s the expected trade-off, and your next move should be expanding reach with the same qualification signal. Success equals lower cost per SQL, with guardrails ensuring CPA doesn’t exceed the stop-loss. Avoid over-interpreting short-term volume decline. Paid still offers unmatched speed: speed to test positioning, enter new segments, and generate signal about what resonates. However, speed aimed at the wrong metric leads to incorrect conclusions. The 78% ROAS on Google non-brand and the 0.9% paid social conversion rate aren't arguments against paid; they highlight the flaws in how most teams optimize, focusing on metrics that predict little about revenue.