The State of SaaS Marketing in 2026

SaaS marketing in 2026 is fundamentally a different practice than it was three years ago. AI Overviews now intercept ~40% of informational search queries before they reach organic results. Paid search CPCs in major B2B SaaS categories have climbed 30-60% in two years. Cookie-based attribution is functionally dead for new acquisitions. And LinkedIn — for the first time in a decade — is producing meaningfully better outbound returns than cold email.

This report covers what we've observed across Makreate's growth-stage SaaS engagements over the last 12 months (US, UK, UAE), combined with public industry data from OpenView's SaaS benchmarks, Gartner reporting on B2B marketing trends, ChartMogul's SaaS metrics reports, and other cited sources.

Quick framing: we work primarily with growth-stage SaaS (Series A through C). Our observations skew to that segment. The patterns probably apply to bootstrapped SaaS and product-led growth motions too, but enterprise SaaS marketing is a different game we're not commenting on here.

1. CAC has structurally increased — and isn't coming back down

Industry-wide, blended CAC across SaaS has climbed 30-60% over the last 24 months depending on segment. OpenView's 2025 SaaS benchmarks put median B2B SaaS CAC at $1,800-$3,200 for self-serve and $12,000-$45,000 for sales-led depending on ACV. Both figures are up materially from 2022 levels and the trend is consistent across segments.

The drivers are well-understood: paid CPCs have inflated (more competitors bidding on the same intent keywords), AI Overviews are reducing organic click-through on informational queries, and the dark-funnel problem (buyers researching outside trackable channels — Slack communities, podcasts, AI chatbots) is structurally bigger. None of these reverse on the timelines anyone hoping for a CAC reset is hoping for.

The teams winning with this CAC reality aren't the ones trying to push CAC back down through aggressive optimisation; they're the ones investing in higher LTV and better retention. LTV:CAC mathematics have always been the unit economics that matters; in 2026, attacking the LTV side of that ratio is where the leverage is.

2. AI search is changing what gets clicks

Google's AI Overviews now appear on a meaningful portion of informational and commercial-investigation queries — public estimates range from 30-50% depending on category. The practical effect: the organic click rate on those queries has dropped 25-45%. For categories with heavy AI Overview presence, the SEO playbook has had to change.

What's working for SEO in 2026 isn't ranking #1 — it's getting cited in the AI Overview. The pattern is consistent: structured, well-scoped pages with explicit factual claims, clear authority signals (author, organisation, citations), and direct-answer formatting are the ones AI Overviews pull from. Long, meandering blog posts that win on link-equity alone are no longer winning the same traffic share.

ChatGPT, Perplexity and Claude are also producing meaningful referral traffic to SaaS websites. Public Perplexity data suggests it's driving 50-200M outbound clicks per month industry-wide. That traffic converts at higher rates than Google organic in our measurement — buyers reaching a vendor through an AI conversation are further down the funnel.

3. LinkedIn outbound is outperforming cold email

For the first time in a decade, well-run LinkedIn outbound is producing better B2B SaaS results than well-run cold email. Reply rates have flipped: a strong cold-email cadence might get 8-15% replies in 2026; a strong LinkedIn cadence is hitting 18-30% in similar ICPs.

The drivers: cold email deliverability has degraded materially as Gmail and Outlook tighten anti-spam (DMARC enforcement, automated quarantine of cold outbound), while LinkedIn's algorithmic signals favour authentic 1:1 conversation. The result is that the channels have effectively swapped pole position for top-of-funnel B2B outreach.

Email isn't dead — it's the workhorse for nurture, transactional, and warm-lead follow-up. But for cold outbound prospecting in 2026, LinkedIn-led multi-channel cadences are the higher-conversion play. Makreate's outreach engagements in 2025 shifted decisively toward LinkedIn-primary cadences with email as a secondary channel; the conversion improvement has been consistent across ICPs.

4. Content marketing is bifurcating

Content marketing in 2026 splits into two distinct disciplines that share almost nothing. The first is AI-search-optimised content: structured, authoritative, citable, designed to be cited in AI Overviews and chatbot conversations. The second is brand-and-community content: long-form thinking, opinionated takes, founder-narrated audio and video, that builds direct audience relationships.

The middle ground — generic SEO blog posts on commodity topics — is dead. AI-generated content has flooded that segment; Google has responded by deprioritising thin content; users have responded by routing around blog-spam to higher-trust sources. Mid-quality blog content is now the single worst content investment a SaaS team can make.

The teams winning at content in 2026 are running two parallel content programs: a tightly-targeted SEO-and-AI-search asset library (typically 30-100 pages, deeply researched, structured for citability) and a brand-led publishing motion (founder posts, podcast appearances, opinionated long-form). Same team often runs both; same writer rarely does both well.

5. Attribution is broken — and we're getting used to it

Multi-touch attribution as a discipline is in worse shape in 2026 than at any point since attribution became a category. iOS 14+ stripped the click-attribution chain for a meaningful portion of paid traffic; the dark funnel (Slack, podcasts, AI chatbots, peer recommendations) is structurally invisible; cross-device tracking is regressing as privacy regulations tighten in the EU, UK and California.

What's filling the gap: self-reported attribution at the form-fill level ('How did you hear about us?') is now the most reliable single signal for top-of-funnel attribution in our 2025 client measurements. It's noisier than multi-touch attribution claimed to be, but it's actually directionally correct, which is more than most MTA models can credibly claim.

Incrementality testing — holding out a segment from a channel and measuring the delta — has become the gold standard for confirming whether a channel is actually working. It's expensive (you have to literally not spend on the channel for a measurable population), but it produces signal that no model can match.

6. What good SaaS marketing teams are doing in 2026

A few patterns show up consistently across the SaaS marketing teams that are winning right now. First, they've made LTV a leading metric, not just CAC — channel decisions get evaluated on LTV:CAC ratio, not CAC alone. Second, they run a thin SEO-and-AI-search content program and a thick brand-and-community program in parallel, with different writers and different success metrics. Third, they treat self-reported attribution as the source of truth at the top of funnel and use multi-touch as a directional supplement. Fourth, they've moved primary outbound spend from cold email to LinkedIn cadences.

What's not working: chasing 2020-era CAC; investing in middle-of-the-road blog content; over-engineering attribution models that the data can't actually support; underinvesting in retention because acquisition feels more urgent.

7. What's coming for SaaS marketing in 2027

Two shifts look near-certain over the next 12-24 months. First, AI-generated content saturation will continue increasing, and Google's response will continue tightening. Original research, opinionated takes, and citable factual content will keep winning; generic blog content will keep losing. Second, AI-assistant referral traffic (ChatGPT, Claude, Perplexity, Copilot) will become a measurable acquisition channel for most SaaS teams — not a dominant one, but big enough to justify dedicated optimisation.

What probably won't happen: a return to pre-2023 CAC. The CAC floor has structurally moved up; the right response is to push LTV, not pine for cheaper acquisition.

What might happen: meaningful improvement in attribution as new identity-resolution and privacy-preserving measurement protocols mature. We'd be surprised if it solved the problem fully in 2027, but the gap should narrow.