Site Speed Is a Marketing P&L Line, Not an Engineering Concern

The Jira board is where website performance lives at most B2B SaaS companies. That is the structural problem. Marketing owns the revenue the site is supposed to produce. Engineering owns the technical health of the site. Nobody owns the gap between them, and that is where the speed cost compounds, invisibly, every quarter.

This piece is the opening of the Performance & Speed pillar. It argues that site speed at growing B2B SaaS companies is a marketing-finance question, not an engineering one. The data on conversion lift, the changes to Core Web Vitals since 2024, the ranking-and-citation implications, and the math on what slow pages cost at modern customer-acquisition prices all point the same direction. Performance is on the marketing P&L. Most marketing teams just have not put it there yet.

Site Speed Is a Conversion Lever, Not a Technical Detail

In 2021, Vodafone Italia ran a controlled experiment that improved Largest Contentful Paint by 31% on its paid-search landing pages. Sales lifted 8%. Lead-to-visit lifted 15%. Cart-to-visit lifted 11%. The test ran with approximately 100,000 clicks and 34,000 daily visits per arm, so the statistical power was high^[3]. The result was causal, not correlated. The size of the lift from a 31% LCP improvement is the easiest data point in this article to underestimate.

The pattern shows up across studies. In 2020, Deloitte’s Milliseconds Make Millions analysis of 37 retail, travel, luxury, and lead-gen brands found that a 0.1-second improvement in mobile site speed raised retail conversion rates by 8.4% and travel by 10.1%, with average order value climbing 9.2%^[1]. The lift compounds per 100ms, which means a site cut from 4 seconds to 2 seconds is moving through 20 increments of that lift, not just one.

For B2B specifically the curve is steeper. Portent’s 2022 study of more than 100 million page views across 20 sites (14 of them B2B lead-gen) found that a site loading in 1 second converts roughly three times higher than one loading in 5 seconds, and roughly five times higher than one loading in 10 seconds^[2]. The conversion penalty for slowness is paid by every visitor on every page, not just the ones who bounce.

A B2B SaaS marketing team I worked with had a homepage averaging 4.2 seconds to interactive on mobile. They were running a six-figure paid program against that homepage. Nobody on the marketing team had run PageSpeed Insights against the URL in over a year, because nobody owned that question. The team was inside one optimization window every month: paid landing pages, ad copy, audience segments. The site itself was not a metric anyone reviewed. That is the pattern this pillar argues against.

What Core Web Vitals Actually Measure

Core Web Vitals are three specific measurements Google uses to evaluate page-experience quality: Largest Contentful Paint (LCP) for loading, Interaction to Next Paint (INP) for responsiveness, and Cumulative Layout Shift (CLS) for visual stability. The official thresholds are LCP at or under 2.5 seconds, INP at or under 200 milliseconds, and CLS at or under 0.1^[5]. These are the same numbers PageSpeed Insights reports, the same numbers Search Console flags, and the same numbers AI Overview systems weigh when deciding whether to fetch and quote a page.

LCP measures how long it takes the largest visible content element on a page to render. For most marketing pages that is the hero image or the H1 with its surrounding block. LCP is a proxy for “how long until this page looks ready.” A bad LCP score (over 4 seconds) means visitors are staring at incomplete content while their decision to stay or leave is forming.

INP replaced FID as a Core Web Vital on March 12, 2024^[4]. The change matters. FID only measured the delay before the first interaction. INP measures the delay before every interaction across the page lifetime and reports the 75th-percentile worst result. INP is structurally harder to pass because a single slow interaction in a session can drag the whole score. Many sites that were green on FID went yellow or red on INP overnight. Marketing teams whose performance audits predate March 2024 are reading old scorecards.

CLS measures how much visible page content moves around as the page loads. The classic failure mode is an image without a declared height pushing text downward just as the visitor goes to click. A high CLS score (over 0.25) means visitors are clicking things they did not mean to click and missing things they did. The conversion cost is direct.

Marketers need to know these three because they are the vocabulary Google uses to evaluate the marketing site. Page experience is a measurement layer, not just an engineering concept. The teams operating without this vocabulary cannot read their own performance scorecards. That is part of the Analytics Trust Gap: not knowing what is being measured, on which dimensions, and what good looks like.

How the Marketing Web Looks in 2026

In 2025, 48% of mobile sites passed all three Core Web Vitals, up from 44% in 2024 and 36% in 2023. The HTTP Archive Web Almanac 2025 Performance chapter, drawing on the Chrome User Experience Report’s real-user data across millions of sites, puts the trend clearly^[6]. Half the marketing web is still in “needs improvement” or worse. The trend is positive but slow.

Broken out by individual Vital, the pass rates for mobile in 2025 are 62% for LCP, 77% for INP, and 81% for CLS^[6]. LCP is the load-bearing failure mode. Most sites are mostly stable visually and mostly responsive to interaction, but they take too long to render the largest visible element. That is a hosting question, an image-optimization question, and a server-response question, in approximately that order.

At the company stage this blog writes about (20-200 person B2B SaaS), the pass rate is materially lower than the global average. The typical marketing site is WordPress, built on a theme bought once and rarely updated, with a stacked plugin layer added incrementally over three to five years and a hosting tier chosen for cost. Of the B2B SaaS marketing sites I have audited in the last two years, fewer than a third passed Core Web Vitals on mobile when I started. Most were closer to two-thirds of the Vitals failing simultaneously.

The state-of-the-web data is the backdrop. It is the answer to “are we behind?” Most B2B SaaS marketing teams that run the check find they are. The next question, the operational one, is “why?” The answer is usually the Website Ownership Gap. Nobody on the marketing team owns the question, nobody on the engineering team prioritizes it, and the site decays inside the gap between them.

The Mobile-Desktop Gap Is the Diagnostic

In 2025, 97% of desktop pages passed good INP. Only 77% of mobile pages did^[6]. That 20-percentage-point gap is the most useful diagnostic in B2B SaaS marketing performance. Your mobile visitors are getting a different site than your desktop visitors. The desktop dashboard you check every morning does not show this.

The 2017 Think with Google / SOASTA research on 900,000 mobile landing pages, while older, remains the most-cited figure on the cost of mobile slowness: 53% of mobile site visitors leave a page that takes longer than 3 seconds to load^[7]. The exact percentage is debated; the directional point is not. Mobile visitors abandon faster than desktop visitors, and the modern mobile web has gotten heavier, not lighter.

This matters more than it looks because B2B SaaS marketing traffic is mobile-heavier than most teams realize. The dashboards optimize for desktop because the marketers running them work on desktop. Mobile share at typical B2B SaaS marketing sites runs 40-60% depending on the audience and the channel mix. Half your traffic is getting the worse version of the site, and the half you observe directly looks fine.

A Series B SaaS marketing lead I worked with had a homepage that scored 1.8 seconds on desktop and 6.2 seconds on mobile. The desktop number was the only one she had seen in nine months because that was the one in the team’s monthly review deck. The mobile number was where the conversion damage was happening. After we surfaced it, the structural issue was visible. Most of the page weight was hero video that was unnecessary on mobile but loaded anyway because the build pipeline did not branch.

The diagnostic move for any marketing team is simple. Open pagespeed.web.dev, paste the URL, toggle to mobile (the default in 2026), and read the field data section. Compare it to desktop. If the two diverge significantly, the marketing team is making decisions on the better-looking dashboard. Run this monthly with the same discipline applied to the 30-minute GA4 audit. The two together cover most of the measurement-layer health a marketing team needs.

Performance Is Now Also an AI-Citation Signal

Google’s official documentation in Search Central states that Core Web Vitals are part of what its core ranking systems reward^[5]. That has been true since the page-experience update in June 2021. The signal is not a hard penalty; it is one input among many. In practical terms it tips the scales between two otherwise equivalent pages.

What is new in 2026 is that the same page-experience signal increasingly shapes whether AI assistants will fetch and quote a page in their answers. ChatGPT search, Perplexity, Claude’s web tool, Google AI Overviews: all rely on a fetch-then-extract pipeline. Pages that fail to load in time get demoted in that pipeline. Pages with layout shift get malformed extractions. Pages with poor INP cannot be interacted with by the headless browsers that some agentic crawlers use to evaluate content.

The compound is what changed the calculation. A slow page in 2021 cost you SERP position. A slow page in 2026 costs you SERP position AND AI assistant citations AND AI-referred discovery-call traffic. The DataGrail signal this blog tracks (B2B prospects discovering Yasser Soliman through AI chat) is the leading indicator of how much that second cost matters. If your page does not load, the AI does not cite. If the AI does not cite, the buyer never finds you.

The structural argument loops back. Performance is a marketing concern because the marketing pipeline now includes AI-mediated discovery, not just SERP CTR. The same audit work that fixes the conversion loss also fixes the citation gap. The team that owns this gets both lifts. The team that does not loses both, slowly, every quarter, with no clear single point of failure to blame.

The Business Case for Marketing Owning Speed

Median B2B SaaS customer acquisition cost reached approximately $1,200 per customer in 2025, up roughly 14% from 2023, with the cost to acquire each dollar of new ARR running around $2^[8]. At those CAC levels, every conversion lost to a slow site costs more this year than last. Performance is a marketing P&L line because the marketing P&L is where the cost lands.

The math is uncomfortable when made specific. A typical B2B SaaS marketing site at the 20-200 employee stage runs ~15,000 monthly mobile sessions, converts demo requests at ~1.5%, closes demos to customers at ~20%, and produces ~45 new customers per quarter from organic and paid combined. A speed fix that lifts conversion 8% (the lower bound of what Vodafone and Deloitte data predict for a meaningful LCP improvement) adds roughly 4 customers per quarter. At $1,200 CAC, the conversion lift is worth roughly $19,200 per year in implicit acquisition savings, and that is before counting the demo-to-close improvements that compound from better lead quality.

That math is the business case. It is rarely the case marketing teams hear because the engineering team frames performance work as platform maintenance and the marketing team has no internal advocate translating Vitals into pipeline. The difference between maintenance and ownership is exactly this gap: maintenance keeps the lights on, ownership notices what the lights are obscuring.

The fix is not “hire a Core Web Vitals specialist.” It is to give one person the ownership of marketing-site performance as a measured outcome, accountable across the audit-and-improve loop. That role can take three forms: a full-time hire (with judgment scope spanning the marketing site as a full-time concern), an agency retainer (where multiple specialists rotate through your account on a shared services model), or an embedded WebOps lead (where one accountable individual carries judgment continuity across quarters). The three ownership models walks through which form fits which company stage. The point of the pillar is upstream of the operational comparison: the cost of NOT having any of those models is large, measurable, and on the marketing P&L every quarter.

The marketing team that owns this gets the conversion lift, the AI-citation lift, and the ranking lift, all from the same audit work. None of the three are visible until someone measures them, and most B2B SaaS marketing teams are not measuring them. That is the gap this pillar exists to close.

Sources

1. Deloitte Digital, Milliseconds Make Millions (commissioned by Google) – March 2020 mobile site speed study across 37 retail, travel, luxury, and lead-gen brands in EU and US; 0.1s improvement raised retail conversion 8.4%, travel 10.1%, average order value 9.2% ↩
2. Portent, Site Speed Is (Still) Impacting Your Conversion Rate – April 2022 study; 100M+ page views across 20 sites (14 B2B lead-gen, 6 B2C ecommerce); B2B sites loading in 1 second convert ~3x higher than 5-second sites, ~5x higher than 10-second sites ↩
3. Google web.dev, Vodafone case study: A 31% improvement in LCP increased sales by 8% – March 2021 controlled A/B test on Vodafone Italia paid landing pages; approximately 100,000 clicks and 34,000 visits per day per arm; LCP improvement of 31% drove an 8% sales lift, 15% lead-to-visit lift, 11% cart-to-visit lift ↩
4. Google web.dev, Interaction to Next Paint becomes a Core Web Vital on March 12 – Official Google announcement, January 31 2024; INP replaced FID as a Core Web Vital on March 12 2024 ↩
5. Google Search Central, Understanding Core Web Vitals and Google Search Results – Official documentation, last updated December 10 2025; good thresholds LCP ≤2.5s, INP ≤200ms, CLS ≤0.1; CWV are part of what Google’s core ranking systems reward ↩
6. HTTP Archive Web Almanac 2025, Performance chapter – Published January 15 2026, updated May 5 2026; CrUX July 2025 data across millions of sites; 48% mobile sites pass all three Core Web Vitals (up from 44% in 2024 and 36% in 2023); mobile-desktop INP pass gap of 20 percentage points (77% mobile vs 97% desktop) ↩
7. Think with Google / SOASTA Research, New Industry Benchmarks for Mobile Page Speed (Daniel An) – February 2017 analysis of 900,000 mobile landing pages across 126 countries using a neural net with 90% accuracy; 53% of mobile users abandon a page that takes longer than 3 seconds to load ↩
8. First Page Sage, Average Customer Acquisition Cost (CAC) By Industry: B2B Edition – 2025 update; median B2B SaaS CAC approximately $1,200 per customer, up roughly 14% from 2023; aggregated benchmark across B2B SaaS clients ↩

Written by Yasser Soliman

Technical Marketer

I've spent 5+ years embedded in marketing teams at B2B SaaS companies. I own the marketing website — performance, analytics, SEO, integrations — so your team ships without bottlenecks.