Enterprise technical SEO has a quiet crisis. The teams are larger, the platforms are more expensive, and the audits are more frequent than they have ever been. Yet when you measure the fundamentals at scale, most sites are failing the same six checks that decide whether a page can rank in Google and be retrieved by an AI engine. We wanted numbers, not anecdotes, so we ran the crawl.
The findings below are uncomfortable because they are not exotic. None of the six metrics require novel research to fix. They require ownership, a rendering budget, and someone whose job is the template rather than the page. What the data shows is that at enterprise scale, the gap between knowing the fix and shipping it is where most organic and AI visibility is lost.
Core Web Vitals: the 42% ceiling
Core Web Vitals remain a ranking factor, and in 2026 they are also a proxy for the render quality an AI crawler experiences when it fetches a page. On mobile field data, only 42% of the sites in our sample passed all three vitals. That is the headline the rest of the section unpacks, and it has barely moved since the last time we ran this benchmark.
The failure is not evenly distributed across the three metrics. Layout stability is close to solved: 74% of sites pass Cumulative Layout Shift, the payoff of a decade of guidance about sizing images and reserving space for ads. Largest Contentful Paint is the middle child at 61%, usually failing on unoptimized hero media and render-blocking third-party scripts. Interaction to Next Paint is where enterprise sites collapse. Only 44% pass INP, and it is INP, not LCP, that most often drags a site out of the passing bucket entirely. Heavy tag managers, hydration cost on component frameworks, and long main-thread tasks are the usual culprits, and they live in the template, not the content.
The practical read is that the marginal work has shifted. Teams that spent 2023 and 2024 fixing layout shift have banked that win, but they are now under-investing in interaction latency, which is both harder to diagnose and harder to attribute to a single owner. INP is a JavaScript problem, which means it is usually an engineering problem, which means it usually has no SEO owner at all.
Figure 1. Share of sites passing each Core Web Vital on mobile field data, plus the all-three pass rate. n=300 sites.
Indexation coverage is leakier than it looks
Indexation coverage is the share of a site's crawlable, index-worthy URLs that are actually indexed and eligible to appear in search. It is the metric enterprise teams most often assume is fine and most often are wrong about. Across the sample, median coverage was 68%, meaning nearly a third of pages that should be earning visibility are not, for reasons ranging from soft signals to hard exclusions the team never intended.
When we classified the reasons a page fell out of coverage, the pattern was consistent. The largest single bucket was duplication and canonical confusion at 34% of excluded URLs: faceted navigation, tracking parameters, and near-identical templated pages competing with themselves. Discovery gaps came next at 22%, pages orphaned from internal linking that a crawler can technically reach but rarely does. Explicit noindex and robots exclusions accounted for 19%, and in roughly a fifth of those cases the exclusion was accidental, a staging rule or a blanket directive that outlived its purpose. Quality-driven exclusions, where Google crawled a page and chose not to index it, made up 25%.
The number that should worry a leadership team is not the 68%, it is the composition. Two-thirds of the loss, the duplication and discovery buckets, is self-inflicted and fully within the team's control. It is not a Google judgment about content quality; it is architecture the site is shipping against itself.
| REASON FOR EXCLUSION | SHARE OF EXCLUDED URLS | IN YOUR CONTROL |
|---|---|---|
| Duplication / canonical conflict | 34% | Yes |
| Quality-driven exclusion | 25% | Partly |
| Discovery / orphaned pages | 22% | Yes |
| noindex / robots exclusion | 19% | Yes |
Redirect chains are the silent tax
A redirect chain is what happens when a URL points to another URL that points to a third before resolving. Each hop costs crawl time, dilutes link equity, and, on mobile connections, adds latency that feeds straight back into the LCP failures from Figure 1. In our sample, 57% of sites carried at least one chain of two or more hops on a meaningful share of their URLs.
Chains accumulate the way technical debt always does: quietly, through migrations. A domain move, then an HTTP-to-HTTPS switch, then a URL-structure change, then a trailing-slash normalization, and suddenly a link from 2019 traverses four hops to reach the page it was always meant to point to. We found sites where the longest live chain was five hops, and where 3.1x more redirects existed than the number of destination pages they served. That ratio is the tell: when redirects outnumber the pages they resolve to, the site is spending crawl budget on plumbing.
The fix is unglamorous and high-leverage. Flatten every chain to a single hop, point legacy URLs directly at their final destination, and audit the redirect map on every migration rather than layering a new rule on top of the old ones. It is the cheapest crawl-efficiency win available, and almost no enterprise site has done it recently.
Figure 2. Prevalence of redirect issues, share of sites in sample. n=300 sites.
Schema is present but rarely valid
Structured data is how a page tells an engine what it is in a language the engine does not have to infer. It matters more in 2026 than it did five years ago, because AI engines lean on it to extract self-contained facts. So the surprising finding is not that enterprise sites lack schema. It is that they have plenty of it, and most of it is broken. 81% of sites deployed some structured data, but only 39% had schema that validated cleanly on their priority templates.
The 42-point gap between deployment and validity is the story. The most common failures were missing required properties, invalid enumeration values, and orphaned references where a node pointed to an entity that did not exist in the graph. These are the errors that cause an engine to silently discard the markup, which is worse than having none, because the team believes the signal is being sent when it is not. Organization and Article schema fared best; Product, FAQ, and BreadcrumbList carried the most errors, often because they were generated by templates that had drifted from the current spec.
Validity, not coverage, is the metric to manage. A single clean Organization block that resolves correctly does more for retrieval than a dozen richly nested types that fail validation. We recommend teams cut their schema surface area to the types engines actually parse and then guard those types with automated validation in the deploy pipeline, so drift is caught before it ships.
| SCHEMA TYPE | SITES DEPLOYING | SITES VALID | VALIDITY GAP |
|---|---|---|---|
| Organization | 77% | 58% | -19 pts |
| Article / BlogPosting | 64% | 46% | -18 pts |
| Product | 52% | 27% | -25 pts |
| FAQPage | 48% | 24% | -24 pts |
| BreadcrumbList | 61% | 33% | -28 pts |
Crawl budget goes to pages nobody wants
Crawl budget is the finite attention a search engine spends on a site. It is rarely a constraint for a small site and almost always a constraint for an enterprise one, where millions of low-value URLs can starve the pages that actually drive pipeline. We measured crawl waste as the share of crawler requests, sampled from server logs where available and from crawl simulation otherwise, that landed on URLs with no business value. The median was 31%.
Nearly a third of every enterprise site's crawl allocation is being burned on parameterized duplicates, infinite faceted-navigation spaces, expired or out-of-stock pages, and internal search results that should never have been crawlable. Every one of those requests is a request the engine did not spend on a fresh comparison page or an updated pricing page. The waste correlates directly with the indexation shortfall from the second section: the same duplication that suppresses coverage is what consumes the crawl budget.
This is the metric where the fix compounds fastest. Blocking low-value URL patterns in robots.txt, adding canonical discipline to faceted navigation, and pruning dead pages does not just recover budget; it tightens the signal the engine receives about what the site is for. Sites in the cleanest quartile of our sample wasted only 12% of crawl on low-value URLs and indexed a markedly higher share of their priority pages.
Figure 3. Share of crawl requests hitting low-value URLs, by sample quartile. n=300 sites, median 31%.
JavaScript rendering still breaks in the wild
The last metric is the one that most directly gates AI visibility. Many AI crawlers fetch raw HTML and do not execute JavaScript the way Googlebot's rendering service does. So when a page's primary content only exists after client-side hydration, the engine sees an empty shell. We compared the raw HTML served to the rendered DOM for every site and flagged those where primary content, the body copy, headings, and links that carry meaning, did not appear without JavaScript. 23% of sites failed.
The severity varied. In the worst cases, the entire main content region was injected client-side, so a non-rendering crawler received a page with navigation and a footer and nothing in between. More common was partial failure: the hero and first section rendered server-side, but comparison tables, FAQs, and the deeper content that engines most want to cite were loaded on interaction or scroll. That partial pattern is deceptive because the page looks fine to a human and to Googlebot, while being functionally invisible to the AI engines a B2B buyer now uses first.
Framework choice was not destiny here. Sites on the same rendering frameworks landed on both sides of the line, separated entirely by whether the team had configured server-side rendering or static generation for content routes. The 23% that failed had shipped client-only rendering for content that needed to be in the initial HTML. The fix is a rendering-strategy decision, not a rewrite, and it is the single highest-leverage change a JavaScript-heavy site can make for GEO.
| RENDERING OUTCOME | SHARE OF SITES | AI-RETRIEVABLE |
|---|---|---|
| Full content in raw HTML | 62% | Yes |
| Partial content client-side | 15% | Degraded |
| Primary content client-side | 8% | No |
| Full page client-side | 15% | No |
What this means for you
Read the six metrics together and a single conclusion emerges: enterprise technical health is not a content problem or a link problem, it is an ownership problem. Every failure in this study, the 58% that miss on Core Web Vitals, the 32% indexation shortfall, the redirect chains, the invalid schema, the wasted crawl, and the client-side rendering, sits in the seam between the SEO team and the engineering team. The pages are fine. The templates, the infrastructure, and the rendering strategy are where visibility leaks.
The prioritization writes itself from the data. Fix rendering first if any of your priority content is client-side, because it is the only metric that can make a page completely invisible to AI engines rather than merely disadvantaged. Then recover crawl budget and indexation together, since they share a root cause in duplication. Then flatten redirects, harden INP, and validate the narrow set of schema types you actually rely on. None of this is novel. All of it is deferred, and the deferral is the cost.
For a marketing leader defending budget, the framing is straightforward. These are not incremental optimizations; they are the conditions under which any content or GEO investment can pay off. Publishing more comparison pages onto a site that renders them client-side, buries them behind redirect chains, and wastes a third of its crawl budget is spending on a channel that structurally cannot deliver. The technical floor has to exist before the content strategy stands on it.
“At enterprise scale, the pages are almost always fine. What fails is the template, the rendering, and the plumbing, and no one owns those.”
See where you are cited today
A free snapshot audit of your rankings and AI citations before we ever talk.
Josh leads work at the intersection of SEO and generative engines at Something Inc., helping B2B brands get ranked and cited across every major AI engine.