SEO for enterprise companies is a different job from SEO for a 200-page site, and most of the advice written about it pretends otherwise. The tactics that work on a small site — write better pages, earn some links, tidy the titles — assume you can touch every page and that one person's decision is enough to ship a change. At 400,000 URLs across six subdomains, four content management systems and three business units, neither assumption holds.
The constraint stops being knowledge and becomes throughput. You usually know what is wrong. The question is which of the forty things you know are wrong will produce revenue if fixed, and whether you can get the fix through a release train that is booked out until next quarter.
This guide covers the failure modes that are specific to large sites, and what to do about each one.
What makes enterprise SEO different
Three things, mostly.
Scale changes which problems matter. On a small site, a duplicate title tag is a minor annoyance. On a site where one template generates 80,000 pages, a duplicate title tag is 80,000 duplicate title tags, and Google's response to that is not to rank them slightly lower — it is to stop treating that section of the site as worth crawling regularly.
Ownership is diffuse. The product team owns the templates. Engineering owns the render path. Legal owns the disclaimer that pushes the content 1,400 pixels down the page. Brand owns the words in the H1. An SEO recommendation that requires all four to agree is not a recommendation, it is a project.
Latency is enormous. A change you specify in January may ship in May and show measurable results in August. This makes prioritisation the highest-leverage skill in the role. You get perhaps six real swings a year at a large organisation, so each one needs to be aimed at something that matters.
Everything below follows from those three.
Crawl budget and index bloat: the enterprise-only problem
Small sites do not have a crawl budget problem. Enterprise sites almost always do, and it is usually the single largest source of wasted opportunity.
The pattern is familiar. Faceted navigation generates a URL for every combination of filters. Internal search results are crawlable. Session parameters, tracking parameters, sort orders and pagination each multiply the URL space. A catalogue of 40,000 real products becomes 3 million crawlable URLs, of which perhaps 25,000 deserve to be indexed.
Googlebot then spends its time on near-duplicate garbage while your genuinely new pages wait days or weeks for a first crawl.
How to diagnose it properly:
- Pull raw server logs, not analytics. Filter for verified Googlebot and count requests by URL pattern. If more than 40% of crawl requests hit URLs with query parameters, you have found your project. Logs are the only source that tells you what the crawler actually did, rather than what a tool guessed it would do.
- Compare crawled versus indexed versus generating traffic. Three numbers, three orders of magnitude apart, is the normal starting state.
- Segment the index coverage report by directory. Aggregate numbers hide everything. A section with 90% "Crawled – currently not indexed" is a template quality problem, not a technical one.
The fixes are ordered by bluntness. Robots.txt disallow for parameter patterns that should never be crawled. noindex for pages that must remain crawlable but must not rank. Canonical tags for genuine duplicates where you want the equity consolidated. Removing the internal links that create the URLs in the first place — the most permanent fix and the one most often skipped, because it requires a front-end change rather than a config change.
A useful rule: if a URL cannot be reached by a human doing something reasonable, it should not be reachable by a crawler either.
Fix the template, not the page
The highest-return work at enterprise scale is almost always template work, because the multiplier is enormous. Changing the way a product page constructs its title tag is one ticket that affects every product page.
Practically, this means the audit output should be organised by template rather than by URL. Instead of "these 3,000 pages have thin content," the finding is "the category template renders no unique copy above the product grid, which affects 3,000 pages, and here is the one component that needs changing."
Templates worth auditing first are the ones that generate the most pages and the ones closest to money. For a retailer, that is category and product. For a marketplace, listing and location. For B2B SaaS, the integrations and comparison directories, which are usually generated from a database and usually neglected. Our Shopify SEO optimization guide works through the same template logic on a smaller platform, and the reasoning transfers directly: the number of pages changes, the diagnosis does not.
Common template-level wins:
- Internal links to related items rendered server-side rather than injected by client-side JavaScript after a user interaction
- Titles and H1s built from real attributes rather than a fixed string plus a product name
- Structured data emitted from the same data layer that renders the visible content, so the two cannot drift apart
- Pagination that exposes deep items to the crawler instead of hiding page 40 behind an infinite scroll
- A copy block on category pages that the merchandising team can actually populate, rather than one nobody has permission to edit
Internal linking is the lever nobody pulls
Large sites usually have plenty of external authority and terrible internal distribution of it. Authority piles up on the homepage and a handful of blog posts, then fails to reach the pages that make money because those pages sit five clicks deep behind a mega-menu that links to the same forty destinations from every page on the site.
Two things to measure. First, click depth from the homepage for your revenue pages — anything beyond four is a problem, and anything beyond six is effectively invisible. Second, internal links pointing to each priority URL. It is common to find a page you have decided is strategically critical with eleven internal links pointing at it, while a 2019 blog post has 6,000.
The fixes are cheap relative to their effect: contextual link modules on high-authority pages, hub pages that consolidate a topic and link out to its members, breadcrumb trails that are real links, and a periodic pass that adds internal links from older content to newer priority pages. This is also the cheapest way to make external link acquisition go further, since every earned link flows through the same internal graph — see our guide to building authority backlinks for how the two efforts compound.
The cannibalization tax
At enterprise scale, several teams write about overlapping topics for years, and nobody deletes anything. You end up with nine pages targeting variations of the same query, none of which rank, because each has a fraction of the signals a single consolidated page would have.
Finding it: export query-level data and look for queries where the ranking URL changes over time. That instability is the signature. A page that oscillates between position 8 and position 19 while the URL Google chooses keeps switching is not a ranking problem, it is a selection problem — Google cannot decide which of your pages is the answer.
Consolidation beats optimisation here. Pick the strongest URL, merge the useful content into it, redirect the rest with 301s, and update internal links to point at the survivor. Teams resist this because deleting content feels like destroying an asset. The asset is the ranking, not the file.
Budget for the political side. "We are deleting 4,000 pages" is a difficult sentence in a large organisation. "We are consolidating 4,000 pages into 600 that will each rank, and here are the three pilot consolidations that lifted their clusters by 40%" is a much easier one. Run the pilot first.
Do you need an enterprise SEO platform?
The market splits into crawlers, rank trackers, log analysers, and platforms that bundle all three with workflow features. Annual contracts commonly land between $20,000 and $150,000.
The honest answer is that the platform is worth it when the bottleneck is monitoring, and not worth it when the bottleneck is implementation. If you have a backlog of known fixes waiting on engineering time, another tool that finds more problems makes the backlog longer, not shorter. Spend the money on a developer instead.
Platforms genuinely earn their price on three jobs: crawling millions of URLs on a schedule without falling over, holding rank data for tens of thousands of keywords with segmentation, and alerting you when a deployment breaks something before the traffic drop shows up in a monthly report. That last one pays for the contract on its own the first time it catches a noindex shipped to production.
What tooling will not fix is a broken relationship with the engineering team. No platform ships code.
Prove revenue, not rankings
Enterprise SEO programmes get cut when they cannot be defended in a budget meeting. Rankings and sessions do not survive that conversation, because nobody outside the SEO team can convert them into money.
Build the number instead:
- Segment traffic by intent, not by channel alone. Brand search is demand you already had; treat it as a marketing effectiveness measure, not an SEO result. Non-brand traffic is the number the programme should be judged on.
- Attach a conversion rate to each segment. Enterprise sites usually have wide variation between sections, and averaging across them hides both the wins and the problems. The conversion rate calculator is enough to do this segment by segment when your analytics setup will not.
- Multiply by value, not price. Use contribution margin per order, or for subscription businesses, lifetime value — our LTV calculator covers the standard retention-based version.
- Compare against the paid cost of the same traffic. The most persuasive slide in any enterprise SEO deck is the paid-search cost of replacing organic non-brand clicks at current CPCs. Running that figure alongside your blended acquisition cost — the CAC calculator handles the arithmetic — reframes SEO as a cost avoidance line item rather than a speculative one.
If you are working with incomplete analytics access, which is common in the first months of an enterprise engagement, our notes on how to determine website traffic cover the estimation methods that will get you close enough to prioritise.
Report a rolling twelve-month view. Enterprise SEO results are seasonal and lagged, and month-over-month reporting on a large site mostly measures weather.
In-house, agency, or consultant
Large organisations tend to need all three, in a specific arrangement.
In-house staff own relationships and prioritisation. They know which engineering team owns which template and which VP needs convincing. This is the part that cannot be outsourced, because it is mostly institutional knowledge.
Agencies bring execution capacity — content production, technical audits, link acquisition — and pattern recognition from having seen the same migration go wrong at six other companies. The failure mode is an agency that produces recommendations nobody implements, which is usually a symptom of the in-house side being understaffed rather than the agency being bad. If you are evaluating vendors, the questions in our white hat SEO company guide apply at any size, though at enterprise scale you should add one more: ask them to describe a large migration they worked on that went badly, and what the traffic recovery curve looked like.
Independent consultants are best used for a bounded, high-stakes problem: a replatform, an international rollout, a manual action. Expensive per day, cheap relative to the cost of getting a migration wrong.
FAQ
How long does enterprise SEO take to show results?
Technical fixes to crawl and indexation can show movement in four to eight weeks, because you are unblocking pages that already exist. Content consolidation typically takes three to six months. New content targeting competitive commercial queries takes nine to eighteen months on most enterprise sites. Any plan that promises material revenue in the first quarter is describing paid search.
What does enterprise SEO cost?
Agency retainers for genuine enterprise scope generally run $10,000–$50,000 per month, with large programmes above that. Add platform licensing of $20,000–$150,000 annually and at least one full-time internal owner. The larger cost is usually engineering time, which rarely appears in the SEO budget but determines whether any of the rest of it produces anything.
Should we build a subdomain or subdirectory for the blog?
Subdirectory, in almost every case. Content on example.com/resources/ inherits the domain's authority directly, while blog.example.com has to build its own. The usual reason for choosing a subdomain is that the CMS is easier to host separately, which is an engineering convenience being paid for with organic traffic. If you are already on a subdomain and it performs well, migrating carries real risk; weigh it as a project rather than a default.
How do we stop breaking SEO with every release?
Put automated checks in the deployment pipeline. Fail the build if a template emits noindex on a page type that should be indexable, if canonical tags point off-domain, if the robots.txt changes, or if render-blocking behaviour on key templates regresses past a threshold. Most enterprise traffic disasters are not strategic errors — they are a staging configuration reaching production on a Friday afternoon.