What Is SEO in 2026? A Practical Guide With Real Case Studies

By Staff Writer July 25, 2026 13 min read

Type almost any question into Google, Bing, or Yahoo today and you’ll notice something that wasn’t true five years ago: often, an AI-generated summary appears above the traditional list of blue links, answering your question before you’ve even clicked anything. This single shift — the rise of AI Overviews — has done more to change how SEO is talked about than any algorithm update in the last decade. And yet, underneath the new interface, the fundamentals of search engine optimization haven’t disappeared. They’ve been sharpened, and the last two years have produced some unusually clear, publicly documented case studies of exactly what happens when a site gets that sharpening wrong — and right.

Source: search.google/ways-to-search/ai-overviews

This guide walks through what SEO actually is in 2026, why it still matters even in an AI-saturated search landscape, how the discipline has shifted since Google’s 2024–2026 wave of core updates, and — because frameworks are only useful once you’ve seen them break something real — three of the most closely analyzed case studies in recent SEO history.

What SEO Actually Means

Search engine optimization is the process of improving a website so that search engines can find it, understand it, and confidently recommend it to people searching for related information. That’s the definition in its simplest form — but “confidently recommend” is doing a lot of work in that sentence, and it’s worth unpacking rather than glossing over.

Google isn’t just matching keywords anymore. Its systems evaluate whether a page demonstrates real expertise, whether it was plausibly written by someone with genuine firsthand experience of the topic, whether it says something the top-ranking pages don’t already say, and whether the site as a whole has a track record of being trustworthy. SEO has moved from a largely mechanical discipline — matching search terms to page text — toward something closer to reputation management combined with technical execution. The case studies below exist precisely because that shift caught a lot of well-resourced, experienced marketing teams off guard.

A Simple Diagnostic: The Relevance-Depth Test

Before getting into specific companies, it’s worth introducing one lens that makes the case studies easier to interpret, rather than treating each as an isolated mystery. Call it the relevance-depth test: for any page, ask two separate questions — is this genuinely relevant to what the site is authoritative about, and does it go deeper than the current top-ranking pages on the same query? A page can fail on either axis independently, and the failure mode looks different each time. Content that’s deep but off-topic for the site (a CRM company publishing an exhaustive guide to cover letters) dilutes authority even if the writing is excellent. Content that’s on-topic but shallow gets outpaced the moment a competitor publishes something more thorough. Keep this test in mind — every case study below fails one axis or the other, and recognizing which one changes what the fix actually looks like.

Why SEO Still Matters When AI Overviews Exist

It’s a reasonable question: if an AI Overview answers the query directly on the results page, why would a business still invest in SEO? The answer is that AI Overviews aren’t created from nothing — they’re generated by pulling from pages Google has already crawled, indexed, and judged relevant and credible, the same pages that would otherwise rank organically. Being cited inside an AI Overview is, in effect, a new and highly visible ranking position, and sites earning those citations have reported meaningful click-through rate gains compared to a standard listing, because appearing as the source Google trusted enough to cite carries its own credibility signal.

At the same time, a large share of purely informational searches now end without any click at all, because the AI Overview satisfies the query on its own. Publishing shallow, easily-summarized content aimed purely at broad informational keywords is a weaker bet than it used to be. Content that goes further — original data, a distinct point of view, something genuinely useful beyond the surface-level answer — is what continues to drive clicks, because it gives someone a reason to read past the summary.

The Three Pillars of SEO

1. Technical SEO

Technical SEO is the infrastructure layer — it determines whether Google can access and properly render your content in the first place. This includes site speed, mobile usability, crawlability, a clean XML sitemap, and the absence of indexing errors. No matter how good the writing is, a page that’s blocked from indexing, painfully slow, or broken on mobile will never get a fair chance to rank.

2. On-Page SEO

On-page SEO covers what you control directly within the content: title tags, heading structure, the actual quality and depth of the writing, internal links, image alt text, and how well the content matches what the searcher wants. This is usually where small businesses have the most direct control, and it’s where the biggest quality gaps show up between content written to rank and content written to genuinely help.

3. Off-Page SEO

Off-page SEO is about the reputation a site builds beyond its own pages — backlinks from trustworthy sites, brand mentions, citations, and digital PR. It’s the slowest pillar to build, since it depends on others choosing to reference you, but it’s also one of the strongest trust signals Google uses.

Case Study 1: How HubSpot Lost the Majority of Its Blog Traffic

HubSpot is one of the more useful case studies in recent SEO history precisely because it wasn’t a small, inexperienced site — it was, for years, held up as a model of content-led SEO. Between early 2023 and early 2025, HubSpot’s blog subdomain saw its organic traffic estimated to fall from roughly 24 million monthly visits to somewhere in the range of 6 to 8 million, according to third-party visibility tools like Semrush and Sistrix, with the steepest single drop occurring after the November and December 2024 core and spam updates. Separate tracking reported the number of keywords HubSpot ranked in the top three positions for falling from around 138,000 to roughly 30,000 over the same period.

What makes this a genuinely instructive case, rather than just an alarming headline, is the “why.” Analysts who dug into the affected pages found a consistent pattern: much of the lost traffic came from content with little relevance to HubSpot’s actual product — pieces on topics like cover letter examples or generic sales quotes, which had driven meaningful traffic for years but did nothing to reinforce HubSpot’s authority in CRM and inbound marketing. Competing publishers with narrower, more topically focused coverage — sometimes with substantially longer, more detailed guides on the same subjects — held or gained ground through the same updates. This is the relevance-depth test playing out at enterprise scale: the content wasn’t necessarily low-quality by traditional standards, but it was off-topic for what Google had come to understand HubSpot’s site was actually authoritative about.

Notably, HubSpot’s business results didn’t collapse alongside its blog traffic — the company continued reporting revenue growth through the same period, a reminder that blog traffic and business health, while related, aren’t the same metric. But for any site relying on blog content as a primary top-of-funnel channel, the lesson holds: breadth without topical alignment is increasingly fragile.

Case Study 2: Forbes Advisor and “Site Reputation Abuse”

Around the same period, Forbes experienced an estimated 60–80% decline in organic search visibility, concentrated heavily in Forbes Advisor, its affiliate content division covering topics like credit cards, insurance, and loans. The core issue Google identified was a practice it formally named site reputation abuse: publishing large volumes of third-party or loosely affiliated content on a subdomain or section of an otherwise reputable, high-authority site, in an attempt to borrow that site’s established trust for topics the publishing team had no direct authority over.

Google’s policy response was a manual action specifically targeting this pattern, applied independent of the otherwise strong domain-level authority Forbes.com carried. The lesson for smaller sites considering a similar structure — hosting third-party or loosely supervised content under a trusted brand’s domain to borrow its authority — is that this is now an explicitly enforced policy violation, not a gray-area growth hack.

Case Study 3: Reddit’s Rise and What It Signals About “Experience”

Not every case study from this period is a cautionary tale. Reddit’s search visibility grew dramatically — one widely cited visibility index tracked over a 1,000% increase in Reddit’s organic search visibility between mid-2023 and early 2024, with the platform climbing from roughly the 68th highest-visibility domain in Google’s US results to among the top five. Reddit threads began ranking, and continue to rank, for a wide range of product comparison, recommendation, and “is this normal” style queries that previously surfaced polished, professionally written content instead.

The most useful interpretation of this shift isn’t “forums beat blogs” — it’s what Reddit threads structurally contain that a lot of professionally produced content doesn’t: specific, first-person accounts of actually using a product or navigating a situation, visible disagreement and debate rather than a single manufactured consensus, and community validation signals (upvotes, reply depth) that function as a crude but real proxy for whether other people found the answer genuinely useful. This lines up directly with the “Experience” component of Google’s E-E-A-T framework, which for years existed mostly as guidance for human quality raters and has become considerably more visible in actual rankings since. The practical takeaway for businesses isn’t to abandon expert-written content — it’s to make sure that content actually contains the specific, lived detail that makes Reddit threads rank, rather than reading as a generic, interchangeable overview of the topic.

What Changed Between 2024 and 2026

Google’s core updates over the past two years share a consistent theme: rewarding original, experience-backed content and penalizing synthetic, derivative content at scale.

“Information gain” became a measurable signal. Google’s updates specifically targeted pages that reorganized or summarized what the existing top 10 results already said without contributing anything new — the exact pattern behind much of HubSpot’s affected content.
E-E-A-T moved from guideline to enforced outcome. Experience, Expertise, Authoritativeness, and Trust had been part of Google’s quality rater guidelines for years, but recent updates made these signals far more consequential in actual rankings, not just in theory — visible in both Reddit’s rise and Forbes Advisor’s fall.
Site reputation abuse became a named, enforced policy, closing off a specific tactic — borrowing a trusted domain’s authority for loosely related third-party content — that had previously existed in a gray area.
AI Overviews became mainstream on informational queries, changing both click behavior and the strategic value of ranking well for broad, easily-summarized topics.

For a business owner or content creator, the practical takeaway is this: the old playbook of publishing keyword-targeted articles and collecting backlinks hasn’t been abandoned, but the bar for “good enough” content has risen substantially, and the reward for genuine topical alignment and firsthand experience has risen with it.

How SEO Actually Works, Step by Step

Crawling — Google’s automated crawlers discover pages, usually by following links from pages it already knows about, or via an XML sitemap.
Indexing — Once crawled, Google analyzes the page and decides whether to store it in its index, the database of content eligible to appear in search results.
Ranking — For every query, Google evaluates relevant indexed pages against hundreds of signals and decides which to show, and in what order.
AI synthesis — For queries that trigger an AI Overview, Google’s generative systems pull from the same indexed, ranked content to construct a summarized answer, often citing sources directly.

A common mistake is treating these as one-time events. In reality, the pipeline runs continuously — pages get re-crawled, re-evaluated, and re-ranked constantly as the web changes, which is exactly why HubSpot’s decline unfolded gradually over roughly ten months rather than overnight.

A Practical Starting Framework, Informed by the Case Studies Above

Step 1 — Confirm the technical basics are sound. Use Google Search Console to check that your site is indexed, mobile-friendly, and free of major crawl errors.

Step 2 — Run your own content library through the relevance-depth test. Before adding anything new, identify existing pages that are off-topic for your core authority (HubSpot’s cover-letter problem) versus pages that are on-topic but thin compared to current top results. These require different fixes — pruning or redirecting the former, deepening the latter.

Step 3 — Write content that adds something the top-ranking pages don’t already have. A real example from your own work, a specific number or result, or a level of specificity competitors haven’t bothered with. Generic restatement of common knowledge is precisely what recent core updates have targeted.

Step 4 — Borrow the “experience” signal Reddit exploits, deliberately. Include specific, firsthand detail — what actually happened, what surprised you, what didn’t work — rather than a smoothed-over, generic overview. This is the same underlying signal, applied to owned content instead of a forum thread.

Step 5 — Never host loosely supervised third-party content under your primary domain’s authority. The Forbes Advisor case made explicit what was previously a gray area; treat any affiliate or syndicated content section as carrying real policy risk if it isn’t genuinely produced and reviewed with the same rigor as your core content.

Step 6 — Be realistic about timelines. SEO is a compounding strategy. Meaningful, sustained results typically take three to six months of consistent effort, and the case studies above show that decline, like growth, tends to unfold gradually before any single update makes it visible.

Common Misconceptions About SEO in 2026

“SEO is dead because of AI Overviews.” Not accurate — AI Overviews are built from the same underlying ranking systems, and being the cited source is now one of the most valuable positions in search.

“More content is always better.” HubSpot’s case is the clearest available counterexample: a large, well-produced content library can lose the majority of its traffic when too much of it sits outside the site’s genuine area of authority.

“A trusted domain can absorb any content placed on it.” Forbes Advisor’s manual action shows this is no longer true in practice, if it ever reliably was.

“Keyword density is what matters most.” Modern ranking systems evaluate semantic meaning, intent, and — increasingly — demonstrated firsthand experience, far more than literal keyword repetition.

Frequently Asked Questions

How long does SEO take to show results?
Most businesses start seeing measurable movement within three to six months, with more substantial results building over six to twelve months, depending on competition and consistency.

Could a HubSpot-style traffic collapse happen to a smaller site?
Yes — the underlying mechanism (broad, off-topic content diluting a site’s perceived area of authority) applies at any scale. Smaller sites are arguably more exposed, since they have less overall domain authority to absorb the dilution.

Is AI-written content bad for SEO?
The issue isn’t the tool used to draft content — it’s whether the final result demonstrates genuine expertise, originality, and specific, firsthand usefulness. Content generated at scale with minimal human oversight or added value is what recent spam policies specifically target.

Final Thoughts

SEO in 2026 rewards fundamentally the same thing it always has: content that genuinely helps the person searching, delivered on a technically sound site, backed by real credibility. What’s changed is the sophistication with which Google can now tell the difference between real expertise and content designed purely to rank — and the past two years have produced unusually clear, public evidence of exactly what that looks like when it goes wrong, and when it goes right.

References

Search Engine Land — HubSpot SEO traffic drop coverage
Aleyda Solis — Analysis of HubSpot’s organic traffic decline
Google Search Central — March 2024 core update and spam policy announcement
Google Search Central — Site reputation abuse spam policy
Orange SEO — Analysis of Reddit’s search visibility growth

Want a relevance-depth audit of your own content library before your next core update surprises you? [Link to your contact/services page here.]