SEO in 2026: What Actually Moves Rankings Now (and What Doesn’t)

Search has changed more in the last two years than in the previous ten. AI-generated summaries, zero-click results, and Google’s spam systems have all reshaped what it takes to rank. If you’re still running the same SEO playbook from 2021, you’re probably working harder for less traffic.

Here’s an honest breakdown of what matters now, what’s a waste of time, and where I’d focus if I were running SEO for a small or mid-sized business.

AI Overviews changed the click economy

Google now answers a huge share of queries directly on the results page. For informational searches (“what is a canonical tag,” “how long does SEO take”), the AI Overview often satisfies the user before they ever see your link. That doesn’t mean SEO is dead. It means the queries that send traffic have shifted.

What still drives clicks:

• Commercial and transactional queries. Someone searching “best CRM for small law firms” or “emergency plumber Commack” wants a vendor, not a definition.
• Specific, hard-to-summarize answers. If your content includes original data, screenshots, real pricing, or hands-on experience, the AI can’t fake it.
• Local intent. “Near me” and city-specific searches still funnel to map packs and local pages.

What’s losing traffic: generic how-to posts that restate what everyone else already wrote. If your article reads like a summary of the top five results, the AI Overview will just summarize it too, and you get nothing.

Experience is the moat now

Google’s helpful content systems have been folding into the core algorithm for a while, and the pattern is clear: content that demonstrates first-hand experience outperforms content that just demonstrates research. This is the “E” in E-E-A-T, and it’s the hardest one to fake.

Practically, that means:

• Write from your own work. If you built the thing, say what broke and how you fixed it.
• Include specifics: real numbers, real tools, real timelines, screenshots of real dashboards.
• Skip the “10 tips” format unless you actually have ten things worth saying.

I’ve watched thin pages on small sites outrank 3,000-word posts on big ones simply because the small site’s author had clearly done the work being described. That’s the direction things have gone, and it’s not reversing.

Technical SEO still does the heavy lifting

None of the content stuff matters if Google can’t crawl, render, and index your site efficiently. This is where a lot of sites quietly bleed traffic. A few things I check on almost every project:

Core Web Vitals (INP replaced FID)

Interaction to Next Paint is now the responsiveness metric. It measures the worst-case delay between a user interaction and the next visual update. If you have heavy third-party scripts firing on click, your INP is probably bad.

// Measure INP in the field with the web-vitals library
import { onINP } from 'web-vitals';

onINP(({ value, rating, attribution }) => {
 // value is in milliseconds; good is under 200ms
 console.log('INP:', value, rating);
 console.log('Worst element:', attribution?.interactionTarget);
});

The usual fix is deferring non-critical JS and breaking up long tasks. If a click handler runs 400ms of synchronous work, INP eats that whole delay.

Structured data that actually earns rich results

Schema markup won’t boost rankings by itself, but it can win you enhanced results that lift click-through. For local businesses, LocalBusiness and FAQPage (where eligible) are still worth the effort. Keep it accurate. Fake review markup is a fast way to get a manual action.

<script type="application/ld+json">
{
 "@context": "https://schema.org",
 "@type": "LocalBusiness",
 "name": "Your Business",
 "address": {
 "@type": "PostalAddress",
 "streetAddress": "123 Main St",
 "addressLocality": "Commack",
 "addressRegion": "NY",
 "postalCode": "11725"
 },
 "telephone": "+1-631-555-0100",
 "url": "https://example.com"
}
</script>

Canonicals and crawl budget

Duplicate content from URL parameters, trailing slashes, and www/non-www splits still wastes crawl budget on a lot of sites. Pick one canonical version and 301 everything else to it. This is boring work that pays off.

Local SEO is still a separate game

If you serve a geographic area, your Google Business Profile is often more valuable than your website’s homepage. For anyone doing Long Island SEO, the local pack is where the money is. The fundamentals haven’t changed much:

• Complete your GBP fully, including services, hours, and attributes.
• Get reviews consistently. Recency matters as much as volume.
• Build city and service-area pages that are genuinely different from each other, not templated swaps of the city name.
• Keep your NAP (name, address, phone) identical across every citation.

The templated city-page thing is a trap. Google is good at spotting pages where only the city name changed. If you can’t write something specific about serving that area, don’t publish the page.

Where AI fits (and where it backfires)

Using AI to draft content isn’t inherently a problem. Publishing unedited AI content at scale is. The failure mode is obvious to readers and to Google: generic phrasing, no specifics, no point of view.

A workflow that works:

• Use AI for outlines, research summaries, and first drafts.
• Rewrite everything in your own voice with your own examples.
• Add the stuff AI can’t know: your data, your clients’ questions, your mistakes.
• Fact-check every claim. Models hallucinate confidently.

If your published post could have been written by anyone with a ChatGPT login, it’s not going to rank for anything competitive.

The foundation is still the foundation

Fast, well-structured sites with clear information architecture win. That hasn’t changed since I started building sites. What changed is how much the content layer matters on top of it. You need both. A beautiful, fast site with thin content goes nowhere. Great content on a slow, broken site goes nowhere either.

If you’re rebuilding a site and want the technical side done right from the start, that’s most of what I do with custom web development. The SEO work is easier when the foundation isn’t fighting you.

FAQ

How long does SEO take to show results in 2026?

For a new site or a big rebuild, expect three to six months before you see meaningful movement, and six to twelve for competitive terms. Sites with existing authority move faster. Anyone promising page one in a month is either targeting terms nobody searches or lying.

Is AI content penalized by Google?

Not for being AI-generated. Google penalizes unhelpful content regardless of how it was made. The problem is that most unedited AI content is unhelpful, so it gets filtered out. Use AI as a tool, edit heavily, and add original value.

Do I still need backlinks?

Yes, but quality beats quantity more than ever. A handful of links from relevant, real sites will outperform hundreds from directories and guest-post farms. Links are still a trust signal. Spam links just waste your time and can hurt you.