SERP: Saat Peringkat Tak Lagi Berarti Visibilitas

A SERP, or search engine results page, is the page a search engine returns after someone submits a query. It can mix organic links, ads, rich results, images, videos, and AI-generated features, and Google says that mix shifts with the query, device, country, and language. Here's the fact that should slow you down. In Google Search Console, every link inside an AI Overview receives the same position value, so "position one" can describe the first conventional result or one of several sources grouped inside a single generated summary.

The definition is simple. Reading a SERP well isn't.

How a SERP Is Assembled for Each Query

A results page is assembled for the query. It isn't one fixed layout.

The Same Query, Different Results Pages

Google's visual elements gallery documentation states that the appearance of result elements can change over time and can differ by device, by the searcher's country, by the language of the query, and by other factors. Query type matters too, since image or video searches tend to return image or video results.

Why this matters: a SERP isn't one fixed layout. Any screenshot captures only one version.

Results, Ads, and Features

A useful way to sort what appears on the page comes from the 2019 Nielsen Norman Group SERP taxonomy, which groups content into three buckets: results, advertisements, and SERP features. In that framework, results link to external pages, ads are paid placements, and SERP features present unpaid information in formats other than a traditional result, such as featured snippets, People Also Ask, carousels, and knowledge panels.

Treat that as an analytical lens, not an official classification. Interface details from 2019 have changed, so current feature names should come from Google's own documentation.

Anatomy of a Google Results Page

Google classifies what appears on its results pages as text results, rich results, image results, video results, and exploration features. Don't force a one-to-one mapping onto the three-bucket model above.

The Standard Text Result

A standard text result is built from three core parts: attribution, a title link, and a snippet. Attribution can include a favicon, the site name, and a visible URL. A byline date, sitelinks, an image, or rich attributes may also appear. Not every result shows every part.

Rich, Image, and Video Results

Rich results usually rely on structured data to display graphic elements or interactive experiences, such as recipe details or event dates. Image and video results are separate presentations, not decorated versions of a text result.

Related Questions and Related Searches

These two look similar but behave differently. A related searches group sends the user to another results page. A related questions group, better known as People Also Ask, expands in place and can show a featured snippet when a question is opened. Here's the thing: changing the query inside Google isn't the same as a click out to your site.

Where AI Overviews and AI Mode Sit

Google's documentation on AI features and your website treats AI Overviews and AI Mode as features within Google Search. Their responses and supporting links won't always match. An AI Overview appears only when Google's systems judge it adds value. It doesn't trigger on every query.

For publishers, the requirements are ordinary ones. A page must be indexed and eligible to appear with a snippet, and Google says no special schema, no AI-specific file, and no additional technical work is needed to be considered as a supporting link. Performance from these features sits inside the Web search type in Search Console, though that doesn't mean all Web traffic comes from AI.

Ranking Position Versus Actual Visibility

Position is the metric most teams report first. It's also easy to overread.

How Search Console Assigns a Position

According to Google's Search Console metrics documentation, position is recorded only when a link earns an impression, and average position reflects the topmost position your property or page occupied, averaged across those impressions. An AI Overview occupies one position on the page. Every link inside it receives that same position value, which says nothing about the order of sources within the summary, and the documentation doesn't say an AI Overview always sits at position one. A click from an AI Overview link to your page counts as a click.

Position is a relative measure. It isn't a pixel coordinate.

What a Shared Position Can Hide

Picture two reports that both show position one. In the first, your page is the first conventional text result. In the second, your page is one of several links inside an AI Overview, sharing that slot with other sources. The number is identical, but the screen space, the competing elements, and the likelihood of a click are not.

The reality is that being included and being clicked are separate outcomes, as the Pew data below shows.

Eligibility Is Not a Guarantee

Many SERP features can't be bought or switched on. A page can only qualify for them.

What Valid Structured Data Does and Doesn't Do

Google's general structured data guidelines are direct on this point. Using structured data enables a feature to be present, but it doesn't guarantee the feature will be present, even when the markup passes the Rich Results Test. Google's algorithm tailors results to each search experience based on variables such as search history, location, and device type, and it may decide that another feature, or a plain text result, serves the user better.

Eligibility can also be lost. Markup that doesn't represent the main content, describes content hidden from users, or misses required properties for its rich result type won't qualify. A structured data manual action removes a page's eligibility for rich result appearance, yet Google states it doesn't affect how the page ranks in web search. Eligibility, appearance, and ranking are three separate questions.

FAQ Markup Isn't People Also Ask

A common shortcut assumes FAQ schema will place a page in People Also Ask. Google's Search Console documentation separates the two. An FAQ rich result is defined by structured data on a site, while People Also Ask is an algorithmic suggestion that can show information and links from sites. Expanding a question isn't a click to your site. A click on your link inside the expanded section is, and impressions for links in that section are recorded once it's expanded.

Two Documented Implementations From 2018

Two Google case studies show structured data work in practice. Both date from 2018. Neither is a controlled experiment or a current benchmark.

• Rakuten Recipe, Japan. The Rakuten Recipe case study describes Recipe structured data applied across a Japanese recipe platform, deployed through its content management system, validated with Google's testing tools, and monitored for rich result issues in Search Console. That era's tools aren't current requirements.
• Eventbrite, United States. The Eventbrite event markup case study describes Event structured data added through a reusable template across event pages, supplying dates and venues for an enriched event search experience in the United States. The team used Search Console and testing tools to catch errors, and the case study notes that other SEO projects ran at the same time, which makes exact attribution difficult.

Worth checking: neither case proves markup guarantees the same presentation today. Both show process, not a promise.

What Click Studies Show About AI Features

Several organizations have measured clicks around AI features. Their findings aren't interchangeable.

Observational Evidence

The Pew Research Center click study tracked the browsing of 900 adults in the United States in March 2025 and matched their searches to results pages collected again from April 7 to April 17, 2025. Users clicked a traditional search result on 8% of visits to pages with an AI summary, compared with 15% of visits to pages without one, and clicked a link inside the summary on 1% of visits (REPORTED; observational, United States). Because the results pages were reconstructed later, what researchers analyzed may differ from what participants actually saw, and the design can't establish cause.

An Ahrefs AI Overviews CTR update reported that the presence of an AI Overview correlates with a 58% lower average click-through rate for the top-ranking page, measured against a modeled estimate without AI Overviews (REPORTED; correlational, desktop Search Console data, December 2023 compared with December 2025, geography not stated). Ahrefs sells SEO tools, which is worth keeping in mind.

Google's Aggregate Claim

Google offers a different frame. In an Google AI Search clicks post, Liz Reid, VP and Head of Google Search, said total organic click volume from Google Search to websites has been relatively stable year over year. The post didn't publish figures or a dataset, so it reads best as a company claim about the aggregate rather than evidence about any single query or position.

When SERP Analysis Misleads You

SERP analysis helps until its numbers get treated as more than they are.

One Manual Search Isn't the SERP

A search from your own browser reflects your device, your location, your language, and that moment. It can't stand in for what a buyer in another city on a different device saw last week. Decisions built on one screenshot optimize for a page nobody else may have seen.

Average Position as a Headline Number

If you're reporting an average position without the query, device, country, and search appearance behind it, you've stripped out everything that gives the number meaning. A shared AI Overview position makes this worse. Pair position with impressions, clicks, and outcomes.

Treating Every Click Study as One Number

The studies above don't measure the same thing. Pew compares visits with and without AI summaries, Ahrefs models a counterfactual click-through rate, the Semrush AI Overviews study followed the same keywords before and after AI Overviews appeared without randomization, and Google reports an aggregate with no published numbers. Zero-click estimates add another layer. SparkToro's zero-click clickstream analysis depends on panel data and on how a click is defined, and its author cautions that year-to-year comparisons built on different panels aren't like for like.

So don't average these findings. Don't frame them as contradictions either.

Reading a SERP Query by Query

A more reliable approach treats each query as its own interface and asks five questions in order.

• Eligibility. Can the page be considered at all? It must be indexed and accessible, and for rich results it needs valid markup that represents the page.
• Appearance. Did it appear, and in which format: a text result, a rich result, an image or video result, or a link inside an AI feature?
• Prominence. How visible was that format next to ads, exploration features, and AI Overviews on that particular layout?
• Engagement. Did the appearance produce impressions and clicks in Search Console?
• Outcome. Did the visit lead to an inquiry, a sale, or another result the business actually cares about?

Each step can fail on its own. A page can be eligible and never appear, appear and never earn a click, or earn clicks that never convert, and a fix aimed at the wrong step simply wastes effort. Teams that want this mapped across a full keyword set, for classic results and AI features alike, can look at Arfadia's search engine optimization services and generative engine optimization services.

Ads, answers, and organic listings all compete for space on a results page. Teams that want help with the paid side can look at search engine marketing, and those working on pages that earn clicks can look at content marketing. For one plan across them, see the coordinated digital marketing services from Arfadia. None of these decides which features a search engine shows for a query.

Frequently Asked Questions

What is a SERP?

A SERP (search engine results page) is the page a search engine returns for a query. It can combine organic links, ads, rich results, images, videos, and AI features, and its layout changes with the query, device, location, and language.

What counts as a SERP feature?

A SERP feature is a distinct element beyond a plain link, such as a rich result, an image or video result, a related questions group, or an AI Overview.

Do AI Overviews need special markup or schema?

No. Google says there are no additional technical requirements, no special schema, and no AI-specific files for AI Overviews or AI Mode. A page must be indexed and eligible to appear in Google Search with a snippet to be considered as a supporting link.

Does valid schema markup guarantee a rich result?

No. Valid structured data makes a page eligible for a supported rich result, but Google doesn't guarantee it will appear, even when the markup passes the Rich Results Test.

Does FAQ markup put a page in People Also Ask?

No. Google treats FAQ rich results and People Also Ask as different features. FAQ rich results come from structured data on a site, while People Also Ask is an algorithmic suggestion that can show information and links from sites.

Does position 1 in Search Console mean the top of the page?

Not necessarily. Position is a relative measure, not a screen coordinate. Every link inside an AI Overview shares the overview's single position, so position 1 can mean the first text result or one of several sources inside a summary.

Related Terms

• Search Engine Optimization (SEO)
• Search intent behind a query
• Featured snippet as SERP feature
• Click-Through Rate (CTR) metric

Sources

• Google Search Central, "Visual elements gallery of Google Search," current documentation. VERIFIED.
• Google Search Central, "AI features and your website," current documentation. VERIFIED.
• Google Search Console Help, "What are impressions, position, and clicks?" current documentation. VERIFIED.
• Google Search Central, "General structured data guidelines," current documentation. VERIFIED.
• Google Search Central, "Rakuten Recipe case study," May 8, 2018. REPORTED, historical.
• Google Search Central, "Eventbrite case study," May 8, 2018. REPORTED, historical.
• Pew Research Center, "Google users are less likely to click on links when an AI summary appears in the results," July 22, 2025. REPORTED, observational.
• Ahrefs, "Update: AI Overviews reduce clicks," February 4, 2026. REPORTED, correlational, vendor study.
• Semrush, "Semrush AI Overviews study," updated December 15, 2025. REPORTED, vendor study, methodology only.
• Nielsen Norman Group, "The anatomy of a search results page," November 17, 2019. Conceptual taxonomy only.
• SparkToro, "In 2026, less than one third of Google searches still send a click." REPORTED, methodology caution only.
• Google, Liz Reid, "AI in Search is driving more queries and higher quality clicks," August 6, 2025. Company claim.