Definition
A search engine results page (SERP) is the page a search engine returns in response to a query. It used to mean ten blue links with ads above them. Today a Google SERP is assembled per query from many modules, and the top of it is often an AI-generated answer rather than a link.
A typical Google SERP in 2026 is built from four layers. At the top, an AI Overview may summarize an answer with linked sources; Comscore measured AI Overviews on 39.4% of US desktop searches in June 2026. Ads can appear above and below AI Overviews in every market that has them, and inside AI Overviews for English searches in 12 countries, including the US. Organic results follow, often interleaved with SERP features such as local packs, People Also Ask, video and image carousels, shopping listings, Top Stories and a knowledge panel. Which modules appear depends on the query's intent, the user's location and device, and freshness.
AI Mode is a different kind of results page. Users choose to open it, it answers in a conversational format with inline links and follow-up questions, and Google says it may use different models and techniques than AI Overviews, so the two can cite different pages for the same question. Google reported more than 1 billion monthly AI Mode users in May 2026. In October 2026 Google announced more inline links in both AI features and began testing a dedicated section that surfaces official sites at the top of a response.
The SERP's job has shifted from sending people elsewhere to answering on the page. SparkToro's 2024 clickstream study found 58.5% of US Google searches ended without a click. In March 2025 browsing data analyzed by Pew Research, users clicked a traditional result on 8% of visits to SERPs with an AI summary and 15% without one, and clicked a link inside the summary on just 1%. Google disputed the methodology. Either way, ranking first can deliver less traffic than it did when the first organic result was the first thing on the page.
AI assistants also produce something like a results page. ChatGPT search, Perplexity and Copilot return a written answer with numbered or inline citations and, sometimes, product cards or maps. These answers are not ranked lists, but the order in which brands and sources appear inside them works much like position on a SERP, which is why GEO tools track answer position and citations rather than rank.
For teams working on visibility, the SERP is now something to map before you write. For each target query, note which modules appear, whether an AI Overview shows and which sources it cites, and what the organic results have in common in format and depth. That tells you what Google believes the intent is and which surface is worth competing for: an organic position, a place in the local pack, or a citation in the AI answer. Zero-click search explains the traffic side, and brand SERP covers results for your own name. Our guide on checking whether your site is cited in AI Overviews shows how to audit the AI layer.
Examples of Search Engine Results Page (SERP)
- A search for 'how to remove red wine from carpet' returns an AI Overview with steps and three cited cleaning sites, a People Also Ask box, a video carousel and shopping ads for stain removers. The first organic result sits well below the fold on mobile.
- 'Dentist near me' produces a local pack with map pins, ratings and opening hours above any organic link. A practice ranked first organically gets fewer calls than the three practices shown in the pack.
- The same question asked in AI Mode returns a conversational answer citing six sources, only two of which appear on page one of the regular SERP, so the team tracking it adds AI Mode to its monitoring.
- A software company searching its own name finds a knowledge panel, its homepage with sitelinks, a G2 review page, a Reddit thread and an AI Overview that repeats a complaint from that thread. It treats the brand SERP as a reputation project as much as a ranking one.
