Search intent evolution and SERP feature optimisation: winning the real estate beyond rankings

A ranking is one row on a page with many shapes

Ask an owner how they do in Google and you get a number. "We're third for bike servicing." It hides most of what the customer sees.

A results page is not ten links any more. Google's own gallery of its parts names text, image and video results and what it calls exploration features, such as the related questions box most people know as "People also ask".¹ Around the ordinary links there can be a boxed answer, a map, a row of products, a strip of videos, stars under a result, and since 2025 in Ireland an AI Overview on top. Each is a piece of the page's real estate. "Third place" describes one of them.

Search Console, Google's free report on how your site does in its search, sees the page this way too. Its position metric counts each element on the page, a knowledge panel or a carousel included, and a featured snippet "typically occupies position 1". Ads do not count. What the report shows is an average, built from the topmost position your site held each time it was seen.²

$\bar{p} = \frac{1}{n}\sum_{i=1}^{n} p_i$

Here $\bar{p}$ is the average position Search Console reports, $n$ is the number of impressions (the times a link to your site was shown), and $p_i$ is the topmost position your site held in impression $i$. It is an ordinary mean (Foundations: the mean).

Worked example: an average position the page never held

Leeside Cycle Works is an invented bike shop in Cork. Aoife runs it. She sells bikes and parts online at leesidecycles.example and runs a servicing workshop behind the shop. Nobody real, every number made up.

Her guide "How often should you service your bike" was shown four times on one day for the same question. Twice Google quoted it in the boxed answer at the top, position 1. Twice it showed the guide as an ordinary link below a map and four other results, position 6.

$\bar{p} = \frac{1 + 1 + 6 + 6}{4} = 3.5$

Search Console says $\bar{p} = 3.5$ (Foundations: the mean). The page was never third or fourth. It was either the answer box or below the fold. The average is not wrong, it just cannot tell her which shape she won. The rest of this article is about seeing those shapes, and which ones a shop can influence.

Search intent is the question behind the words

Every one of those shapes serves what the searcher wants. To return relevant results, Google says, it first needs "to establish what you're looking for", which it calls "the intent behind your query". It gives the example that a search containing "cooking" or "pictures" is a sign that "showing recipes or images may best match your intent".³

The old version of this article listed "four classic categories" of intent. There is no single classic list. The academic split goes back to a 2002 paper by Andrei Broder, which sorted web searches into navigational (take me to a site), informational (tell me something) and transactional (let me do something, often buy).⁴ Google's own guidelines for its human search quality raters use a different set of four. A Know query is made "to find information or explore a topic", and a Know Simple query is the special case whose correct answer fits in one or two sentences or a short list. A Do query is made "to accomplish a goal or engage in an activity", such as to download, buy or be entertained. A Website query looks for a specific site or page. A Visit in person query looks for a business, or a kind of business, nearby.⁵

The same guidelines say the line between Know and Do "is sometimes blurry" and that "many queries have more than one likely user intent". They are instructions for people who rate results, not a ranking formula, and they say no single rating can directly change how a page appears.⁵ I use the labels because they predict the shape of the page well, not because Google scores pages by them.

Worked example: Aoife's six queries, labelled

Aoife labels six of her top queries. "leeside cycles opening hours" is a Website query with a Visit in person flavour. "bike repair near me" is Visit in person. "how often should a bike be serviced" is Know Simple: one or two sentences settle it. "how to adjust disc brakes" is Know and Do at once, understanding with a spanner in hand. "gravel bike under €1,500" is Do, with research still going on. "best gravel bike for commuting" is Know on its way to Do.

Six queries, two of them mixed. That is normal. The label is a guess, and the next step is to check it against what Google actually shows.

Read the results page before you write anything

The quickest way to learn what Google thinks a query means is to look at the page it builds. A map with three businesses means it read local intent. A strip of videos means "show me". A row of products with prices means "I want to buy". The format is Google's reading of the intent, made visible.

One caution. Search Console's help says results "are specific to the time, place, device, and recent history of the person searching".⁶ Your view from the office chair is one sample. Look on a phone and a computer, signed out, and again a week later.

The flow runs from the top. The decision in the middle is the honest one: some features belong to a kind of business you are not.

Worked example: what Google showed Aoife

Aoife checks her six queries on a phone and a laptop, signed out, two Tuesdays running. Her notes:

Query	What the page showed	What she can earn
leeside cycles opening hours	Her business details from her Business Profile, then her site	Already hers. Keep the profile right
bike repair near me	A map with three local workshops, then links	A place in the map, through her Business Profile
how often should a bike be serviced	A boxed answer, then a related questions box	The boxed answer, with a quotable paragraph
how to adjust disc brakes	Videos near the top, then guides	A video result, with a proper video page
gravel bike under €1,500	A row of products with prices, then shops	Product results, with product markup and a feed
best gravel bike for commuting	Review articles from magazines, related questions	Little. Magazines own this one

Five of the six have something worth working for, each a different kind of work. The sixth she leaves alone. A "best gravel bikes" roundup against the cycling magazines would cost her a month and earn very little, and she is grand with that.

Featured snippets and People also ask are chosen, never marked up

The boxed answer is a featured snippet. Google describes it as a box "where the format of a regular search result is reversed, showing the descriptive snippet first", and notes that featured snippets also appear inside the related questions box.⁷ Its gallery says that when someone expands a question in People also ask, "they're shown a featured snippet", and about which questions appear there it says "you can't control what shows up here".¹

Can you mark a page up so that it becomes a featured snippet? Google's answer is one word: "You can't." Its systems decide whether a page "would make a good featured snippet" for a search, and if so they elevate it.⁷ The controls run the other way. The nosnippet rule blocks every snippet for a page, featured or not. Text inside an element with data-nosnippet is kept out of snippets. A lower max-snippet value makes a featured snippet less likely, without a guarantee.⁷

Two consequences. First, the old advice to add FAQ markup "for People also ask" was wrong twice over: the related questions box is filled by Google's own algorithms, not by markup, and the FAQ rich result has since disappeared altogether (more on that below). Second, there is a quiet cost to winning the box. Since 22 January 2020, when a page is shown as the featured snippet, Google stops showing that same URL again elsewhere in the main results.⁸ You do not get the box and the blue link. You get the box.

What you can do is write a passage that can be quoted whole, the question as a heading and the answer in the first sentence. The full version is in the article on passage indexing. Here is the short one.

Worked example: Aoife's servicing paragraph

Her guide used to open like this:

Servicing is something we get asked about all the time and the truth is every rider is different, so before we get into intervals let's talk about what a service actually involves.

Nothing in it answers the question. She rewrites the section:

How often should a bike be serviced? Once a year for most riders, or every 1,600 kilometres if you ride more than that. Commuters who ride through a Cork winter should add a quick check in spring, because grit and rain wear the chain and brake pads far faster than a summer of riding does.

The interval is her own workshop's rule of thumb, stated as hers. The first sentence could stand alone in a box. If Google quotes it, the page swaps its blue link for the top of the page. If not, the reader still gets the answer in the first line, which is the point anyway.

Rich results need structured data, and structured data guarantees nothing

Stars, a price and "In stock", a video with chapters: these are rich results, usually powered by structured data, a block of code that describes the page in the schema.org vocabulary. Google's Search gallery lists the types it currently supports, from Article and Breadcrumb to Product, Review snippet and Video, and says that following each guide makes a site "eligible" for them.⁹ Google recommends writing it as JSON-LD, because it is the easiest to maintain.¹⁰

Eligible is the important word. Google's general guidelines say that it "does not guarantee that your structured data will show up in search results", and that markup "enables a feature to be present, it does not guarantee that it will be present". The same guidelines say markup must describe what is on the page, and a structured data problem can lead to a manual action, which costs the page its eligibility for rich results while leaving its ranking alone.¹¹

Google does publish successes. Its introduction to structured data reports that Rotten Tomatoes measured a 25% higher click through rate on 100,000 pages with structured data than on pages without, and that Nestlé found pages shown as rich results had "an 82% higher click through rate".¹⁰ These are cases chosen by Google, comparing pages with and without the feature, not controlled experiments. Before borrowing anyone's uplift, write the rate down:

$\text{CTR} = \frac{\text{clicks}}{\text{impressions}}$

Here $\text{CTR}$ is the click through rate, $\text{clicks}$ the number of times someone clicked through to the page from Google, and $\text{impressions}$ the number of times the page was shown (Foundations: ratios and rates). An uplift is a relative change in that rate (Foundations: percentages).

Worked example: what a borrowed 25% would mean for Aoife

Aoife's bike pages get 30,000 impressions a month at a CTR of 2.0%, so $30{,}000 \times 0.020 = 600$ clicks. If her pages behaved like the Rotten Tomatoes case, the CTR would rise by a quarter:

$\text{CTR}_{\text{new}} = 0.020 \times 1.25 = 0.025$

That is 2.5%, and $30{,}000 \times 0.025 = 750$ clicks a month, 150 more (Foundations: percentages). I would treat that as an upper hope, not a forecast. Her pages are not film reviews, her rich result will not show every time, and only measuring tells her the real figure.

Features come and go, so build for the reader first

On 8 August 2023 Google said FAQ rich results would only be shown for well known, authoritative government and health websites, and limited HowTo rich results to desktop. On 13 September 2023 it removed HowTo results from desktop too, and the type was deprecated.¹² On 12 June 2025 it announced it was phasing out seven more structured data features, Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement and Vehicle Listing, because they were "not commonly used" and no longer added "significant additional value for users".¹³ In January 2025 it stopped showing breadcrumbs in mobile results, keeping them on desktop. And on 7 May 2026 the FAQ rich result stopped appearing in Google Search altogether. Google removed its documentation the following month.¹⁴

The timeline reads from the top. Every step removed a feature that many guides, this one included, used to recommend. A page written for the reader lost a decoration. A page written for the decoration lost its reason to exist.

Worked example: Aoife's FAQ markup on 40 pages

In 2022 an agency added FAQPage markup to 40 of Aoife's pages. Should she rip it all out?

What Google said in 2023 still applies: "Structured data that's not being used does not cause problems for Search, but also has no visible effects in Google Search."¹² No emergency, then. But on 11 of the 40 pages the questions in the markup appear nowhere on the page, which breaks the rule that markup describes visible content.¹¹ She removes those now, and the other 29 whenever each template is next touched. The questions readers really ask stay on the pages as plain headings with plain answers.

Shopping results: product snippets and merchant listings

For a shop, the product row is the most valuable piece of the page. Google splits product structured data into two classes. Product snippets are for pages where people cannot buy, such as an editorial review. Merchant listings are for "pages where customers can purchase products from you", with room for sizes, shipping and returns. Adding what merchant listings require usually makes a page eligible for product snippets too.¹⁵

The merchant listing rules are strict in useful ways. Only pages where a shopper can buy are eligible, and only pages for a single product or its variants, not category pages. It requires a name, an image and an offer, and the offer must be a single Offer, because "the merchant has to be the seller of the product".¹⁶ Google also says that structured data on the page plus a Merchant Center product feed, with free listings switched on, maximises eligibility, because some experiences combine the two.¹⁵

Worked example: the markup on one of Aoife's bikes

Aoife's own gravel bike, the Ballyhoura 2, sells for €1,349 and has 38 reviews from buyers on its product page, averaging 4.6 out of 5. Her developer puts this in the page:

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Ballyhoura 2 gravel bike",
  "image": [
    "https://leesidecycles.example/images/ballyhoura-2-1x1.jpg",
    "https://leesidecycles.example/images/ballyhoura-2-4x3.jpg",
    "https://leesidecycles.example/images/ballyhoura-2-16x9.jpg"
  ],
  "brand": { "@type": "Brand", "name": "Leeside" },
  "aggregateRating": { "@type": "AggregateRating", "ratingValue": "4.6", "reviewCount": "38" },
  "offers": {
    "@type": "Offer",
    "url": "https://leesidecycles.example/bikes/ballyhoura-2",
    "price": "1349.00",
    "priceCurrency": "EUR",
    "availability": "https://schema.org/InStock"
  }
}

Three images in the three shapes Google suggests, one offer in euro, the stock status, and a rating that matches the 38 reviews visible on the page. She checks it in Google's Rich Results Test and sends the same price and stock through her Merchant Center feed, so the two never disagree.

Review stars, and the rule that catches most shops out

Stars under a result come from review snippet markup. Google supports them for products, recipes, software, books and several other types, and for local businesses and organisations only on sites "that capture reviews about other" businesses.¹⁷ That last clause is the trap. If a business controls the reviews about itself, its pages using LocalBusiness or Organization markup are ineligible for stars, and Google names the common case: a widget of your own Google or Facebook reviews embedded on your own site.¹⁷

The other rules are plain. The reviews in the markup must be visible on the page. They must be about one specific item, not a category. You must not copy ratings from other websites. And Google forbids "fake or undisclosed incentivized reviews" in the page or the markup.¹⁷ In Ireland that last one is not only Google's rule either. The EU's Omnibus Directive put two review practices on the list that is banned in all circumstances: submitting or commissioning false reviews, or misrepresenting reviews, to promote a product, and claiming that reviews come from real buyers without reasonable steps to check.¹⁸ Ireland wrote both into the Consumer Protection Act 2007 through the Consumer Rights Act 2022, in force since the end of November 2022.¹⁹ So Aoife may ask every buyer for a review. She may not write, buy or commission fake ones.

Worked example: Aoife's two kinds of stars

Aoife's homepage shows a widget with her Google reviews, 4.8 from 212 reviews, and the agency's markup wraps it in LocalBusiness with an aggregate rating. That will not produce stars: she controls the page and it is about her own business. So it goes.

Her product pages are different. The 38 reviews of the Ballyhoura 2 are written by buyers about one product, shown in full, with no reward for a good one. Those are eligible. Same shop, two answers, because the rule is about who controls the reviews of what.

Images and videos: two formats with their own rules

For images, Google's recommendations are mechanical.²⁰ Use a real img element, because "Google doesn't index CSS images". Write alt text, which Google calls the most important attribute for an image and which also serves people using screen readers. Place images near the text they belong to. Use short descriptive filenames, which give "very light clues" and no more.²⁰

Video is stricter, and the old version of this article missed the rule. To be eligible for video features, including video results on the main results page and key moments, Google asks for a dedicated watch page, a page whose main purpose is to show one video. It lists "a blog post that reviews an embedded video" and "a product page with a 360 video" as examples of pages that are not watch pages. The watch page itself must be indexed and "performing well in Search" before its video is considered, and the video must not be hidden behind other elements.²¹ Key moments, the chapters under a video, can be set with Clip structured data, with SeekToAction, or in a YouTube description, and Google prioritises the ones you set over the ones it guesses.²¹

Worked example: Aoife's disc brake video

Aoife's four minute video on adjusting disc brakes sat two thirds of the way down a 2,000 word guide. It never appeared in video results, because the guide is a page with a video in it, not a watch page.

She makes a page at leesidecycles.example/videos/adjust-disc-brakes with the video at the top, a short description, the transcript below it and VideoObject markup with three Clip entries: "Check the rotor" at 0:20, "Centre the caliper" at 1:05, "Adjust the pad" at 2:40. The video stays embedded in the guide as well, which Google says is fine.²¹ Her hero image on the homepage, a CSS background until now, becomes an img with the alt text "Mechanic truing a road bike wheel in the Leeside workshop". Neither change is clever. Both remove a reason to be skipped.

Local results and knowledge panels: features you feed, not mark up

The map with three businesses is the local pack, and it is filled from Business Profiles. Google says local ranking rests on relevance ("how well a Business Profile matches what someone is searching for"), distance ("how far each business is from the customer who's searching") and prominence (how well known a business is), and that "there's no way to request or pay for a better local ranking".²² The knowledge panel, the box of facts about a business or a person, is "automatically generated" from sources across the web, and its subject can claim it and suggest changes.²³ Sitelinks, the extra links under a result, are automated too.²⁴

Worked example: "bike repair near me" from two sides of the city

Someone two kilometres from Aoife's workshop searches "bike repair near me" and sees her in the map. Someone five kilometres away, out past Douglas, sees three other workshops. Her website cannot change that. She can change relevance and prominence: the right categories in her profile ("Bicycle repair shop" as well as "Bicycle store"), her real opening hours, photos of the workshop, and replies to her reviews. Distance is geography, and she leaves it to the map.

Measuring the real estate you actually hold

Search Console's Performance report can split clicks and impressions by search appearance, alongside queries, pages, countries and devices.⁶ That shows whether a feature earns anything.

The counting rules differ by feature. A link inside a People also ask item gets an impression only when someone expands the question, and the click that expands it does not count as a click. A featured snippet follows the standard rules. An image thumbnail that someone clicks to enlarge is not a click, until they leave Google for your page.² A small People also ask count means few people opened the question, not failure.

To see which feature carries your traffic, take each one's share of the total:

$s_k = \frac{c_k}{\sum_{j} c_j} \times 100$

Here $s_k$ is the share of clicks from appearance $k$ as a percentage, $c_k$ the clicks with that appearance, and $\sum_{j} c_j$ the clicks across all the appearances you are comparing (Foundations: percentages, Foundations: ratios and rates).

Worked example: Aoife's quarter, by appearance

Three months after the changes, Aoife exports her clicks by search appearance.

Along the bottom are four search appearances, up the side the clicks each brought in over the quarter. The product features carry most of it. Of the 4,550 clicks across the four appearances, product snippets bring

$s_{\text{product}} = \frac{2{,}140}{4{,}550} \times 100 = 47.0$

so 47.0%, and videos bring $\frac{240}{4{,}550} \times 100 \approx 5.3\%$ (Foundations: percentages). The video has not failed: it is three months old, and 240 clicks is a start she did not have. The takeaway is that her product data is now her most valuable search asset, so keeping prices and stock right is a weekly job, not a launch task.

Where AI answers fit in

Since March 2025 an AI Overview can sit above all of this in Ireland, and since October 2025 AI Mode can replace the page. What that does to clicks is in the article on AI Overviews and organic traffic. How a page becomes a cited source inside those answers is in the article on AEO and GEO.

One point joins the two. Google's guide for site owners says its AI answers can include images, videos, product listings and local business details, and that its image and video guidance, Merchant Center and Business Profile help in both the AI responses and the rest of Search. It needs no special markup, and structured data is still worth using for rich results.²⁵ The work in this article is part of what those answers are built from.

What I do on the sites I build

I start with the query list, not the page list. I label each important query, check the label against the results page on a phone and a computer a week apart, and write down which feature, if any, the business could realistically earn.

Then the dull, durable work. A quotable first sentence in every section that answers a question. Product markup that matches the page and the feed. Review markup only where the reviews are about the product and visible. A watch page for every video worth finding. Real image elements with real alt text. A complete Business Profile. Measured by search appearance, quarter on quarter.

And I have dropped most of what the old version of this article recommended. FAQ markup for People also ask: the box was never fed by markup, and the FAQ rich result is gone. A video inside the article: it needs its own page. A client whose impressions "grew by 60%": I cannot source it, so it is cut. The goal is to answer the question in the shape the person wanted, on a page honest about what it is. When Google changes the decorations, as it did five times in three years, that page loses nothing that mattered.

Footnotes

1. Google, Visual elements gallery of Google Search, Google Search Central documentation, last updated 4 February 2026, https://developers.google.com/search/docs/appearance/visual-elements-gallery

2. Google, What are impressions, position, and clicks?, Search Console Help, accessed 8 November 2026, https://support.google.com/webmasters/answer/7042828?hl=en

3. Google, How does Google determine ranking results (Meaning of your query), Google Search, accessed 8 November 2026, https://www.google.com/search/howsearchworks/how-search-works/ranking-results/

4. Andrei Z. Broder, A taxonomy of web search, ACM SIGIR Forum, volume 36, issue 2, pages 3 to 10, September 2002, https://doi.org/10.1145/792550.792552

5. Google, Search Quality Rater Guidelines (General Guidelines), section 12.7 Understanding User Intent, 11 September 2025, https://guidelines.raterhub.com/searchqualityevaluatorguidelines.pdf

6. Google, Performance report (Search results): Overview and basic setup, Search Console Help, accessed 8 November 2026, https://support.google.com/webmasters/answer/7576553?hl=en

7. Google, Featured snippets and your website, Google Search Central documentation, last updated 10 December 2025, https://developers.google.com/search/docs/appearance/featured-snippets

8. Barry Schwartz, Everything You Need To Know About The Google Featured Snippet Deduplication Change, Search Engine Roundtable, 30 January 2020, https://www.seroundtable.com/google-featured-snippet-deduplication-change-summary-28921.html

9. Google, Structured data markup that Google Search supports (Search gallery), Google Search Central documentation, last updated 16 October 2026, https://developers.google.com/search/docs/appearance/structured-data/search-gallery

10. Google, Introduction to structured data markup in Google Search, Google Search Central documentation, last updated 10 December 2025, https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data

11. Google, General structured data guidelines, Google Search Central documentation, last updated 10 July 2026, https://developers.google.com/search/docs/appearance/structured-data/sd-policies

12. John Mueller, Changes to HowTo and FAQ rich results, Google Search Central Blog, 8 August 2023, updated 14 September 2023, https://developers.google.com/search/blog/2023/08/howto-faq-changes

13. Henry Hsu, Simplifying the search results page, Google Search Central Blog, 12 June 2025, https://developers.google.com/search/blog/2025/06/simplifying-search-results

14. Google, Latest documentation updates (entries of 22 January 2025, 8 May 2026 and 15 June 2026), Google Search Central, accessed 8 November 2026, https://developers.google.com/search/updates#deprecating-the-faq-rich-result-feature

15. Google, Introduction to Product structured data, Google Search Central documentation, last updated 10 December 2025, https://developers.google.com/search/docs/appearance/structured-data/product

16. Google, Merchant listing (Product, Offer) structured data, Google Search Central documentation, last updated 8 September 2026, https://developers.google.com/search/docs/appearance/structured-data/merchant-listing

17. Google, Review snippet (Review, AggregateRating) structured data, Google Search Central documentation, last updated 8 September 2026, https://developers.google.com/search/docs/appearance/structured-data/review-snippet

18. European Parliament and Council, Directive (EU) 2019/2161 (the Omnibus Directive), Article 3, inserting Article 7(6) and Annex I points 23b and 23c into Directive 2005/29/EC, Official Journal of the European Union, 2019, https://eur-lex.europa.eu/eli/dir/2019/2161/oj/eng

19. Law Reform Commission, Consumer Protection Act 2007, revised, sections 46(3B) and 55(1)(za) and (zb), as inserted by the Consumer Rights Act 2022, Revised Acts, https://revisedacts.lawreform.ie/eli/2007/act/19/revised/en/html

20. Google, Google image SEO best practices, Google Search Central documentation, last updated 2 March 2026, https://developers.google.com/search/docs/appearance/google-images

21. Google, Video SEO best practices, Google Search Central documentation, last updated 18 December 2025, https://developers.google.com/search/docs/appearance/video

22. Google, Tips to improve your local ranking on Google, Google Business Profile Help, accessed 8 November 2026, https://support.google.com/business/answer/7091?hl=en

23. Google, About knowledge panels, Knowledge Panel Help, accessed 8 November 2026, https://support.google.com/knowledgepanel/answer/9163198?hl=en

24. Google, Sitelinks, Google Search Central documentation, last updated 10 December 2025, https://developers.google.com/search/docs/appearance/sitelinks

25. Google, guide for site owners on generative AI features in Google Search (page title in American spelling), Google Search Central documentation, last updated 10 July 2026, https://developers.google.com/search/docs/fundamentals/ai-optimization-guide