Untapped SEO opportunities: how to find them in your own data

What an untapped SEO opportunity is

An opportunity is untapped when the demand already exists and something you control stands between it and a click. That is different from classic keyword research, which looks for new demand. Here the evidence is in your own data: Google has already decided your page is relevant enough to show, or searchers already look for a format you have and have not exposed.

Most opportunities fall into a handful of patterns:

Start with your own data for a practical reason too. Google's guidance on third-party SEO tools, published in June 2026, says those tools have no access to its internal ranking data and that their predictions are their own, and it points site owners to Search Console instead. Third-party estimates earn their place for competitors, where you have nothing better.

Get the data out of Search Console

The Performance report for Search results shows four metrics for every query and page: clicks, impressions, click-through rate (CTR, clicks divided by impressions) and average position. Two definitions matter for everything that follows. An impression is counted when a link to your site appears on the page of results the searcher is viewing, whether or not they scrolled to it; a result on a page nobody opened gets no impression. Average position is the topmost position your site or page held, averaged over the searches where it got an impression.

The report has limits that hide exactly the long tail you are looking for:

• The table shows at most 1,000 rows. Rarer queries and pages drop out of the table while still counting in the chart totals.
• Anonymized queries are left out of the query table to protect privacy, so its rows never add up to the chart.
• History goes back 16 months, as it has since Google introduced the new Search Console in 2018. That is enough for a year-over-year comparison, and no more.

For a small site the interface is enough. Past a thousand meaningful rows, use the Search Analytics API, which returns up to 25,000 rows per request and pages through the rest, or the bulk data export to BigQuery, which delivers every row except anonymized queries.

Two filters make the rest of this guide faster. The branded queries filter, announced in November 2025 and available to all eligible sites since March 2026, splits branded from non-branded queries. The non-branded view shows how new users find your content, which Google's announcement describes as the source of organic growth. It is not available for URL-path or subdomain properties, or for sites with few impressions. The regex filter (Custom (regex) under Queries or Pages) uses RE2 syntax, matches anywhere in the string unless you anchor it with ^ or $, and ignores case by default.

Search Console Insights, moved into the main interface in June 2025, adds a lighter view of top, trending up and trending down pages and queries. Google calls trending-up queries a source of new content ideas, and says trending-down pages may need a refresh. Since October 2025, properties with many queries also get query groups, which cluster misspellings and rephrasings of one search so you can judge the demand behind it as a whole.

Queries with impressions but few clicks

Google's Performance report help describes this case directly: a low CTR can mean people see your page in results but do not think it answers their query. To list the candidates, open the Queries tab, turn on impressions and CTR, sort by impressions and read down for queries with a CTR well below your site's average. Then do the same on the Pages tab.

Before rewriting anything, find out why the CTR is low. Google's guide to analyzing queries with a bubble chart plots average position against CTR, with bubble size for clicks, and lists the usual causes for queries where you rank near the top but few people click:

• Competitors show rich results, such as review stars or video thumbnails, that draw the eye away from a plain link.
• You rank by accident for a query your page is not really about. That may be fine to ignore.
• The searcher already got the answer on the results page, such as opening hours or a phone number. If that answer serves your goal, the low CTR is working as intended. For AI answers on the results page, see AI Overviews and your traffic.

When the result itself is the problem, work on the title link and the snippet. Google builds the title link from several sources: the <title> element, the main visible title, headings such as the H1, og:title, other prominent text and the anchor text of links pointing at the page. A descriptive, specific <title> that matches the page's main heading is the most reliable lever, because Google replaces titles it finds half-empty, obsolete or inaccurate. Snippets come mainly from the page content, and Google uses the meta description when it describes the page more accurately. Changes show only after Google recrawls and reprocesses the page, which its documentation puts at a few days to a few weeks.

Pages in striking distance of page one

"Striking distance" is SEO shorthand, not a Google term, for queries where a page ranks just below the results most people click: in practice, an average position of roughly 8 to 20. Google's position help shows what that means. Position 11 can be the first result on page two when page one held ten results, and a result on a page the searcher never opened records no impression and no position. A page in this range is being shown, but to a fraction of the people searching.

The bubble-chart guide singles out queries with a below-average position but an above-average CTR: people who do see the result choose it, so moving it up should pay off. Its advice is concrete. If you have no page for the query, consider creating one that gathers what your site knows about the subject; if you have one, add content that meets the need better.

For a larger site, the BigQuery export builds the list in one query. The export stores position as a zero-based sum, so Google's sample queries add one to get the average position the report shows, and anonymized queries arrive as empty strings:

/* Queries where a URL averages position 8 to 20 over the last 90 days */
SELECT
  url,
  query,
  SUM(impressions) AS impressions,
  SUM(clicks) AS clicks,
  SUM(clicks) / SUM(impressions) AS ctr,
  (SUM(sum_position) / SUM(impressions)) + 1.0 AS avg_position
FROM searchconsole.searchdata_url_impression
WHERE search_type = 'WEB'
  AND query != ''
  AND data_date BETWEEN DATE_SUB(CURRENT_DATE(), INTERVAL 90 DAY) AND CURRENT_DATE()
GROUP BY url, query
HAVING avg_position BETWEEN 8 AND 20 AND impressions >= 50
ORDER BY impressions DESC
LIMIT 500

Treat average position as a guide rather than a target. Google recommends watching trends in impressions and clicks more than position alone, because one position number can mean very different things on different result layouts.

Question and long-tail queries you already rank for

Long-tail queries are the longer, more specific searches that each bring few impressions but add up across a site. They are also where a page is often shown for a question it only half answers. Two regex filters on the Queries tab surface them. The first matches queries that start with a question word; the second matches queries of six words or more:

^(who|what|when|where|why|how|which|can|does|do|is|are|should)\s
^(\S+\s+){5,}\S+$

Sort the result by impressions. A cluster of similar questions landing on one page usually means the page needs a section that answers them directly, in the words searchers use. On larger properties, query groups in Insights do this clustering for you.

Two warnings from Google apply here. Its guide to optimizing for generative AI features, added in May 2026, says that creating separate content for every variation of how people might search, primarily to manipulate rankings, violates the scaled content abuse spam policy. Its helpful content guidance lists writing about things only because they seem to be trending as a warning sign. The fix for many phrasings of one question is one good answer, not one page per phrasing.

For demand you do not see yet, Google Trends shows whether interest in a term is rising, and Google notes that rising terms can be less known and so less competitive. Its Explore tool compares up to five terms at once. For approximate search volumes of variations, the bubble-chart guide points to Keyword Planner in Google Ads.

Pages that rank for queries they do not answer

Filter the Performance report to a single page and read its queries. Pages often collect queries they were never written for. Some are near misses, where the page mentions the topic in one line; others are a different intent altogether, such as a pricing page shown for how-to questions. Click a query to see every page of yours Google showed for it. Two of your pages sharing one query can be a sign that neither is the clear answer.

The decision is simple to state:

1. Same intent, missing detail: add a section to the page that already ranks. This is usually the cheapest fix.
2. Different intent: write a separate page for it, and link the two.
3. Two pages competing for one intent: merge them into the stronger page and redirect the other with a 301.
4. Unrelated query: ignore it, or remove the wording that attracts it if the traffic is unwanted.

Content and keyword gaps against competitors

A content gap is a subtopic that competing pages cover and yours does not. The manual way to find one is to search your target queries, open the pages that outrank you, and list what they explain that you skip: a comparison, a step, a definition, a table of options. Commercial tools automate the keyword side; Ahrefs' Content Gap, for example, lists keywords your competitors rank for and you do not. Those lists come from each vendor's own crawler and estimates, which is why Google's guidance on third-party tools steers you back to Search Console for data about your own site.

Filter any gap list through one question from Google's helpful content guidance: would your existing or intended audience find this useful if they came to you directly? A gap is worth closing when your business has something real to say about it. Entering a topic without real expertise, mainly because it looks like search traffic, is one of the warning signs on the same list.

Internal links to the pages that deserve more

Of all the links Google looks at, internal links are the ones you control completely. Google uses links to find pages and to judge their relevance, and its link best practices are plain: every page you care about should have a link from at least one other page on your site, with anchor text that is descriptive and reasonably concise.

To find pages that deserve more links:

1. Take the striking-distance list from above.
2. In the Links report, open the top internally linked pages and check how many internal links each of those pages receives. The report is a sample and its tables stop at 1,000 rows, so on a larger site a crawler gives the complete count.
3. Find pages that already earn traffic on the same subject, and add a link from their body text, in a sentence that tells the reader why to follow it.
4. Look for orphan pages: URLs in your sitemap that no internal link reaches.

Google says there is no ideal number of links per page. The goal is that a reader, and Googlebot, can get from your strongest pages to the one that needs help in a click or two.

Video SEO: the opportunity in videos you already have

Video is where "untapped" is most literal. Google can show videos in the main results, in Video mode, in Google Images and in Discover, but only videos it can index, and a video placed on the wrong kind of page is not eligible at all. The requirements, from Google's video SEO best practices:

• A watch page. To be eligible for video features, a video needs a page whose main purpose is to show that one video. A blog post that reviews an embedded video, or a product page with a 360-degree view, is not a watch page. Such a page can still appear as a text result, and in Google Images with a video badge.
• An indexed page that performs. The watch page must be indexed, and Google says it must be performing well in Search before its video is considered for indexing.
• Markup Google can find. Embed with <video>, <embed>, <iframe> or <object>, check that the player appears in the rendered HTML, and never require a click or swipe to load it.
• A stable thumbnail. Every video needs a valid thumbnail at a URL that does not change. CDN links that expire quickly can stop the video being indexed.
• Metadata: VideoObject structured data, a video sitemap or Open Graph tags.

VideoObject requires three properties: name, thumbnailUrl and uploadDate. Google recommends contentUrl, the URL of the video file itself and the most effective way for it to fetch your video, or embedUrl when that is not available, plus description and duration. Since September 24, 2026, the documentation also lists creator (or author). A minimal block, placed in a <script type="application/ld+json"> element on the watch page:

{
  "@context": "https://schema.org",
  "@type": "VideoObject",
  "name": "Replace a laptop battery safely",
  "description": "Opening the case, disconnecting the old battery and fitting the new one.",
  "thumbnailUrl": "https://example.com/video/battery/thumb.jpg",
  "uploadDate": "2026-09-01T09:00:00-04:00",
  "duration": "PT6M30S",
  "contentUrl": "https://example.com/video/battery/video.mp4"
}

Key moments let searchers jump to a segment. Google detects segments on its own, but gives priority to the ones you define: Clip markup with a start time, end time and label for each segment (supported in every language Google Search offers), SeekToAction markup that tells Google where timestamps go in your URLs (a shorter list of languages), or timestamps in a YouTube description. Video previews need Google to be able to fetch the video file itself.

If your videos live on YouTube, two things apply. Google may index a video both on your page and on YouTube, and it still recommends structured data on your own watch page. And since July 2026, Search Console platform properties cover YouTube, Instagram, TikTok and X, so you can see which Google searches find your channel. Google's guide to that data suggests comparing /watch URLs with /shorts/ to see which format search favours for you. On your own site, track progress in the Video indexing report, which gives reasons such as "Video isn't on a watch page", and in the Performance report with the Videos search appearance filter.

Images and multimodal search

Images are the second format to check, and searches that start from an image now have their own data. Since September 24, 2026, Search Console reports a web multimodal search type covering Google Lens, Circle to Search on Android, image uploads to Google Search and Chrome's "Search this image". The queries dimension is not available for that type, because the search is an image rather than text.

Google's image SEO best practices come down to making images findable and giving them context:

• Use an <img> element with a src attribute; inside <picture> is fine. Google does not index CSS background images.
• List images Google might not find in an image sitemap, which may include URLs on a CDN's domain.
• Use a supported format: BMP, GIF, JPEG, PNG, WebP, SVG or AVIF.
• Write alt text that describes the image in the context of the page. Google calls alt text the most important attribute for image metadata, and warns that keyword-stuffed alt text can make a site look like spam.
• Place images near the text they illustrate. Short, descriptive file names help too, though Google says they give only very light clues.
• For images you license, image metadata structured data lets Google Images show the creator, the licence and credit information.

To see what images already earn, switch the Performance report's search type to Image and read the queries and pages as you did for web results.

Rich results that fit the content

Rich results are the other reason a competitor's result can outdraw yours at the same position. The opportunity is to mark up what a page genuinely is, using a type Google still supports, and to validate it. As of June 2026, Google's structured data gallery lists types including Article, Breadcrumb, Product, Review snippet, Recipe, Event, Job posting, Local business, Organization, Profile page, Discussion forum, Q&A, Software app and Video. Markup makes a page eligible; Google does not guarantee a rich result.

Validate markup with the Rich Results Test while you build, and watch the rich result reports in Search Console after release, since Google warns that templating or serving issues can break markup after deployment. The Search appearance tab of the Performance report shows which result types already earn you impressions.

Prioritize, change one thing, measure

A list of opportunities is only useful in order. Rank each item by the demand already in play (impressions for the query or page), how close it is (position and CTR) and the effort to fix it. A new title is a small edit, a new section is a writing task, and a watch page template is a development project.

Then change one thing at a time on a given page, and record it. Custom annotations, added in November 2025, let you right-click the Performance chart and note a change in up to 120 characters; everyone with access to the property sees them, so keep personal details out. When you compare before and after, use the weekly or monthly view added in December 2025, which removes the day-of-week mismatches from a daily comparison. Google's help is candid that periodic shifts in interest, news events and competitors' improvements also move the numbers, so look for a change that lines up with your own date and compare it with similar periods in the past.

If you would rather hand this over, our SEO service starts with a technical audit: a crawl of the whole site, a read of Search Console and, where you have them, your server logs, with the issues ranked by impact and effort. Content plans are built from search demand, and every brief maps to one search intent and one page. More on the subject is under SEO.