A featured snippet is the highlighted box that sits above every regular organic result, position zero, pulling a quick, direct answer straight from a page in the index. Once that box shares the page with AI Overviews, ads and shopping boxes, position one stops mapping to one click rate.
In public testing of improved click estimates shared in a DataForSEO thread in early September, the revised numbers read the actual results page, AI Overviews, featured snippets, ads, shopping boxes, and the implied click rate for position one ran anywhere from 0.04 percent to 64.86 percent depending on what else was on the page. The snippet is one of the surfaces that breaks position-equals-traffic.
Position Zero: Where the Box Sits
Semrush's guide and the reference explainers agree on the shape: a highlighted text box at the top of the results page giving a quick, direct answer pulled from a webpage, called position zero because it sits above all the regular organic results.
Why the spot matters is screen real estate. The box takes the top of the page, up to about half a mobile screen, and pushes competitors below the fold. It answers immediately for people who want a quick fact, still earns clicks from people who want more, feeds voice assistants their read-aloud answer, and signals that Google trusts the source. Google's own Search Central guide, on featured snippets and your website, is the official statement of how the box is chosen and how to opt a page out of it.
The Four Types
The catalog is stable.
- Paragraphs. A short block of forty to sixty words that defines a term or explains a concept.
- Lists. Bulleted or numbered steps, rankings or how-to guides, pulled from lists or headings in the source.
- Tables. Rows and columns for comparisons, pricing or schedules.
- Videos. Short clips that jump to the timestamp answering an action-based query.
Match the shape you target to the shape already showing. A definition query wants a paragraph, a steps query wants a list, a comparison wants a table, an action query wants a video timestamp.
Featured Snippets vs AI Overviews
The two boxes get confused because both sit at the top and both answer. Semrush draws the line this way: a featured snippet pulls a direct passage from a single webpage, while an AI Overview generates a multi-source summary using a model such as Gemini. Both can appear on the same page, and the snippet keeps a highly visible spot on queries where no Overview triggers.
The practical difference is attribution. A snippet credits one page, with a link, and that page gets the click. An Overview cites several and may resolve none of them cleanly. When you read a results page, note which of the two is present, because they move traffic in different directions.
How to Get Featured Snippets
The earning method is one vendor's, Semrush's, and it matches what practitioners describe.
- Ask the question in a subheading and answer it directly in forty to fifty words, the direct statement first and the context after.
- Target long-tail question queries, where snippets typically appear.
- Organize lists under clear headings and put data in clean tables with keyword-bearing column heads, so rows and columns pull cleanly.
- Mine People Also Ask questions as snippet-shaped targets. Turning those questions into headings is the same move, one layer earlier.
All of that improves the chance of being picked. None of it is a promise of inclusion, and Google's guide says as much.
Reading the Actual Results Page
The click-estimate test above is the pattern to copy. Do not estimate from position alone. Pull the live page, note which surfaces are present (Overview, snippet, ads, shopping) and let that presence adjust the estimate before any difficulty or intent call. A position one with a snippet and a shopping box above it is not the same inventory as a bare position one, even at the same rank.
Make the pull a metered API fetch with a retry backoff and a cache in front of it, and stamp each presence read with keyword, locale and week. The date is what turns a sighting into something a later estimate, difficulty note or brief can cite and re-check.
A worked read shows why the presence matters more than the rank. Take a keyword where your page sits at position one. On a bare page, ten blue links and nothing else, position one collects the large share of clicks the old curves promised. Put a featured snippet above it, sourced from someone else's page, and a fraction of searchers get their answer in the box and never scroll; your position has not moved and your traffic has. Put an AI Overview above both and the box now cites several pages and may resolve none cleanly, so the traffic that leaves the page leaves in a different direction again. Add a shopping carousel and the first organic result is now below the fold on a phone. Same rank, four different pages, four different click rates, and the DataForSEO thread's 0.04 to 64.86 percent spread is that range measured across real pages.
So the row you record is not "position 1." It is: keyword, locale, date, snippet present or absent, snippet type and source, Overview present or absent, other surfaces present, and then the rank. The rank is the least informative field in it.
People asking about snippets ask three things: what one is, how the box works, and how to get one, with a durability tail (are they still a thing?) and a failure tail (why is mine not showing?). The durability answer is that they are still there, sharing the top of the page with Overviews rather than displaced by them. The not-showing answer is almost always a dated pull away: the box arrived, changed type, or dropped, and only a series of dated reads tells you which.
Snippet Presence with a Pull Date
For your own tracking, record the same four things every week: snippet present or absent, its type, its source URL, and what else shared the page. Compare week to week to see the box arrive, change type or drop. qequals keeps weekly snapshots of the live page, which gives you that series without building the storage yourself; the reading is still yours to do. That dated series is the proof a later click estimate, difficulty note or intent claim stands on.
