The fastest way to learn what kind of page a query wants is to search it and describe what comes back. Record each result as a page type, a format and a title promise, then plan your own page against that record. No ranking system guarantees an outcome, but page one is better evidence than a hunch. Picture the editor of a running gear site deciding how to cover "trail running shoes": the results page tells that editor whether a buying guide, a category page or a video deserves the effort.
What the results page shows before you write a word
Google's guide to its ranking systems names several systems that work on understanding queries. BERT is described as a system that helps Google "understand how combinations of words express different meanings and intent." RankBrain helps Google understand how words relate to concepts, so it can return relevant results even when a page lacks the exact words of the query. Neural matching connects concepts in queries with concepts on pages. The guide also states that MUM is not currently used for general ranking in Search.
For an editor, the lesson is practical. Stuffing the query words into a heading does not match a page to an intent. A page matches when it answers the concept behind the words in the form the searcher expects, and the expected form is visible in what the results page already offers.
Why observation beats labels from a tool
A keyword tool might tag a query as "informational" or "commercial". The tag cannot tell you whether the searcher wants a shortlist, a size chart or a store. Page one shows what the system chose to present for that query on that day. Treat it as evidence you collected yourself, not as a rule.
What page one cannot tell you
The results page reveals page types, not page quality. Ten sites can publish the same type and all of them can be thin. Take the type from the results and take the depth from your reader's questions. Results also move, so write down the date, the device and the language next to every observation.
A scoring sheet for page one
A repeatable sheet keeps you from remembering only the results that confirm your plan. Give every visible result a row and fill in the same columns each time.
Row fields
- Position and the type of page: guide, category, product page, ranked list, forum thread, manufacturer page, glossary entry.
- Format of the opening screen: short definition, numbered steps, comparison table, product grid, embedded video.
- The promise in the title: how-to, best-of, price, comparison, a year, a question.
- Extra elements around the result: images, videos, related questions, related searches.
Reading the title promise
Titles repeat patterns. When three of five results promise a ranked shortlist, a reader arriving from that query expects a shortlist. You do not have to copy the wording, but your title has to keep the same promise honestly. A title that promises a shortlist on a page that is really a sales grid disappoints the reader, and the reader leaves.
| What you counted | What it suggests | Page decision |
|---|---|---|
| Product grids and store pages fill the page | The searcher is close to a purchase | Plan a category or product page, with guidance as a supporting block |
| Step lists and explainers fill the page | The searcher wants to learn | Plan an explainer with a clear answer near the top |
| Ranked lists and comparisons fill the page | The searcher wants options side by side | Plan a comparison with stated criteria |
| No single type dominates | The query is ambiguous or serves several needs | Choose one main job for the page and narrow the query |
Table: from counts to a page decision. The words "fill the page" describe your tally, not a statistic.
Naming what you see with Google's vocabulary
The Search Gallery page defines the visual elements of a result, and its names make a good shared language for the team. A text result is based on the textual content of a page. A rich result relies on structured data in the markup of the page to show graphical or interactive elements. An image result is based on an image embedded in the page. A video result is based on a video embedded in the page. Exploration features help searchers expand and refine their search.
Parts of a text result
The page lists the title link, the snippet, the byline date, sitelinks and rich attributes as separate elements. The byline date is the date Google estimates the page was updated or published. If several results on page one carry recent dates, that hints at a freshness expectation, though it proves nothing by itself.
Image and video results
An image result rests on an embedded image and a video result rests on an embedded video. If those formats appear for your query, a page that wants to take part needs the media on the page itself. Whether the extra production cost is justified is a separate question, handled at the query level.
Related searches and related questions
The related searches group is a cluster of searches other people made. The related questions group is a cluster of questions connected to the first search. Neither suggests a new page type. Both hint at what the reader asks next, which makes them useful raw material for subheadings.
Result type and the work it implies
| Result type you aim for | What must exist on your page | Cost you carry |
|---|---|---|
| Text result | Content that answers the query, and a clear title | Writing and editing time |
| Rich result | Matching structured data on top of the content | Developer and testing time |
| Image result | An on-topic image embedded in the page | Photography and editing |
| Video result | An embedded video | Filming, editing and hosting |
Google decides which result types appear. The table only lists the materials you prepare and promises no particular result.
Title promises and what readers expect
| Title promise | What the reader expects | What the page must contain |
|---|---|---|
| "How to ..." | Steps they can follow | Numbered steps, one action each, with a clear finish line |
| "Best ..." | A ranked or grouped shortlist | Stated criteria, the reason for each pick, first-hand evidence |
| "... vs ..." | A side-by-side judgment | A comparison table and a rule for choosing |
| "... price" | Current figures or a way to see them | Prices or a path to them, and what changes the price |
| A bare product name | Products, or the right brand page | Product list or brand page, not an essay |
Use the table as a gap check on your own draft. If your title makes one of these promises, the page has to hold the matching material in the first screens.
Turning counts into a purpose: four outcomes
Practical suggestion: classify each query by what the searcher wants to end up with. This is a working method for choosing a page format. Google's documentation does not define these four outcomes.
| Outcome | Signs on page one | Format to build | Success test |
|---|---|---|---|
| An answer | Explainers, glossary entries, related questions | Answer-first article | The reader does not need a second page |
| A shortlist | Ranked lists, comparisons | Criteria-led comparison | The reader narrows the options |
| A purchase | Category and product pages | Category or product page | The reader reaches the product |
| A destination | Brand pages, sitelinks | Strengthen the brand page | The reader reaches the right page in one step |
Do not infer the outcome from the wording alone
A query containing "buy" can still return mainly explainers, and a query containing "how" can return a shop. The wording is a clue. The outcome is whatever the result types say. When the two disagree, follow the result types.
One promise per page
Write the page's promise in one sentence and test every section against it. If the promise is "the reader can name the shoe that fits their terrain", a long history of the sport has no place in the page. Move it elsewhere or cut it.
The tally rule
Decision rule: count the result types on page one. Build the page around the type that fills the most rows, fold the second type in as a section, and give the third type a page of its own. Leave the leading type only when you can name a concrete benefit to the reader, such as your own measurements, your own test photos or a comparison nobody else publishes. Being different for its own sake is not a reason.
A tally worked through
Example scenario: page one shows five category pages, two explainers and one video. Build the category page, add a short selection block that borrows from the explainers, and skip the video unless it shows the shoe in real use. When the tally ends level, split the work across two pages and put a short pointer to the sibling page in each introduction.
When page one is a mixed bag
Sometimes no type dominates. That can mean searchers want different things, or that the system is hedging between interpretations. You have three moves.
- Narrow the query. Add a word that fixes the intent, such as "trail running shoes for wet rock", and target that more specific query.
- Choose the main job. A page that tries to be a guide and a store at once can do both badly.
- Add a short routing note in the introduction. Two sentences can send a reader to the right section or the right page.
Practical suggestion: search the ambiguous query on different days and devices. A stable mix points to real ambiguity. A changing mix points to unstable results, and you should hold off on a large investment.
Duplicates, site diversity and freshness
Why a second page of yours may not help
The ranking guide describes deduplication systems that show only the relevant results to avoid unhelpful duplication. It also describes a site diversity system that generally avoids showing more than two page listings from the same site in top results, while allowing more when they are especially relevant. If two of your pages answer one query, decide which page owns it.
Queries that expect fresh content
The "query deserves freshness" systems are designed to show fresher content when it would be expected. Google gives the example of a newly released movie, where recent reviews matter more than older articles from production. After an earthquake, news can surface even for a general query. If page one is full of dated results, keep your date visible and update the substance of the page, not only the date.
Carrying the sheet into the page
The reading changes the page itself, not only the choice of type.
Title link
The title link is the name of your page in a result. If the competing titles promise steps, your title must be able to deliver steps. Write what the reader will get instead of lining up the query words.
Opening screen
For an answer outcome, put a short answer first. For a purchase outcome, show products or a clear path to them. For a shortlist outcome, put the comparison summary near the top. If the opening screen and the outcome disagree, the reader scrolls to find what they came for.
Subheadings
Related questions are raw material for headings. Do not paste them. Order them the way a reader thinks: the core question first, the detail next, the decision last. Under each heading, put something new: a table, a rule or a worked example.
Photos and video
Where image or video results appear, use your media as evidence and not as decoration. A photo of the sole after twenty trail runs says more than a stock image of a shoe on a rock.
A 20 minute reading routine
Checklist: use these steps in order and stop at twenty minutes.
- Write the query and note the device, language and date.
- Enter every visible result in the sheet with type, format and title promise.
- Mark whether image, video, related searches or related questions appear.
- Count the types and write down the top two.
- Note how many results carry dates and how recent they look.
- Mark any site that appears more than once.
- Place the query in one of the four outcomes, or two when needed.
- Write the page decision: format, main job, supporting section.
- List two or three questions the visible pages leave unanswered.
- Set the date for the next look.
Variables that change what you see
Signed-in state and location can affect what you see. Practical suggestion: repeat the search under a second condition, such as a private window on another device, before you rely on the tally.
A filled-in record
Example scenario: query "trail running shoes", phone, English, page one with five store categories, two guides and a video. Main type: category. Supporting type: guide. Outcome: a purchase. Decision: add a selection block to the existing category page and write no new guide. Confidence: high, because the types agree. Next check: in three months.
Keeping the record useful
A reading that nobody revisits turns into an archive. Give each query a record with the query, date, device, language, type counts, main and supporting type, whether media appeared, the date weighting, the decision, the target address and the next check date. The fields look like a lot, but each one answers a later question: why did we build it this way?
When to repeat the reading
Practical suggestion: look again every three to six months for queries whose results are heavily dated, and before major edits for the rest. This interval is a working habit and not a rule.
Checking the call after publication
After the page goes live, open Search Console and see which queries show impressions for it. If it shows nothing for the target query, do not assume the page is weak. The results page may have changed, the type call may have been wrong or the page may miss the intent. Repeat the reading and compare it with the first one before you edit.
Pairing the sheet with your own site's data
The sheet describes the market. Your own data describes your position in it. Before you build, collect three facts about your site: which of your pages already receive impressions for the query, which page Google shows for it, and whether that page matches the type you found on page one. A mismatch between the page shown and the type wanted is the cheapest fix available. If your guide appears for a query where page one is full of category pages, the problem is the page type and not the writing.
Three cases
- Your page of the right type already appears: improve depth, evidence and clarity. A new URL adds nothing.
- Your page of the wrong type appears: build the right type and link it from the wrong one where the reader would want it.
- No page of yours appears: build the type the sheet shows, then check after publication.
What to record about the owner page
Write down the address, the last meaningful update, the format of its opening screen and the queries it already attracts. When two pages compete, the owner page is the one whose type matches the sheet. The other page keeps a different job or links to the owner.
Trail running shoes: four queries, four jobs
Example scenario: the editor of a running gear site searches four related queries. The observations in the table are invented to show the method and do not describe real results.
| Query | Invented page one | Reading | Decision |
|---|---|---|---|
| trail running shoes | Store categories, two buying guides, one video | Purchase with a learning side | Category page with a short selection guide block |
| trail vs road running shoes | Explainers, a forum thread | An answer | Short answer-first explainer with a comparison table |
| best trail running shoes for beginners | Ranked lists with tests | A shortlist | Comparison with stated criteria and first-hand notes |
| how to clean trail running shoes | Step lists, a video, related questions | An answer in steps | Numbered steps with photos, video optional |
The same product family produces four different page jobs. A single "trail shoe guide" would answer none of them well. The plan has four rows, and each row has a separate owner and a separate promise.
The shortcut the editor could take
The editor could fold the cleaning steps into the category page to save effort. The invented table argues against it: step lists have their own pages. Folding them in lengthens the category page, delays the path to a product and leaves the how-to query without a dedicated answer.
Pitfalls when reading page one
Judging from one screen
Phones and desktops can show different elements. Read on the device most of your readers use, then note what changes on the other.
Stopping the read after three results
Three results are not the whole mix. Read the full page and the extra elements before you tally anything.
Taking a result type for proof of quality
Category pages filling page one do not mean you have a category worth ranking. Without a matching product range, a guide or comparison may be your only honest route, and the results page may still prefer another type.
Cloning the leading layout
Copying a rival's layout copies its gaps as well. The type comes from the results page; the value comes from your reader's questions.
Splitting one page between two audiences
A page that is half guide and half store tends to satisfy neither group. If both jobs are needed, build two pages and link them where the reader benefits.
Leaving the sheet out of the brief
A table that stays with the editor leaves writers and designers building to their own assumptions. Put it on the first page of the brief.
Before you commit: other inputs
Page one is one input among several. Before you commit, look at the pages you already have: a page near the same query may deserve strengthening instead of a new URL. Look at your Search Console queries to see whether you already earn impressions for the query and which page appears. Read the guide to People Also Ask and SERP features to plan how the exploration features shape your subheadings, and the piece on matching search intent in articles for the planning side.
For the official definitions, see Google's guide to ranking systems (Opens in a new tab), its Search Gallery (Opens in a new tab) and the explanation of what each ranking system does. Google documentation can change, so check the current page before you act.
Frequently Asked Questions
What does reading page one give me that a keyword tool does not?
It shows which page types Google actually presented for the query, which is better evidence than a tool label alone.
What goes into each row of the sheet?
Each result is recorded as a page type, a format and a title promise. Images, videos, related searches and related questions are noted as well.
How do I plan when page one is a mixed bag?
Narrow the query with a word that fixes the intent, choose one main job for the page, and add a short routing note in the introduction. Repeating the search on other days shows whether the mix is stable.
Why might a second page of mine not appear beside the first?
Google's ranking guide describes a site diversity system that generally avoids showing more than two page listings from the same site in top results. Decide which page owns the query before you publish another.
When should I repeat the reading?
Practical suggestion: every three to six months for heavily dated results, and before major edits for the rest.
How does freshness change the reading?
Where page one is full of dated results, keep your date visible and update the substance of the page, not only the date.
Conclusion
Reading page one turns a format decision into an observation you can file. Describe each result by type, format and title promise, name what you see with Google's vocabulary, place the query in one outcome and choose the page's main job. Narrow ambiguous queries, record the date and repeat the reading as results move.
If you want help turning those notes into finished pages, take a look at the content writing service.
Post a comment
Please log in to leave a comment. Login