Choose the format by the job the searcher is trying to finish, not by the keyword. A person typing a query wants to learn something, do something, pick from options or make one decision, and each job fits a different page: a guide, a roundup, a comparison or a single review. If you run a blog or an editorial team, name the job first, then open the first page of results and note which formats appear and which question none of them answers.
The documentation has no statement on format selection. What Google's pages do supply is narrower: readers may search with different words than you use, a page should leave the reader able to reach their goal, and a title should describe the page without exaggeration.
Start from the reader's task
Take the query "wines like Rioja." One reader wants a shortlist for the weekend. Another wants to know what makes a Rioja distinctive so they can judge similar wines. A third has finished a bottle and needs the next one today. The words are identical; the tasks are different, and so are the pages that would serve them.
The SEO Starter Guide (Opens in a new tab) notes that users who know a lot about a topic might use different keywords than beginners. It also says Google's language matching can connect a page to a query even when the page does not use the exact words. Read together, these two points move the weight off the keyword and onto the person: you cannot infer the page you owe a reader from their words alone.
The helpful content page (Opens in a new tab) asks whether someone will leave feeling they have learned enough to reach their goal, and whether they will need to search again for better information. A format that fits the task is the cheapest way to pass both tests. A roundup answering a how-to question, or a long essay answering a comparison question, fails them before the first paragraph is read.
Four reader jobs
Practical suggestion: sort queries into four jobs instead of a long taxonomy. Four is easy to apply during planning and still separates the formats that matter.
| Reader job | What they want to leave with | Wording in the query | Format that fits |
|---|---|---|---|
| Learn | An explanation they can repeat | what is, why, meaning | Explanatory guide |
| Do | A procedure they can follow | how to, set up, steps | Step-by-step guide |
| Pick | A shortlist to choose from | best, wines like, ideas | Roundup list |
| Compare | A verdict between named options | X vs Y, difference | Comparison with a table |
| Decide | A yes or no on one item | worth buying, should I | Single review |
Navigational and news-style queries sit outside this table. A reader typing a site name wants that site, and a reader after the latest edition wants a page that is kept current. Neither is a format puzzle for a content plan.
Applying the jobs to ambiguous queries
A table of jobs stays abstract until a query is run through it. These queries do not settle into one job on first reading; the results page supplies the second reading.
| Query | First reading | What to look for in the results |
|---|---|---|
| dry sherry styles | Pick or learn | List bottles or style definitions? |
| wine tasting night picks | Pick | Seasonal lists or evergreen lists? |
| how to read faster | Do | Step lists, or product pages for apps? |
| X vs Y whisky | Compare | Side-by-side tables, or two separate reviews? |
| new X whisky release | News | Official announcements or retailer pages? |
Reading the first page of results like an editor
A guess about the job becomes a decision only after you look at the results. The same Starter Guide describes the title link as the headline part of a search result and the snippet as text sourced from the page. Those two elements are what a searcher sees, so they are what you read.
Pass one: titles
Copy the top five titles into a note. Mark the words that repeat. If four of five carry a list number, the page type that is already being served is a roundup. If two carry "vs", comparison is in play.
Pass two: formats
Open each page for ten seconds and write one word for what it is: guide, list, comparison, review, store page, glossary entry. Ten seconds is enough to classify. You are not auditing quality yet.
Pass three: the gap
Ask which question a reader would still have after finishing all five. The unanswered question is where your page can contribute. If you find none, you have no gap, and the topic may not be worth a new page.
A dated worksheet
Example scenario: a worksheet for the query "wines like Rioja". The rows are invented to show the layout and do not describe a real results page.
| Position | Title pattern | Format | Gap noted |
|---|---|---|---|
| 1 | Best wines like X | Roundup | No occasion grouping |
| 2 | If you loved X, read these | Roundup | Same titles as position 1 |
| 3 | X read-alikes explained | Guide | Explains traits, no shortlist |
| 4 | Store collection page | Store | Sales only |
| 5 | Reader forum thread | Discussion | Unsorted suggestions |
Write the date at the top of the sheet. A results page changes, and an undated note turns into a claim nobody can check.
What a results page can and cannot tell you
Reading results is observation, not measurement. Be explicit about the two sides.
- It can show which formats are being offered for the query today and where the shared coverage ends.
- It cannot show why a page sits where it does, which format ranks best, or what Google wants. The documentation has no statement that matching the dominant format is required.
If every result is a roundup, you are not obliged to write one. If every result is a roundup and the searcher's job is comparison, a comparison page may serve readers better than the fifth roundup. If the roundups are strong and you have nothing to add, the honest conclusion is to leave the topic alone.
Noise on the page
A results page can also carry ads, answer boxes and image strips. They may cloud the format reading, so the thing to read is the page type behind each result. Practical suggestion: run the same query on a different day and device, and log the pattern that repeats rather than the element that changes. The documentation has no statement on how result page layouts vary, so a single look should never be logged as settled fact.
Format blueprints
Once the job is clear, the outline follows. Each format promises something in its first screen, and the rest of the page must keep that promise.
| Format | What the first screen must give | Body parts | Failure mode |
|---|---|---|---|
| Guide | A short answer and who it is for | Definition, steps, exceptions | Reader wanted options, got a lecture |
| Roundup | The selection rule | Items with the same fields, a rule to choose between them | Items with unequal depth, no rule |
| Comparison | Which option fits which reader | Shared criteria, table, verdict per criterion | Criteria differ per option |
| Review | A one-line verdict | Experience notes, strengths, weaknesses, who should skip it | Praise only, no weakness |
| Landing page | The action | Short copy, one next step | Used for an informational query |
Titles carry the first signal of the format. The Starter Guide says a good title is unique to the page, clear and concise, and the helpful content page asks for a descriptive summary that avoids exaggeration. A title promising "The 12 best" over a page with seven entries breaks that standard and the reader's trust together.
Tables and lists inside other formats
The choice is not always exclusive. A guide can carry a comparison table, and a roundup can give each entry a short review. The test is whether the headline promise still governs the page. A page titled "how to choose" should spend most of its length on choosing, with the table summarising it.
A decision tree you can paste into a brief
Practical suggestion: answer five questions in order and write each answer in the brief.
- What job is the searcher finishing: learn, do, pick, compare or decide?
- Which format dominates the first five results?
- Which question do those five leave unanswered?
- Can you fill that gap with something of your own, such as tasting notes or a tested method?
- In the format you would choose, can the reader finish the job on one page?
| Answers | Decision |
|---|---|
| Dominant format matches the job, no gap | Use the same format and differentiate through evidence and clarity, or skip the topic |
| Dominant format matches the job, gap exists | Use it and give the gap its own section |
| Dominant format does not match the job | Pick the format that fits the job |
| Formats are mixed | Apply the mixed-intent rule |
| You have nothing of your own to add | Hold the topic |
Two runs of the tree
Example scenario: the query "how to start a tasting club". Four of five results are step lists, one is a store page. The job is "do", and the dominant format matches. The gap: none of the lists says what to do when attendance drops. Decision: a step-by-step guide with a section on keeping a club alive, drawing on the site team's own tasting nights.
Example scenario: the query "bourbon or rye". Results are two reviews of single bottles and a forum thread. The job is comparison and no result offers a table. Decision: a comparison with shared criteria such as price band, cocktail use, sweetness and finish, followed by a verdict by drinker type.
Mixed intent: one page or two
Some queries split between jobs. "Red wines" might be a reader wanting a shortlist and another wanting a definition of the style. The results mirror that split. Two routes exist.
- Choose the primary job and answer the secondary one in a section, such as a roundup with a short "how to choose" part.
- Separate the jobs into two pages that link to each other and do not repeat each other.
Route one fits readers who can finish on one page. Route two fits distinct audiences. Do not take route two just to produce more pages: the helpful content page lists producing lots of content on many topics in hopes that some of it might perform well among the warning signs of search engine-first content.
Order of sections on a mixed page
Put the primary job's answer first. The secondary section gets its own heading so a reader with that second need can find it. The order follows what the reader asks first, not what you find more interesting.
Example scenario: a wine and spirits review site
Example scenario: a wine and spirits review site plans pieces around one distillery's range with three query groups. The worksheet that follows is invented for illustration.
Group one asks in what order to taste the range. The results mix a numbered list with a guide to tasting order. The job is pick and learn. The site's tasters have tried every bottling and can say which ones can wait. Format: a numbered tasting order with a one-line note per bottle. Evidence: the tasters' notes.
Group two sets two bottlings from the range against each other. Results are single reviews with no side-by-side view. The gap is a comparison by drinker type. Format: a comparison table with the same criteria on every row (sweetness, smoke, finish, price band) and a verdict for each kind of drinker.
Group three asks whether the newest bottling is worth buying. The searcher is at the decision. Format: one review that says in the first paragraph who will enjoy it and closes with a plain verdict. Weak points are named, and the title promises nothing it cannot back.
The editor links the three pages so a reader can move from discovery to decision. The format came from the job and the gap, with the dominant results as a reference rather than a template.
The same decisions as brief lines
| Query group | Job | Format | Gap | Evidence |
|---|---|---|---|---|
| Tasting order | Pick | Numbered list | No skip guidance | Tasters' notes |
| Two bottlings | Compare | Comparison table | No reader-type verdict | Shared criteria sheet |
| Worth buying | Decide | Single review | Unclear audience | Named weaknesses |
Alternatives the editor rejected
Example scenario: writing down what was not chosen is as useful as the choice. For the comparison group the editor rejected a roundup of "ten best bottlings in the range" because the searcher already knows the two bottlings. For the tasting-order group the editor rejected a long essay on the distillery's history because the searcher wants a sequence. For the review group the editor rejected a bare rating because it would not tell the reader who should skip the bottle.
Writing the choice into the brief
A format decision that stays in someone's head vanishes when the writer starts. Put it in the brief as fields. For the other brief fields see how to write an SEO content brief.
| Brief field | Content |
|---|---|
| Reader job | Compare |
| Decision moment | Torn between two options |
| Format | Comparison table, verdict at the end |
| Gap | No result splits the options by reader type |
| Evidence | The team's own tasting notes |
| Title rule | Names both options, promises nothing extra |
What the format changes in the opening
A comparison opens with which option suits which reader. A how-to opens with the result and the number of steps. A review opens with the verdict. For the mechanics of an opening, see how to write an article introduction, for short answer blocks see writing answer blocks, and for heading levels see heading structure.
Should the article show live results
Live examples carry two risks. Results change, so the example ages and misleads. Naming specific pages also invites judgments about those pages. A dated, invented worksheet avoids both. If a screenshot is unavoidable, caption it with the date and the conditions.
Substitutes for a live example
- A dated pattern sentence, such as "three roundups, one guide and one store page appeared".
- An invented worksheet labelled as an example.
- Your own content map showing which query group got which format.
Checking the choice after publication
A format decision is a hypothesis. Search Console data for the page's queries and reader behaviour can test it. For keyword discovery to feed this process, see how to do keyword research. The documentation has no timeframe for how long a page takes to be assessed, so change one thing at a time and log the date.
| Observation | Possible reading | Next step |
|---|---|---|
| Impressions, few clicks | Title or snippet does not match the job | Revisit the title |
| Clicks, quick exits | Page does not finish the job | Rework the opening and the format |
| No impressions | Too early or the topic is a poor fit | Wait, then reassess coverage |
These are readings, not diagnoses. Without further evidence do not swap the entire format.
A change log that stays useful
Each change gets a dated line: what changed, why, which observation prompted it. Without that line, two months later you cannot tell which edit produced which shift. One line per change is enough.
A one-page note for the writer
If the decision stays in the brief, the writer may never see it. Send a short note with these lines.
- The chosen format and the reason in one sentence.
- The reader's question in the writer's own words.
- The answer the opening must give.
- The words allowed in the title and the promises to avoid.
- The evidence the page must carry and where it comes from.
- Any section that does not belong in this format.
A format record for teams with several writers
When several people write for one blog, each picks a format by habit unless the decisions are recorded. A small record prevents this: query type, chosen format, reason and observation date. It also shows a new editor why earlier choices were made. Practical suggestion: keep it next to the brief template and review the rows every quarter. A row that nobody has reviewed for a long time is a prompt to rerun the three passes, because the results page may have moved on. A row that survived a review with the same reason attached is evidence that the format choice was grounded in the reader's job, and a new writer can rely on it without asking for the whole history again.
| Record field | Why it is needed |
|---|---|
| Query type | Ties similar queries to the same decision |
| Chosen format | Spares the writer from rebuilding the outline |
| Reason | Gives a basis when the decision is challenged |
| Observation date | Shows when a decision has aged |
| Owner | Names who updates the row |
Checklist before drafting
- The query is written in the reader's own words.
- The reader job is named; a mixed query has a primary and a secondary job.
- The first five results are logged on a dated sheet.
- The dominant format and the gap are written down.
- The format decision and its reason are in the brief.
- The title describes the page without exaggeration.
- The opening keeps the format's promise.
- No live result example appears undated.
Mistakes to avoid
Copying the dominant format blindly
The dominant format is a clue. The gap and your evidence decide.
Reading the format off the keyword
The same words come from different jobs. Look at the results before choosing.
Promising one format in the title and delivering another
A comparison title with no table costs the reader's trust.
Presenting observation as a rule
Telling a team "Google wants a roundup here" turns a reading of results into a claim nobody can source.
Ignoring the reader's vocabulary
A team's internal label for a style may not match what readers type. Log queries in reader words.
Frequently Asked Questions
Are the search intent categories defined in Google's documentation?
No. Learn, do, pick, compare and decide are an editorial aid. The documentation has no statement on these categories.
Do I have to look at the results page to choose a format?
It is not required, but it is useful. The results page shows which formats are on offer today. It is an observation, not a rule.
If the top five results are all roundups, must I write a roundup too?
No. Start with the reader's job. If the job is comparison and no result serves it, a different format can fill the gap.
Should a mixed query get one page or two?
One page works when a reader can finish on it: pick the primary job and answer the secondary one in a section. Two pages work when the jobs serve different audiences; link them and keep them distinct.
Should I put live search results in my article?
It is not recommended. Results change and the example ages. Use a dated, invented worksheet labelled as an example instead.
How do I know my format choice was wrong?
The documentation gives no single measure. Practical suggestion: read impressions, clicks, whether readers stay, and the fit between title and content together.
Conclusion
Format selection is a decision about the reader's task, not a matter of taste. Name the job, log the first five results on a dated sheet, find the dominant format and the gap, then write the choice and its evidence into the brief. The reader's own wording and the title standard come from the SEO Starter Guide, and the goal-reached test comes from the helpful content page. Treat the job categories as working labels, not as rules from Google.
This week, pick one planned post, type its query in a clean window and run the three passes. If you want help producing articles in the formats you plan, take a look at our content writing service.
Post a comment
Please log in to leave a comment. Login