AnswerThePublic for content ideas: the four reports, and how to turn an export into a plan

AnswerThePublic turns one seed word into the questions, comparisons and phrases people actually type around it, and that list is the fastest raw material for a content plan. Put in a seed such as "home insurance", and it returns the question forms (what, why, how, when, where, who, which, can, are, will), the preposition forms (for, with, without, near, to, is), the comparison forms (vs, versus, and, or, like) and an alphabetical sweep of the longer phrases. You export the lot, group it into the handful of real questions underneath, and each group becomes one page. The tool gives you the vocabulary. The grouping and the judgement are the work that makes it pay.

What the tool actually returns

Checked on 10 October 2026, answerthepublic.com presents its output in four families, plus a set of related searches. It is worth knowing which family answers which kind of question before you start, because they are not interchangeable.

• Questions. Grouped by the question word. These map almost one to one onto headings and FAQ entries, and they are the highest value part of the export.
• Prepositions. "home insurance for landlords", "home insurance without a survey", "home insurance near me". These reveal segments and constraints you may not be selling to yet.
• Comparisons. "home insurance vs landlord insurance", "buildings and contents". These are decision stage phrases, and they convert better than anything else in the export.
• Alphabeticals. The seed followed by each letter of the alphabet. Noisy, but it surfaces brand names, local terms and spellings you would never guess.
• Related searches. Adjacent seeds. Use these to decide what you research next week rather than what you write today.

Output comes as a visual wheel and as a plain list, and the list exports to CSV. Work from the CSV. The wheel is excellent for showing a client the shape of a market in one screen, and it is poor for the grouping work, because grouping needs sorting and the wheel cannot be sorted. There is a free tier and there are paid plans; its pricing page carries the current terms, so read it there rather than trusting a figure quoted in any article.

The seed decides the quality of everything after it

One seed word, chosen badly, produces three hundred rows of noise. Three rules fix that.

Use the term your buyer uses, not the term you use internally. If your industry says "fleet telematics" and your buyers say "van trackers", seed the buyer's phrase. The tool reflects real typed language back at you, so feeding it internal vocabulary returns the questions of people who already work in your industry.

Seed the problem as well as the product. "Payroll software" returns product questions. "Late payroll" and "payroll mistakes" return the panic that sends somebody looking for payroll software in the first place, and those pages catch the reader earlier.

Run three seeds, not one. The product term, the problem term and the category term. Merge the three exports into one sheet before you group anything. Patterns that appear across all three seeds are the real centre of the topic.

Turning the export into a plan

The export is a list of phrasings. A content plan is a list of questions. Collapsing one into the other is the step most people skip, and skipping it is how a site ends up with five thin pages competing for the same result.

1. Dump all seeds into one sheet. One column for the phrase, one for the seed it came from.
2. Sort alphabetically and read the whole thing once. You are looking for repeated nouns, not for individual winners.
3. Tag each row with the question underneath it. "how much is home insurance", "home insurance cost", "average home insurance price" and "home insurance monthly cost" are four phrasings of one question. Tag all four "cost".
4. Count the tags. A tag with fifteen phrasings behind it is a page. A tag with one phrasing behind it is a paragraph inside another page.
5. Check intent per tag. Open the results page for the main phrasing and look at what Google is already rewarding. If every result is a calculator, your article will not win that query, and the answer is to build the calculator or to pick a different tag.
6. Commit one page per tag and list every phrasing in that tag inside the page, as headings, as FAQ questions, or as sentences in the body.

Report family	What it tells you	Where it lands on the page
Questions	What the reader does not know yet	H2 and H3 headings, FAQ block
Prepositions	Segments, constraints, locations	Dedicated sections, or separate pages where demand is large
Comparisons	The shortlist the reader is choosing from	A comparison table, or its own page
Alphabeticals	Vocabulary, brands, local terms	Body copy and internal links
Related searches	The next topic to research	Next month's plan

Pair it with a volume source and with your own data

AnswerThePublic is a phrasing engine. It is at its strongest when two other inputs sit next to it. The first is a volume source, so you can tell a question asked by thousands from a question asked by four people. The second, and the better of the two, is your own query data in Search Console: the questions you already appear for and lose, which are the cheapest wins on any site because the demand and the partial relevance already exist. Our five step keyword research method sets out that order, and our page writing workflow picks it up once the question is chosen.

Use the questions as headings almost verbatim. A heading that matches the typed question gets lifted into featured snippets and into assistant answers far more often than a cleverly worded alternative, and Google's own guidance on titles and headings rewards descriptive text over marketing language.

This page covers using AnswerThePublic to generate and organise content ideas. For the full research method including volume qualification and difficulty, see the keyword research method. For getting video questions, see YouTube SEO keywords.

Where people get stuck on AnswerThePublic for content ideas

Is AnswerThePublic free?

It offers free searches and paid plans. The number of free searches and the plan contents change, so check the pricing page on the day you decide rather than relying on a figure in an article.

Where does the question data come from?

From search autocomplete patterns, which is why the phrasings read like real typing rather than like marketing copy. That is also why the tool shows you phrasings rather than search volumes, and why a volume source belongs next to it.

How many pages should one seed produce?

Usually between three and eight. If you are getting twenty from a single seed, the tags are too narrow and you are about to build pages that compete with each other.

Should every question become an FAQ?

No. Promote a question to a heading when the answer needs several paragraphs, and keep it in the FAQ block when two or three sentences settle it. Questions with no real answer belong in neither.

Does it work for a local business?

Yes, and the preposition report is the part to read first. "near me", "in [city]" and "for [customer type]" phrasings tell you which service and location pages are worth building.

How often should I rerun a seed?

Quarterly for a stable category, monthly where the category moves, so you are always planning from current language.

From the question export to the question asked aloud

The phrases AnswerThePublic hands you are the same ones buyers now put to an assistant, only longer and more specific. So test the export against one: pick the five questions where you most want to be the answer and ask each in ChatGPT, Perplexity and Gemini without naming your company. Read who is named and which pages are cited.

A content plan built on the export tells you what to write; the assistant's reply tells you whether what you already wrote is the page being used. AI Knows Us collects those names and sources for a business so the same questions can be compared month to month.