"I can't write." "I don't have any ideas." Those two sentences come up in almost every training session I have run since I started working in search, in 2012. They are sincere, and they rest on a stubborn misunderstanding: people assume content comes from inspiration, when it comes from a list — the questions your customers already ask you, on the phone, by email, over the counter or in the Google search bar.
For years I demonstrated the method on a deliberately dry example: the vehicle number plate. No glamour, no storytelling, a subject most people assume is exhausted in three lines. In practice it carries dozens of distinct questions, with very different intents, and a few better left alone. Here is the sequence: explore, sort, then turn the list into pages rather than one catch-all article.
Inspiration is not a method, collection is
One thing worth knowing before you start: the pool of questions does not dry up. In October 2019, Google stated on its official blog that, of the billions of searches it handles daily, 15% of queries are phrasings it has never seen before. The figure is old and covers the whole engine, not your sector, but the order of magnitude says what matters: people rephrase constantly, and nobody has ever "covered everything" on a subject.
Google's guidance on helpful content points the same way. It asks one question before you publish: does the page offer original information, reporting, research or analysis, and does it rest on first-hand experience? Starting from what your customers actually ask is what lets you answer yes. You do not need imagination. You need a notebook.
Where to find questions people really ask
I rank my sources by decreasing reliability: the first are direct observation, the last are estimates.
- Your own conversations. For three weeks, write down every question asked on the phone, in meetings, at the counter or in reply to a quote. The richest source, and the least used. A plate maker who hears "can I fit it myself?" every week has a page, not a footnote.
- Search Console. The Performance report lists queries that genuinely showed your site. Mind one documented limit: to protect user privacy, Google leaves rare, anonymised queries out of the table while still counting them in the chart totals. The most specific questions are the ones you cannot see.
- Your site's internal search. If it is enabled and tracked, it gives the exact wording of people already on your site.
- Search engine suggestions. Autocomplete, "People also ask" blocks, related searches at the foot of the page. Free, and worth collecting by hand.
- Communities. Trade forums, help groups, reviews, questions on marketplace product pages. That is where you read the real vocabulary, which is never the vocabulary of your brochures.
- Keyword tools, last. They organise and rank. They invent nothing the earlier sources have not given you.
A displayed volume is not a measurement
This is the point I insist on most. Search volumes shown by tools are not counts. They are estimates. Google documents this for its own keyword planning tool: the monthly volume is a rounded average calculated over twelve months, it groups a term with its close variants, and it depends on the locations and networks selected. Third-party tools work from panels and models they do not publish.
So never rank an editorial plan to the nearest number. Think in orders of magnitude — rare, regular, frequent — and cross that with a test of accessibility. Who already holds the top positions, with what kind of site, and do you have something more precise to say?
| What you observe | How to read it | Usual decision |
|---|---|---|
| High estimated volume, results held by national players | Generic query, blurred intent, high cost of entry | Treat as a main page, without expecting a quick result |
| Low estimated volume, very specific question | Wording close to the way your customers speak | Dedicated page: this is often where conversion happens |
| Volume displayed as zero, but the question comes up weekly | The tool is blind to rare queries | Write it anyway, trusting what you hear on the ground |
| High estimated volume, intent unrelated to your offer | Traffic that will never become a customer | Set it aside |
Sort by intent before you write a line
Once the list exists, I file every question into one of four families. On the number plate example: understanding ("what size is a number plate?"), comparing ("acrylic or pressed metal?"), doing it yourself ("how do I fit a plate on a motorbike?"), and buying or getting in touch ("number plate fitting near me"). Each family calls for a different outline, a different level of detail and a different call to action. The last one depends more on your local search presence than on the article itself.
One page cannot serve four intents seriously. The page explaining the rules is not the page selling the fitting service, and merging them produces a long text nobody reads to the end.
The questions best left alone
Exploring this keyword, I kept running into one question: "how to hide your number plate". I once saw a comparison site tackle it head on, listing tricks for making a plate unreadable before concluding, at the foot of the article, that you should not do any of it. Capture the traffic first, cover yourself afterwards.
The editorial decision here is straightforward, because the penalties are real. Take the French rules, since that is where the example came from: the official service-public.gouv.fr page on number plates, checked on 1 January 2026, sets a fine of up to €750 for a missing or non-compliant plate, and cloning another vehicle's plate is a criminal offence carrying up to seven years' imprisonment and a €30,000 fine. Those are French figures and they transfer to no other country — but the editorial point does. A business in that trade has every reason to explain the framework and none at all to publish a how-to. The traffic gained is paid for in credibility lost.
Check the rules before you publish
On any regulated subject, read the rule that applies in your own market rather than copying a competitor's article. Who may manufacture a plate, which models are approved, what must be marked on it: all of that differs by country, and all of it changes. Find the page published by the relevant authority, note the date you checked it, and write from that. The content that ages worst is the content that copied something already stale.
Structure it as pages, not as a catch-all article
The classic mistake is to pile every question you collected into one 3,000-word guide. It ranks poorly, because it matches no query precisely, and it converts poorly, because a reader who came with one question has to wade through everything else.
I prefer a simple structure: one reference page per offer, targeting the main query and explaining the service; satellite pages, one per question or small cluster of close questions; and internal links tying the satellites back to the offer page. On our example: a "plate manufacturing and fitting" page, then separate pages on legal dimensions, on fitting a plate to a two-wheeler, on classic vehicles, and on what to do after a plate theft. Each answers one question, each can rank alone, each points back to the offer. That is the principle behind our SEO content work and behind how we approach organic search generally.
One warning: two phrasings of the same question share one page. "How much does a number plate cost" and "number plate price" are the same intent.
Measure, then prune
A list of questions is not a fixed plan. Six months after publication, open Search Console and look, page by page, at the queries captured, the impressions and the average position. You will find two useful things: pages ranking on questions you had not anticipated, which you can extend; and pages capturing nothing, usually better merged than left to sleep.
To follow that without spending your days on it, a Looker Studio (formerly Data Studio) dashboard plugged into Search Console is enough for a small or mid-sized business: one view by page, one by query, one comparison with the previous quarter. That is the monitoring we set up in a traffic management engagement, and the same logic applies when you scope an audit of what you already have. Exploring a keyword is not something you do once. It is a loop, and the loop is what makes it hold up.
Common questions
How do I find the questions my customers actually ask?
The best sources are internal: write down the questions you receive by phone, by email and in meetings, and read your site's internal search results. Then add the Performance report in Search Console, Google's own suggestions, and forums or reviews in your sector. Keyword tools come last — they rank a list, they do not create one.
Can I trust the search volumes shown by SEO tools?
Not as a measurement. Google states that the monthly volume in its keyword planning tool is a rounded twelve-month average that groups close variants and depends on the locations and networks selected. Third-party tools rely on undisclosed panels and models. Treat the figures as orders of magnitude, never as a forecast.
One page per question, or a single complete guide?
One page per search intent, not per phrasing. Two questions with the same answer and the same next step belong on one page; understanding and buying belong apart. The single guide of several thousand words tends to rank poorly, because it matches no query exactly.
Why do some queries never appear in Search Console?
To protect user privacy, Google leaves queries searched only a very small number of times out of the table, while still counting them in the chart totals. That is why the rows do not always add up. Very specific questions, often the most qualified, are the ones that stay invisible.
Should I cover every question a tool surfaces?
No. Some high-volume questions are irrelevant to your business and will never produce a customer. Others carry a problematic intent, such as getting around a legal obligation: explain the applicable framework rather than publishing a how-to. Deliberate triage beats poorly targeted traffic.
How long before this kind of work shows anything?
There is no fixed answer, and anyone offering one is guessing. What you control is the review cycle: publish, then read Search Console page by page a few months later.