Standard keyword research starts with search volume. On 2026-09-18 our Search Console
record for the previous 30 days held 11 impressions and 1 click. There is no volume data
to mine in that, and most new sites sit in the same position. This is the method we use
instead, built from our own roadmap file, and every number in it is ours.
What our own numbers actually show
Eleven impressions and one click is not a dataset, it is a hint. It tells us one person
in Turkey saw the site in search and clicked. It cannot tell us which keyword deserves
more content, because eleven impressions spread over any number of queries resolves to
noise per query.
The distribution layer explains the emptiness. On 2026-09-08 only the homepage was
indexed, out of a sitemap with 1495 URLs, and Google had last downloaded that sitemap on
2026-07-03 before we resubmitted it. Keyword demand cannot be measured on pages a search
engine has not indexed. Any keyword research done at this stage is planning, not
measurement, and it has to be honest about that.
Start from what you ship, not from volumes
Volume tools estimate demand for words. A new site inverts the problem: you own assets
and need the words that fit them. Our roadmap file, version 1, updated 2026-06-25,
defines 11 clusters, and each one is anchored on tools that already exist in the code:
1. Six clusters at priority 1: JSON ecosystem, image optimization, PDF tools, CSS and
HTML design generators, encoding and security, and the regex library.
2. Three clusters at priority 2: text productivity, SEO tools, and AI productivity.
3. Two clusters at priority 3: calculators and converters, and web and network tools.
Each cluster lists its tool keys, its core keywords, and its supporting pages. The regex
cluster, for example, carries 18 core keywords, including regex tester, regex cheat
sheet, email regex, and iban regex, and names regex-cheat-sheet as a supporting page.
That is how a keyword becomes a planned article: it inherits priority from the tool
cluster it belongs to, because the tool is the reason the page can exist.
Expand with intent modifiers
With no volume data, expansion comes from grammar. The roadmap defines 21 intent types:
tool, how to, best, free, online, alternatives, versus, error, cheat sheet, examples,
template, generator, validator, converter, decoder and encoder, viewer, analyzer,
optimizer, checker, calculator, and reference.
Each intent turns a seed into a distinct page promise. The seed regex produces regex
tester under tool, how to test regex under how to, regex cheat sheet under cheat sheet,
and regex101 alternatives under alternatives. One seed, four intents, four pages with
different jobs. Our keyword research tool automates exactly this: it takes a seed term,
applies common modifiers, lets you select the results, and exports them to CSV, with
volume columns you fill in yourself.
Map keywords to page types before writing
The roadmap names four pillars: tools, reference, guides, and templates and examples.
Every keyword gets exactly one pillar before anyone writes anything:
1. Tool intent, generator, converter, validator, checker, calculator: it lands on a tool
page or waits in the backlog. Our backlog holds 45 named tools, from uuid generator
to instagram caption generator.
2. Error intent: it lands on a reference page under the matching collection. The roadmap
defines 14 collections, covering HTTP statuses, MIME types, file extensions, CSS
properties, SQL keywords, and more.
3. Cheat sheet, examples, template: it lands on a reference or template page.
4. How to, best, alternatives, versus: it lands on a guide, like this one.
The mapping kills the worst new-site failure: writing guide number 300 about a topic
where the site owns no tool, no reference, and nothing original to say.
Validate with your own numbers only
Third-party volume estimates are someone else's model, and acting on them at our size
means spending weeks of writing on unvalidated guesses. We set the validation rule
explicitly:
1. Build the page from the roadmap, publish it, and get it indexed.
2. Wait one full 30-day window, then read Search Console for that page.
3. A page with impressions joins the working set, and its queries seed new modifier
rounds.
4. A page with zero impressions after two windows gets a title and content check before
anything else, because with most pages unindexed the first suspect is distribution,
not quality.
The roadmap carries the same honesty. Its phase targets, 2,000 to 5,000 indexable pages
first, then 10,000 or more, are capacity goals for the page factories we control, and
its long-term 100,000 keyword coverage figure is an ambition, not a forecast. Nothing in
it is volume-validated yet, and it says so by dating itself.
Checklist before you research keywords on a new site
- You listed your existing tools and pages before opening any volume tool.
- Every keyword you plan maps to one page type from a fixed pillar list.
- Modifier expansion covered error and alternative intents, not only how-to.
- Volume columns in your sheet are marked as estimates or left empty, not invented.
- You reread your 30-day console data before copying a keyword list from any article.
- Your roadmap file carries a date, and you know what changed since it was written.
Keyword research with no data is not guesswork if you anchor it on what you can build
and validate it only against numbers you generate yourself. Eleven impressions was
enough to prove the loop runs. It was never enough to pick a keyword, and any guide that
implies otherwise is written for a site that already ranks.
What does your own 30-day query data show? Expand a seed from your strongest tool in the
[keyword research tool](https://webrecast.com/en/keyword-research) and check the ideas
against your pillars before you write a word.
