Long-tail keywords convert 10 to 50x better than broad head terms. That single number should reorganize your entire keyword strategy, because most founders spend their first months chasing the exact terms that will never send them a customer.
When we grew promptbuilder.cc to 50K clicks, the lesson that mattered most was intent, not volume. This post turns that lesson into a repeatable keyword research workflow you can run this week, with free tools.
The day-one mistake: chasing head terms
Every founder opens a keyword tool, types their category's biggest term, sees a huge monthly volume, and decides that's the target. "Project management software." "AI writing tool." "CRM."
Those head terms are the worst possible targets for a new domain. Three reasons:
- You can't win them. The first page is owned by incumbents with thousands of backlinks and years of topical authority. A fresh domain does not outrank them by writing a better page.
- Even if you ranked, the traffic is mushy. Head terms mix students, researchers, competitors, and job seekers in with actual buyers. Volume is not demand.
- They burn your only scarce resource. You get a handful of strong pages per month as a solo founder. Spending them on unwinnable terms is how sites die at month four with zero traffic.
The head term is where your market ends up, not where you start.
The filter that works in 2026: intent first, volume second
Here's the filter, in priority order:
- Intent first. Would the person typing this query pay for what you sell? If the answer is no, the keyword is out, regardless of volume.
- Difficulty under 30. Use any tool's difficulty score as a rough gate. Under 30 means a new domain has a real shot within months, not years.
- Volume last. 30 to 200 searches a month is fine. Ten queries at 50 searches each beat one query at 5,000 if the ten are buyer queries.
You don't need 100,000 visitors. You need 100 with intent. A hundred visitors a month who are actively trying to solve the exact problem you solve will produce more signups than ten thousand drive-by readers.
Run the "credit card test" on every keyword before it enters your list. Ask: is the searcher close enough to a purchase that a clear CTA on this page would feel helpful, not pushy? "best invoicing tool for freelancers with EU VAT" passes. "what is an invoice" fails.
One page, one buyer: modifiers and jobs-to-be-done
Low-competition, high-intent keywords almost always follow one of two shapes.
Shape 1: head term plus modifier. Take the head term you can't win and append a qualifier that narrows it to a specific buyer: "for freelancers," "for agencies," "free," "open source," "with API," "[competitor] alternative." Each modifier splits off a smaller, winnable query with sharper intent than the parent term.
Shape 2: job-to-be-done queries. These don't mention your category at all. They describe the task: "how to send invoices that chase themselves," "track keyword rankings without a tool," "turn a spreadsheet into an API." The searcher wants an outcome, and your product is the shortcut.
The mapping rule is strict: one page targets one buyer with one job. Don't write "Invoicing for Freelancers, Agencies, and Small Teams." Write three pages. Each page can then speak directly to that buyer's situation and make one specific promise, which is exactly what both Google and the reader reward.
The free workflow: four sources, zero subscriptions
You do not need Ahrefs or Semrush to build your first keyword list. They help later, for validating difficulty at scale, but every input below is free.
1. Autocomplete mining. Type your head term into Google followed by each letter of the alphabet: "invoicing for a...", "invoicing for b...". Do the same with "vs," "alternative," "without," and "how to." Every suggestion is a real query real people type. Repeat in ChatGPT or Perplexity to catch how people phrase the problem conversationally.
2. SERP scraping by hand. Search your candidate keyword and read the results page itself. "People also ask" boxes are a keyword list. "Related searches" at the bottom is a keyword list. And check who ranks: if the first page is Reddit threads, thin listicles, or forum posts, that's a green light. If it's all DR 80+ product pages, move on.
3. Search Console queries. Once you have any traffic at all, open the Performance report and sort queries by impressions. You're looking for terms where you get impressions but rank on page two, and for query phrasings you never thought to target. This is the only keyword source that's literally about your own site.
4. Competitor scan. Open the blog and sitemap of two or three competitors one size up from you. Their page titles reveal their keyword map. Note which modifiers and jobs they cover, then target the gaps: the modifiers they skipped, the buyer segments they ignore, the "vs" pages they were too polite to write.
Build the keyword map: query, page, intent, CTA
Raw keyword lists rot in spreadsheets. Turn yours into a keyword map, where every row is a page you'll actually ship:
| Query | Page | Intent | CTA |
|---|---|---|---|
| invoicing tool for freelance designers | /for/freelance-designers | Commercial | Start free trial |
| [competitor] alternative for small teams | /alternatives/[competitor] | Transactional | Compare plans |
| how to automate late payment reminders | /blog/automate-payment-reminders | Informational | Free guide |
| invoice template with EU VAT | /templates/eu-vat-invoice | Transactional | Use template |
Four columns, four decisions. The query column comes from the workflow above. The page column forces you to commit to a URL. The intent column tells you how hard the CTA can sell. The CTA column makes sure every page has a next step, so traffic becomes signups instead of bounces.
Aim for 20 to 30 rows before you write anything. That's a full quarter of weekly content, prioritized before you've spent a single writing hour.
Worked example: pick a niche, run the system
Say you're building an AI meeting-notes tool. The head term "AI note taker" is a bloodbath, so you run the workflow instead:
- Autocomplete: "ai meeting notes for sales calls," "ai meeting notes without bot joining," "meeting notes for therapists."
- SERP check: "meeting notes for therapists" returns two Reddit threads and a 2019 blog post on page one. Difficulty is effectively nil, and the searcher has a compliance problem generic tools ignore. Green light.
- Map it: query "HIPAA compliant AI meeting notes for therapists," page /for/therapists, intent commercial, CTA "Start free trial."
- Ship the page, answer the compliance question in the first paragraph, and repeat next week with "without bot joining."
Fifty searches a month, near-zero competition, and every single visitor is a buyer with a problem your product solves. Twenty pages like that outperform one lucky head-term ranking, and they're winnable on a DR 5 domain.
That's the whole system: filter by intent, narrow to one buyer per page, source queries for free, and map every keyword to a page and a CTA before you write.
Want the full playbook this workflow comes from, plus a new breakdown every week? Grab the free guide, and follow along on X.