Seed Keywords: Where the List Actually Starts
Seed keywords come from competitor sitemaps, sales calls and support tickets, not a keyword tool. Here's where they actually originate.
Seed keywords come from four places, and none of them is a keyword tool: your competitors’ sitemaps, your sales call transcripts, your support tickets, and your own category or navigation pages. A keyword tool takes a seed and multiplies it into hundreds of variations — it does not originate the seed itself. If you open Ahrefs or Semrush with a blank cursor, you’re skipping the one step that actually determines whether the rest of the list is any good.
Most keyword research guides start at step two. They assume you already have a seed list and jump straight to expansion, clustering, and volume filtering. This one covers step one: where the seeds genuinely come from, in what order, and how many you need before a tool is even useful.
What Is a Seed Keyword, Exactly?
A seed keyword is a short, high-level term that names a topic your business actually competes on — usually one to three words, with no modifier attached. “Invoicing software” is a seed keyword. “Best invoicing software for freelancers in India under ₹2,000/month” is not; that’s a long-tail keyword an expansion tool generates from a seed. Seeds don’t need search volume attached yet. Their job is to define the boundary of a topic, not to rank on their own.
A useful test: if you can’t picture ten different pages your site could plausibly own under that term, it’s too narrow to be a seed — it’s already a long-tail keyword you found early.
Where Do Seed Keywords Actually Come From?
Four sources cover almost every seed list we’ve built for clients at PalV’s DM. None require a paid tool.
1. Competitor Sitemaps and Navigation
Pull the XML sitemap of three to five direct competitors (append /sitemap.xml or /sitemap_index.xml to their root domain, or check robots.txt for the actual path). Scan the URL slugs and page titles for topic words — not exact-match keywords, just the nouns they’ve organized their site around. Then open their main nav and blog category list separately; nav labels tend to be broader (“Pricing,” “Integrations,” “Templates”) while blog categories are closer to seed keywords (“expense tracking,” “payroll compliance”).
2. Sales Call Language
Pull five to ten recent discovery-call transcripts or recordings — most CRMs and tools like Gong, Fireflies, or even a basic Zoom transcript export will have these. Search for the exact phrases prospects use to describe their problem before they’ve heard your product’s name for it. A CRM company might internally say “customer relationship management,” while prospects on calls say “keep track of my leads” or “stop losing customer info in spreadsheets.” Those prospect phrases are seeds; your internal category name usually isn’t one people search.
3. Support Tickets and Chat Logs
Export three to six months of support tickets or live chat transcripts and pull out recurring nouns and problem phrases, not just bug reports. Customers describe what they were trying to do when something broke, and that “trying to do” language is often closer to real search behavior than anything in your product copy.
4. Your Own Category and Product Pages
List every category, subcategory, and product-line page in your own site navigation, without filtering for whether you think people search for them. This gives you a seed list anchored to what you actually sell, which keeps the eventual expanded list from drifting into topics you can’t serve.
Where seeds come from, ranked by effort
- Your own nav/category pages: zero cost, ready in 10-15 minutes, but limited to what you already sell.
- Competitor sitemaps: 30-45 minutes per competitor, reveals topics you haven’t thought to own yet.
- Support tickets/chat logs: 1-2 hours to scan 3-6 months of history, surfaces real customer phrasing.
- Sales call transcripts: 1-2 hours for 5-10 calls, the highest-intent language of the four because it’s pre-purchase, not post-purchase.
How Many Seed Keywords Do You Need Before Expanding?
Fifteen to thirty is a workable range for a single site or a single service line. Fewer than ten and your expanded list will be thin and repetitive; more than forty and you’re probably mixing seeds with long-tail terms you found early, which just clutters the expansion step. For a multi-service business — like an agency offering SEO, PPC, and web development separately — build a separate seed list of ten to fifteen per service line rather than one combined list of eighty. It keeps clustering cleaner later.
How Do You Pull Seed Keywords From a Competitor’s Sitemap Without a Paid Tool?
- Go to
competitor.com/sitemap.xml. If that 404s, checkcompetitor.com/robots.txt— most sites list their sitemap path there. - Open the sitemap (or sitemap index, which links to sub-sitemaps like
sitemap-posts.xml) directly in a browser; it renders as a clickable XML list. - Copy URL slugs into a spreadsheet — one column per competitor.
- Strip the boilerplate (dates, IDs, pagination) and keep the topic words:
/blog/best-payroll-software-2026/becomes “payroll software.” - Repeat for 3-5 competitors, then dedupe the list in a spreadsheet using a simple “remove duplicates” pass.
Screaming Frog’s free tier (up to 500 URLs) will crawl and export a sitemap faster than doing it by hand if a competitor’s site has more than a couple hundred pages.
What’s the Difference Between a Seed Keyword and a Head Term?
They overlap but aren’t identical. A head term is defined by search volume — it’s simply the highest-volume version of a topic (“shoes” versus “running shoes for flat feet”). A seed keyword is defined by function — it’s the input you feed into an expansion tool, regardless of its own volume. Every head term can work as a seed keyword, but plenty of good seed keywords have low or unknown volume, especially in B2B or niche categories where the head term itself only pulls a few hundred searches a month.
How Do Seed Keywords Turn Into a Full Keyword List?
Once you have 15-30 seeds from the four sources above, run each one through an expansion tool: Ahrefs’ Keyword Explorer, Semrush’s Keyword Magic Tool, or the free Google Keyword Planner (it still returns volume bands even without an active ad campaign, though the ranges are wide). Each seed typically returns 200-2,000+ related terms depending on how broad it is. From there you filter by relevance, group by search intent, and cluster by SERP overlap — but none of that works if the seed feeding it was guessed rather than sourced.
| Source | What it gives you | Typical seed yield |
|---|---|---|
| Own nav/category pages | Seeds anchored to what you sell | 5-15 |
| Competitor sitemaps (3-5 sites) | Topics competitors already validated | 10-25 |
| Support tickets/chat logs | Post-purchase customer phrasing | 5-10 |
| Sales call transcripts | Pre-purchase problem language | 5-10 |
What Mistakes Kill a Seed Keyword List Before It Starts?
Three come up constantly in client audits. First, using internal product names as seeds instead of the plain-language problem customers describe — “workflow automation platform” instead of “stop doing this by hand.” Second, pulling seeds only from a keyword tool’s own “related searches” box, which just reflects what’s already ranking and misses gaps competitors haven’t filled. Third, treating a single high-volume seed as good enough and skipping the other three sources, which produces a list that’s technically searched but doesn’t match how your specific buyers talk.
This is exactly the groundwork we do before touching an expansion tool in our full keyword research process — seed sourcing is step one of six, and skipping it is the most common reason a finished keyword list underperforms.
If you’re building a list for a product or feature that’s genuinely new to the market — where competitor sitemaps don’t exist yet because no direct competitor does either — the sourcing logic changes. See our guide to keyword research for a brand-new product category for how to pull seeds from adjacent categories instead.
FAQ
Do seed keywords need to have search volume?
No. A seed keyword’s job is to define a topic boundary for expansion, not to rank on its own. Many good seeds — especially in B2B or niche markets — show low or zero volume in tools, but still expand into dozens of long-tail terms that do get searched.
How many seed keywords should I start with?
Fifteen to thirty per site or service line is workable. Fewer than ten produces a thin expanded list; more than forty usually means you’ve mixed in long-tail terms that aren’t really seeds anymore.
Can I use my own brand name as a seed keyword?
Only for branded-search tracking, not for topic expansion. Your brand name will return searches from people who already know you, not new prospects — it doesn’t help you find the problem-language your target audience uses before they’ve heard of you.
What’s the fastest free way to find competitor seed keywords?
Pull their XML sitemap directly (competitor.com/sitemap.xml or check robots.txt for the path) and scan URL slugs and blog category names. It takes 30-45 minutes per competitor and needs no paid tool.
Should seed keywords come from one source or several?
Several. Relying on only one source — usually just a keyword tool’s suggestions — tends to miss either the topics competitors have validated or the exact phrasing customers use, and often both.
Ready to turn a real seed list into a full content plan? Our one-time keyword research service starts exactly where this article does — sourcing seeds from your competitors, your sales calls, and your support history, not just a tool’s suggestion box.
Written by Palash, founder of PalV’s DM. 5+ years in SEO, 1,000+ articles published.
