How do you find keywords worth targeting?

The keywords worth targeting are the ones your actual customers already use when they're trying to solve the exact problem you solve—not the broad industry terms that sound impressive in a pitch deck. Start by getting specific about who you're serving and what they're searching for at the moment they need help, then validate that specificity with real data before investing in content. Narrow beats broad, especially early.

Start with the problem, not the product category

Most founders make the same mistake: they target the category name for what they've built ('project management software,' 'HR platform') instead of the specific failure mode their customers are trying to escape. A person searching 'how to stop missing client deadlines with a remote team' is further along in buying intent than someone searching 'project management.' The first person is describing a wound; the second might just be doing research.

The fastest way to find these problem-level phrases is to go back to the conversations you've already had with early customers. What exact words did they use to describe the situation before they found you? What did they type into Google the week before they signed up? If you haven't asked, ask now. Email your five most engaged users and ask them: 'Before you found us, what were you searching for?' The answers will be more useful than any keyword tool you can buy.

This approach connects directly to what Paul Graham argues in his essay on doing things that don't scale: the feedback you get from engaging directly with your earliest users is the best you'll ever get. Keyword research is just one more form of listening to those users—translating their language into a targeting strategy.

Use the narrow market principle to find winnable terms

Paul Graham's observation about Facebook starting at Harvard before expanding illustrates something directly applicable to SEO: a deliberately narrow market lets you build real momentum before you compete broadly. The same logic applies to keywords. You cannot rank for 'accounting software' as a new startup. You can rank for 'accounting software for independent dog groomers in the US' if you actually serve that market and write content that is genuinely useful to them.

The question to ask for any candidate keyword is: could we realistically be the most useful page on the internet for someone searching this exact phrase? If the answer requires beating Salesforce, HubSpot, or a site with ten years of domain authority, skip it for now. Find the version of the keyword where the answer is yes. This usually means going two or three levels more specific than feels comfortable.

Niche keywords have a structural advantage beyond winnability: the people searching them convert better. Someone searching 'invoicing tool for freelance translators' is almost certainly a freelance translator who needs invoicing software right now. They are not browsing. They have a problem and money. Specificity is a proxy for intent, and intent is what turns traffic into revenue.

Practically, this means building a list of modifiers that describe your actual customer segments—industry, job title, company size, geography, workflow, technology stack—and combining them systematically with your core problem terms. That matrix of combinations is your first keyword universe to evaluate.

Evaluate keywords on three dimensions, not just volume

Search volume is the metric everyone looks at first and it's often the least useful signal for early-stage founders. A keyword with 200 monthly searches that converts at 8% is worth far more than one with 20,000 searches that bounces 90% of the time because the intent doesn't match what you sell.

The three dimensions that actually matter are: intent match (does this person want what we offer at the moment they're searching?), competitive reality (can we plausibly rank in the top five given our current domain authority and the quality of content we can produce?), and economic value (if this person becomes a customer, is the deal size worth the cost of acquiring them through organic search?). A keyword can fail any one of these tests and be worth dropping regardless of its volume.

For intent, the simplest check is to look at the pages currently ranking for a keyword and ask whether your product belongs in that list. If the SERP is dominated by educational blog posts, the searcher is probably learning, not buying—you can still target it, but your content needs to meet them where they are, not pitch at them. If the SERP shows comparison pages, review sites, or product landing pages, there's transactional intent and that's where you want to be.

For competitive reality, look at the domain ratings and content quality of the top-ranking pages. Tools like Ahrefs or Semrush can show you this numerically, but you can also just read the pages and ask honestly: could we write something better or more specifically useful than what's there? If the ranking content is generic and you have real domain expertise in the niche, that gap is your opportunity.

Build a feedback loop between keyword targeting and customer reality

Keyword research is not a one-time exercise you do before launching a blog. It's a signal-gathering process that should run continuously alongside your product and sales work. As you talk to more customers, you'll hear new language—new ways of describing the problem, new comparison points they're using, new job titles you didn't know were in your market. Each of those is a potential keyword cluster you haven't explored yet.

Set up a simple system: when a sales call or support ticket introduces a phrase you haven't seen before, log it. Once a month, run those phrases through your keyword tool and check whether there's search volume behind them. Sometimes there isn't—the phrase is idiosyncratic to one person—but often you'll find that five of your customers use similar language because it reflects how an entire professional community talks about the problem, and that community is searching for answers.

The other half of the feedback loop is tracking what's actually working. Which pages are bringing in users who convert? Which keywords send traffic that leaves immediately? After six months of publishing, you'll have enough data to double down on the content types and keyword patterns that produce real customers, and stop investing in the ones that produce only pageviews. That's when keyword strategy stops being guesswork and starts being a compounding asset.

“The feedback you get from engaging directly with your earliest users will be the best you ever get.”

— Paul Graham, source

The one thing to do

Write down the exact words your three best customers used to describe their problem before they found you, check whether people are searching those phrases, and publish the most useful page on the internet for each one.

Frequently asked questions

How do I find keywords if I don't have customers yet?

Talk to the people you want as customers before you build content. Reddit threads, industry Slack groups, and competitor review pages on G2 or Capterra are full of the exact language your target users use to describe their problems. Mine those sources first, then validate with a keyword tool.

Should I target high-volume keywords or low-volume niche ones?

For an early-stage startup, low-volume niche keywords almost always win. They're more winnable competitively, they attract buyers rather than browsers, and ranking for ten specific terms builds domain authority that eventually helps you compete for broader ones.

How many keywords should I focus on at once?

Pick three to five keyword clusters that map to real customer segments and go deep on each before expanding. Spreading thin across dozens of terms early on produces weak content that ranks for nothing and teaches you little about what's actually working.

How long before I know if a keyword strategy is working?

Expect six to twelve months before organic search produces meaningful, consistent inbound traffic for most early-stage sites. You can shorten the feedback loop by tracking ranking movement monthly and using paid search to quickly test whether a keyword actually converts before investing heavily in content.

Sources

More playbook answers · Growth Prophet home