The fastest way to waste a content budget is to write for the words you assume people search, rather than the words they actually type. The good news is that real customer language is not hidden. It is sitting in five places most businesses already have access to, and four of them are free. This guide walks through each source in turn, what a result actually looks like, and how to turn raw phrases into something you can act on.
Why "the terms your customers use" is different from "keywords"
A keyword is an abstraction. A search term is the literal string a person entered. The gap between them is where most opportunity lives. Someone shopping for accounting software might search "cheap bookkeeping app for sole traders" rather than the tidy head term "accounting software" a marketer would have guessed. That messy, specific phrasing tells you the budget, the buyer type, and the job to be done all at once.
So as you work through the sources below, capture the exact wording. Note whether each term reads like someone researching, comparing, or ready to buy — that is its search intent, and it decides what kind of page should answer it. A "what is" query wants an explainer; a "best X for Y" query wants a comparison; a "buy X near me" query wants a product or location page.
1. Keyword research tools (for scale and volume)
Dedicated tools are the right starting point when you need breadth: thousands of variations on a theme, each annotated with rough monthly search volume and a difficulty estimate. You feed in a seed term — a short, central phrase — and the tool expands it.
How to use one
- Pick three or four seeds that describe your core offer in plain language. For a dog-walking business that might be dog walker, dog walking rates, pet sitting, and dog daycare.
- Run each seed through the tool's keyword explorer. A typical result is a table of related queries — for dog walker you might see dog walker near me, how much does a dog walker cost, dog walker insurance, and 15 minute dog walk — each with an estimated volume and a 0-to-100 difficulty score.
- Filter aggressively. Sort by word count to isolate long-tail phrases, exclude branded competitor names, or keep only questions to find content angles.
- Use the difficulty column to triage, not to decide. Treat it as a calibrated guess — every vendor models keyword difficulty differently, so the relative ordering matters more than the absolute number.
Platforms differ mainly in data depth and price. A full suite like the one in our Semrush review gives you competitor terms and intent tagging alongside raw volume, while a lighter, cheaper option such as the bundle covered in our Mangools review is often plenty for a single site. See the best keyword research tools roundup if you are still choosing.
Do mine the competitor-terms reports — they hand you a validated list someone else paid to discover. Don't chase the highest-volume head terms first; they are usually the hardest to rank for and the least specific about what the searcher wants.
2. Google Search Console (the terms you already win and almost win)
If your site has any traffic, this is the single most valuable source you own, because it reports the real queries Google showed your pages for — not estimates. It is free for any verified property.
How to read the Performance report
- Open Performance > Search results and enable all four metrics: clicks, impressions, average CTR, and average position.
- Switch to the Queries tab. You now have a list of the exact phrases that triggered your pages, with how often each appeared and how often it was clicked.
- Hunt for "almost there" terms: sort by impressions, then look for queries sitting at position 8 to 20 with low click-through. These are phrases Google already associates with you but where you rank just below the fold — often the cheapest wins available, because a modest improvement to an existing page can lift you onto page one.
- Find high-impression, zero-or-low-CTR queries. They reveal mismatches: people are seeing you for a term but your title or page does not answer it, which usually signals a missing or mistargeted page.
Do filter by a single page (use the Pages tab, click a URL, then return to Queries) to see exactly which terms each article earns. Don't ignore the long tail of one- and two-impression queries — collectively they describe how customers actually phrase things, in their own words.
3. Search-terms reports in Google Ads and Amazon
Advertising platforms expose the most commercially honest data of all: the terms people searched immediately before spending or nearly spending money. You usually need an active campaign to populate them, but the insight transfers straight to organic and product work.
Google Ads search terms report
Found under a campaign's Insights and reports > Search terms, this shows the actual queries that triggered your ads — distinct from the keywords you bid on. You might bid on the phrase running shoes and discover the report is full of running shoes for flat feet, waterproof trail running shoes, and are Hokas good for running. The first two are product and content gold; the third is a research query you can answer with a guide. Conversion data sits beside each term, so you can see which phrasing actually pays.
Amazon search terms
For sellers, Amazon's advertising console offers an equivalent Search Term Report for sponsored campaigns, plus Brand Analytics (for Brand Registry sellers) showing top search terms by frequency on Amazon itself. Because Amazon is a buying engine, these terms skew transactional and product-attribute-heavy — stainless steel water bottle 1 litre rather than hydration tips.
Do harvest converting search terms as exact-match keywords and as on-page copy. Don't treat the ad report as a pure organic list — paid intent runs hotter, so a term that converts in ads may need a different page to satisfy an organic searcher.
4. Autocomplete and People Also Ask (free, fast, qualitative)
These cost nothing and need no account — they are Google quietly telling you what real people type and ask next.
- Autocomplete: start typing a seed into Google and watch the suggestions. They are drawn from genuine high-volume queries. Append a letter to branch the tree: "dog walker a…" surfaces dog walker app, dog walker agreement; "dog walker c…" surfaces dog walker cost, dog walker contract. Prefixing works too — "why is dog walker…" or "cheap dog walker…".
- People Also Ask: the expandable question boxes in the results. Each click loads more, so one query can yield a dozen real customer questions — perfect raw material for FAQ sections and subheadings.
- Related searches: the cluster at the very bottom of the results page, useful for spotting adjacent topics you would not have brainstormed.
Do use these to capture natural phrasing and question formats that volume tools sometimes miss. Don't rely on them alone for prioritization — they tell you a term exists, not how much demand it carries, so validate the promising ones in a keyword tool or Search Console.
5. On-site search (what visitors ask once they arrive)
The search box on your own site is a confession booth. Every query is a customer telling you, in their own words, exactly what they came for — and frequently failing to find it. This is the most underused source on the list.
How to capture it
- Turn on site-search tracking in your analytics platform (in GA4 the view_search_results event captures this automatically when your search uses a query parameter; otherwise configure it).
- Review the top internal search terms monthly. Pay special attention to searches that returned no results or led to an immediate exit — those are unmet demand you can fill with a new page or product.
- Cross-reference the vocabulary. If visitors search "returns" but your page is titled "refund policy", adopt their word. On-site search reveals the exact nouns your audience uses, which you can then echo in titles and headings.
Do watch for the same term recurring — three people searching the same missing thing usually means thirty more left silently. Don't dismiss low-volume internal queries; unlike Google, every one of these came from someone already on your site and already interested.
Putting the five sources together
No single source is complete. Keyword tools give you scale, Search Console gives you proven reality, the ad reports give you commercial intent, autocomplete and People Also Ask give you natural phrasing, and on-site search gives you the words of people already at your door. The practical workflow is to pull from all five into one sheet, then tag each term by intent and by the page type it implies.
| Source | Cost | Best for | Data type |
|---|---|---|---|
| Keyword tools | Mostly paid | Scale and volume estimates | Estimated |
| Search Console | Free | Terms you already rank for | Real |
| Ads / Amazon reports | Ad spend | Commercial, converting terms | Real |
| Autocomplete / PAA | Free | Phrasing and questions | Qualitative |
| On-site search | Free | Unmet demand from visitors | Real |
Start with the free, real-data sources you already own — Search Console and on-site search — because they describe customers you have, not hypothetical ones. Layer keyword tools on top for reach, sanity-check phrasing against autocomplete and People Also Ask, and mine your ad reports for the terms that actually open wallets. Do that consistently and your content will start answering the questions people are genuinely asking, in the language they actually use.