Free Keyword Research for a Coffee Shop in One Afternoon

Thomas Garner

Oct 7

11 min read

Short answer: you do not need a keyword tool. You need one afternoon, a private browser window, and a method for turning what people actually type into a short list of pages your site should have. This post is that method, built entirely from free sources: Google autocomplete, the map pack, People Also Ask, and Search Console if your site already exists.

A lot of owners hear target your keywords and assume it means buying a tool and staring at a spreadsheet of numbers. It does not have to. What a cafe actually needs is a page map, not a keyword list: a few dozen real phrases people search, sorted by what they are trying to do, each one pointed at a page on your site, or honestly marked as not worth a page yet. This method gives you that map for free. It will not give you search volume, and it will not pretend to. It gives you direction, which for a single location cafe is usually enough.

Autocomplete: what Google suggests before you finish typing

Open a private browser window so your own search history does not skew the results, and start typing a handful of seed phrases into Google's search box without hitting enter. Watch what autocomplete suggests. Good seeds for a cafe include coffee shop near me, best coffee in your city, cafe open now, coffee shop with wifi, coffee shop with pastries, oat milk latte near me, coffee roaster in your city, local coffee beans, coffee shop for meetings, and dog friendly coffee shop. Run each one in both a small city and a mid sized city if you can, since the suggestions genuinely differ by area size, and record what comes back for each.

Autocomplete suggestions are Google showing you, in real time, phrasing that real searchers use, which is worth more for a small cafe than any invented list of keywords a tool might generate. The catch is that autocomplete gives you no sense of how often any of these phrases get searched. It tells you the shape of the demand, not the size of it. Treat every suggestion as a direction worth considering, not a number to chase.

As you run through your seed list, also pay attention to People Also Ask, the expandable question box that often appears partway down a Google results page. These are phrased as full questions, which makes them especially useful raw material for an FAQ section or a blog post, since a person is more likely to type or say a full question than a fragment. Record each question exactly as it appears, along with which seed search triggered it, so you can trace where each idea came from later.

Keep two files as you go, not one. The first is a raw dump, every suggestion, every question, exactly as it appeared, untouched. The second is a cleaned sheet you build afterward, once you have looked at the full raw list and started noticing duplicates and phrases that clearly do not apply to your cafe. Keeping both matters because it lets you, or anyone reviewing your work later, trace a final decision back to the original suggestion rather than trusting a cleaned list on faith. Note how many distinct phrases you collected in the raw dump and how many you discarded as duplicates or off topic when you build the cleaned sheet, since that ratio is a useful gut check on how much real variety your seed searches surfaced.

Maps and People Also Ask: what a real search returns

Search coffee shop in your own city, exactly as a customer would type it, and look at the full page, not just the map pack. Note what appears: the pack itself, the organic results below it, and any related searches shown at the bottom of the page. What you are looking for is which page types show up for this kind of query, whether that is a homepage, a location page, a menu page, or a blog post, because that tells you what kind of page Google currently treats as a reasonable answer to this exact phrasing.

This step matters because it grounds your keyword list in reality rather than guesswork. If every result for a phrase is a homepage, that suggests the phrase belongs on your homepage, not a separate page built just for it. If results include specific location or neighborhood pages, that suggests a phrase like that deserves its own page if you serve more than one area. Reading the actual results, rather than assuming, is the fastest way to avoid building a page nobody was going to find useful anyway.

Take note of the organic top five as its own row in your table, separate from the pack. The pack is largely governed by Maps signals rather than website content, while the organic list below it is a more direct read on what kind of written page Google is currently surfacing for that exact phrase. A cafe deciding whether a phrase deserves a dedicated page, a paragraph on an existing page, or nothing at all should weigh the organic list more heavily than the pack for that decision, since the organic list is closer to what your own site content is competing against.

Related searches at the bottom of the results page are worth the same treatment as autocomplete: free, direct evidence of real phrasing, with no volume attached. Collect them the same way, noting the seed query that produced them, and fold them into your growing list alongside the autocomplete and People Also Ask results.

Do the same walk for at least one seed phrase built around a specific need rather than just location, such as coffee shop with wifi or dog friendly coffee shop. These amenity focused searches tend to return a more mixed set of results, sometimes a homepage, sometimes a list style article from an outside site, sometimes a review site. That mix is itself useful information: it tells you there is real search interest in the amenity, and that the field is more open, since fewer cafes appear to have built a page specifically answering it. An open field like that is often the easiest place to start.

Search Console: what people already found you for

If your site already exists, Search Console is the one free source in this method that shows real queries tied to your actual pages, not a general guess at what people might search. Hillcane's own site offers an honest, small case for why this step matters and why the timing matters too. Wix Analytics, the platform Hillcane runs on, keeps only 62 days of Search Console style data before it rolls off, which means an owner on a similar platform needs to export monthly rather than assume the data will still be there later.

Pull your query report and look at three things for each query: how many times it showed up in search results, how many times someone clicked, and roughly where your page tended to rank. A query with real impressions but very few clicks is worth noticing, since it often means Google already understands your page is somewhat related to that phrase but the title or description is not convincing enough to earn the click. A query you rank for that you never intentionally targeted is also worth noticing, since it is free evidence of a phrase working for you already, without you having tried.

For a genuinely new site, this data will be thin at first, and that is fine. Treat it as a small, dated case rather than a trend, the same way Hillcane treats its own numbers. The value is directional: even a handful of real queries from Search Console tells you more about what is happening on your specific site than any general list of coffee shop keywords ever could.

One more use for this data that owners often overlook: comparing it against the phrases you collected from autocomplete and People Also Ask in the earlier steps. If a phrase shows up in both places, real autocomplete suggestion and real Search Console impressions, that is a genuine signal worth weighing more heavily than a phrase that only appeared once in your raw collection. This is not a volume number, but it is a form of confirmation, two independent free sources agreeing that a particular phrase matters, and it is exactly the kind of low cost cross check this whole method is built around.

Turning terms into pages

Once you have your raw collection, autocomplete suggestions, People Also Ask questions, related searches, and any Search Console queries, the next step is grouping, not counting. Sort every phrase into one of four intents: find a place, someone deciding whether this business exists and where it is; check details, someone confirming hours, menu items, or amenities; decide, someone comparing this business against alternatives, often using words like best or near me; and learn, someone asking a broader question that a blog post could answer better than any single product or location page.

Map each intent to a page type. Find a place phrases point to your homepage or a location page. Check details phrases point to your menu, hours, or amenities information, wherever that lives on your site. Decide phrases often belong on the homepage or a comparison minded page if you have reason to build one. Learn phrases are blog post territory, and often make the best starting points precisely because they require the least new page architecture, just a well written post answering the actual question.

By the end of the afternoon you should have a filled table: phrase, intent, target page or none. Some phrases will map to a page you already have. Some will point at a gap worth filling. And some, honestly, will not be worth a page at all, at least not yet, and that is a legitimate outcome of this exercise too. Pick the three phrases with the clearest intent and the most obvious existing gap on your site, and write those pages first. That is the whole method: not a keyword list, a page map, built for free, in one afternoon.

Resist the temptation to do this once and consider it finished. Search behavior around a specific cafe shifts with seasons, with menu changes, with new amenities, and with whatever is happening in the neighborhood. A method built entirely from free tools costs nothing to repeat, which is exactly its advantage over a paid subscription that gets cancelled after the first month. A light version of this same afternoon, maybe an hour instead of a full afternoon, once every few months, keeps your page map honest as your cafe and your customers' habits both keep changing.

It is worth restating plainly what this method cannot do, since being honest about the limits is part of what makes it trustworthy. It will not tell you how many people search a given phrase each month. It will not rank phrases by competitiveness. It will not replace a paid tool for a business with the budget and the need for that level of precision. What it will do is give a single location cafe with no budget a real, evidence based starting point, built from what actual searchers type, rather than a guess.

Frequently Asked Questions

Do I need a paid keyword tool to do this properly?

No. This method uses only free sources: Google autocomplete, the map pack and the organic top five beneath it, People Also Ask, related searches at the bottom of the page, and Search Console if your site exists. None of them give search volume, and this post invents none. For a single location cafe, direction usually matters more than volume, and an afternoon with a private browser window and seeds like coffee shop near me or dog friendly coffee shop gives you direction for free.

What keywords should a cafe target?

Rather than a fixed list, this method has you collect real phrases from your own seed searches, sorted into four intents: find a place, check details, decide, and learn. The right phrases differ by city and by what your cafe offers, which is exactly why this method builds your list from your own searches rather than a generic list.

How long does Wix keep Search Console style data?

Wix Analytics, the platform behind hillcane.co, keeps Search Console style data for 62 days before it rolls off. So exporting monthly is the safe habit; waiting longer risks losing the window entirely. When you export, keep three things per query: impressions, clicks and rough average position from the Performance report. This retention figure applies specifically to Wix, and other platforms may keep data for a different length of time, so check yours.

My site is brand new. Is Search Console even worth checking?

Yes, but treat what you find as a small, dated case rather than a trend, the same framing this post uses for Hillcane's own early numbers. Even a handful of real queries, each with impressions, clicks and rough position from the Performance report, says something specific about your site that a generic list cannot. If you are on Wix, export monthly, since Wix Analytics keeps this data for only 62 days before it rolls off.

What is the difference between autocomplete and People Also Ask?

Autocomplete shows suggested completions as you type into the search box, before you hit enter, which reveals real fragment level phrasing such as cafe open now or oat milk latte near me. People Also Ask shows full questions in an expandable box partway down the results page, which makes it good raw material for an FAQ section or a blog post. Record both exactly as they appear, along with the seed search that triggered them.

Should I run my seed searches in a private browser window?

Yes. A regular browser window can personalize autocomplete and results from your own search history and location habits, which skews what you see away from what a typical customer encounters. A private window gives a cleaner read. Run each seed, such as best coffee in your city or coffee shop for meetings, in both a small city and a mid sized city if you can, since suggestions differ by area size, and record every result in your raw file.

What if a phrase does not clearly fit any of the four intents?

Some phrases will not fit cleanly, and that is fine to note rather than force. The four groups, find a place, check details, decide and learn, organize the bulk of what you collect; each maps to a page type, such as a homepage or location page for find a place and a blog post for learn. Mark an odd phrase as none in your phrase, intent and target page table, and revisit it at your next quarterly pass.

How many phrases should I end up with after an afternoon of this?

This method sets no target number, since the total depends on how many seed searches you run and how many cities you test. The suggested seeds alone number ten, from coffee shop near me to dog friendly coffee shop. Keep a raw file and a cleaned sheet, and note how many phrases you collected versus discarded as duplicates, which is a gut check on variety. What matters most is that every kept phrase maps to a page or is marked not yet.

Can I use this method for a second location in a different city?

Yes. Run the same seed searches with the second city's name substituted in, in a private browser window, and compare the pack, the organic top five and the suggestions against your first location's results. Different cities often surface different phrasing, which is why this method checks more than one city size. If the organic results include specific location or neighborhood pages, that suggests each location deserves its own page with its own title.

What do I do with the three phrases I pick to write first?

Build or update the page each one maps to, using the phrasing and intent you identified rather than guessing. A find a place phrase belongs on your homepage or a location page; a check details phrase belongs wherever your menu, hours or amenities live; a learn phrase is often best served by a focused blog post answering the question directly. Then repeat a lighter, roughly one hour version of the whole method every few months.

Work with Hillcane

You do not need a keyword tool. You need one afternoon and four free sources.

If building the page map feels like more than one afternoon can hold, Hillcane can help turn your own searches into a working content plan. Reach out at hillcane.co/contact or call (256) 384-2449.