Almost every keyword research guide starts by opening a tool. This one starts by asking what the tool is for, because in this category the answer changes the whole exercise.
The industry-standard source of search volume is Google's own Keyword Planner, and it is an advertising product. A dispensary cannot buy the advertising it plans. What is left is either a third-party estimate that Google says has no access to its ranking data, or your own Search Console, which is free, first-party, and incomplete in a way we can measure exactly.
The tools were built for advertisers, in a category that cannot advertise
Read how Google describes Keyword Planner and notice the nouns: "View estimates on the number of searches a keyword gets each month", "View the average cost for your ad to show on searches for a keyword", and a forecast that "shows you how many conversions, clicks, or impressions you're likely to get for your keywords based on your spend". Forecasts are "refreshed daily and based on the last 7-10 days, adjusted for seasonality".
That is a media planning instrument. It exists so an advertiser can size a budget. And Google's advertising policy disallows ads for substances that alter mental state for recreation, with a narrow carve-out for certified advertisers promoting topical hemp-derived CBD in three jurisdictions. A dispensary is being sold planning data for a channel that is closed to it.
The alternative most agencies reach for is a third-party tool, and Google is blunt about what those are:
Google doesn't evaluate or endorse third-party SEO tools, and these tools don't have access to Google's internal ranking data. Be wary of tools that claim to be "acceptable" or "approved" by Google Search.
Google Search Central, Do you need an SEO?, read September 8, 2026
None of that makes the tools worthless. It makes their numbers models rather than measurements, and it means a plan that rises or falls on a volume figure is resting on somebody's estimate of Google rather than on Google.
What Search Console shows, and what it withholds
Search Console is the only place a store sees its own queries, with its own impressions, clicks and average positions. It is also the only source in this whole discussion that is a measurement rather than a model.
It is not complete, and the shortfall is worth quantifying on your own property rather than waving at. The arithmetic takes a minute:
- Sum the query table. Total the impressions and clicks columns across every row of the queries export.
- Compare against the device totals. Those are in the same export, on the devices tab.
- The difference is what Google did not name. Express it as a percentage of each.
Expect the gap to be larger than you assume, and larger the smaller the site, because the rarest queries are the ones that go unlisted. It is common for the clicks you cannot attribute to a named query to outnumber the ones you can.
Two honesty notes. First, we went looking for a Google page stating the rule behind that withholding and did not find one on 8 September 2026, so treat the gap as an observed behavior rather than a documented policy. Second, the pages table in the same export will usually not reconcile with the device totals either: Google's documentation attributes chart-and-table differences to how the data is aggregated, property against page. Neither number is wrong. They are counting different things.
The practical consequence is a change of posture. Search Console does not hand you a keyword list. It hands you evidence about the queries you already touch, and the rest of the job is judgment.
"Dispensary near me" is not a keyword you can research
This is the query every dispensary wants and the one the tools handle worst, for a reason that is documented rather than mysterious.
Google's local ranking documentation describes distance as how far each business is from the customer who is searching, and says that if a customer does not share where they are, Google uses what it knows about their location. The phrase is identical for everyone who types it. The result set is different for everyone who types it.
So a volume figure for "dispensary near me" is a national aggregate of millions of separate local contests, and it cannot be split by city because the split never happens in the query. Any tool showing you a city-level figure for it has modeled that number, not measured it.
What replaces it is a position measurement across the area you serve rather than a volume estimate. The Map Pack guide covers the mechanics, and the practical version is simple: sample the same search from points across your trade area and see where your visibility stops.
A method that does not need a volume number
Six steps, in order, none of which require buying anything.
- Export your own queries first. Search Console, last three months, Web, with the comparison to the previous period. Everything after this is easier when you start from what you already touch.
- Sort them into families before you look at the numbers. Define the buckets in advance, so the data cannot talk you into a flattering split afterwards. Ours are brand, on-topic non-brand, geographic, and off-topic legacy. A store's would be more like brand, product and category, proximity, and questions.
- Find the queries where you appear but do not get clicked. Impressions with no clicks at positions 8 to 30 are the most actionable rows in the export: Google is already willing to show you and something is stopping the click, which is usually the title, the page, or the fact that a Map Pack sits above the result.
- Write down the questions your staff actually get asked. This is the input no tool has. What people ask at the counter is a live sample of what they typed on the way in.
- Check the results before you write. Search the query. If the top results already answer it well, and better than you could, that is a query to skip regardless of what a volume figure says.
- Decide what you can source. A page you can build from something checkable, your own hours and stock, your state's license file, your own data, will outlast a page assembled from other people's summaries.
Nothing in that list needs a monthly search volume. It needs one export, one afternoon and a willingness to skip queries you cannot serve better than the incumbent.
Where keyword-first goes wrong
Google publishes the failure mode, framed as questions to ask yourself. Answering yes to these is, in its words, "a warning sign that you should reevaluate how you're creating content":
- "Is the content primarily made to attract visits from search engines?"
- "Are you producing lots of content on many different topics in hopes that some of it might perform well in search results?"
- "Are you using extensive automation to produce content on many topics?"
- "Are you mainly summarizing what others have to say without adding much value?"
- "Does your content leave readers feeling like they need to search again to get better information from other sources?"
The second question describes the standard workflow precisely: export a keyword list, sort by volume, turn each row into a title, publish on a calendar. It is not that the workflow is banned. It is that Google has written down what it looks like from the outside.
The same guidance covers automation directly, asking whether the use of AI generation "is self-evident to visitors through disclosures or in other ways", whether you explain how it was used, and why it was useful. And the scaled content policy catches the output regardless of the tool: many pages generated primarily to manipulate rankings, unoriginal content with little value, "no matter how it's created". If keyword lists are your input to bulk generation, that policy is the one to read first, and the multi-location spoke covers it in full alongside the doorway rules.
How to read the split once you have it
Sort your queries into the families you defined and three patterns show up on nearly every dispensary property.
Brand queries flatter everything. People searching your store name were always going to find you, and they will carry the best position and click-through rate in the export. Report them separately or they will hide whatever the rest of the site is doing.
The specific queries land before the general ones. Head terms, the words you most want, tend to sit deep or show no impressions at all, while narrow ones with a place or a question in them arrive first and arrive at real positions. That is the normal shape of early progress, not a sign the work failed, and reading it the other way round is how good programs get canceled in month three.
Anything the site used to be about will still be in there. Old topics, discontinued products, a previous owner's pages. If those sit in the same total as everything else, a quarter where your actual category grew can read as a quarter where nothing happened. Split them out and keep them split.
Had you built the plan from a volume tool instead, you would have written for the head terms and concluded that the first months produced nothing. The export says something more useful: which specific, answerable queries Google is already willing to show you for, and which general ones it is not yet willing to rank you for. That is a plan, and it came out of a free report.
The working checklist
- Verify Search Console for your domain, if that has never been done.
- Export three months of queries with the comparison period on.
- Measure your own withheld share by summing the query table against the device totals, so you know how much of the picture you are missing.
- Define your families before analyzing, and write them down.
- List every impression-bearing query at positions 8 to 30 with no clicks. Work that list first.
- Collect twenty questions from the counter and check which of them appear in the export.
- Stop treating "near me" as a keyword and start measuring position across your trade area.
- Skip any query you cannot answer better than what currently ranks.
- Publish nothing you would not publish if search did not exist.
- Re-export monthly and compare families, not positions.
Frequently asked questions
What is the best keyword research tool for a dispensary?
Your own Search Console, and it is free. Every volume tool is either Google's advertising planner, whose numbers are described by Google as estimates and whose forecasts are built around your ad spend, or a third-party model. Google states plainly that third-party SEO tools do not have access to its internal ranking data and that none is approved by Google. A dispensary also cannot buy the ads the planner exists to plan, since Google's advertising policy disallows ads for substances that alter mental state for recreation.
Does Search Console show every query my dispensary appears for?
No, and the gap is larger than most people expect. You can measure it on your own property in a minute: sum the impressions and clicks columns across every row of the queries export and compare the totals against the devices tab in the same export. The difference is traffic that arrived from queries Google does not name, and on a small site the unnamed clicks routinely outnumber the named ones. We could not find a Google page stating the rule behind that, so treat it as an observed behavior rather than a documented policy.
How do I find search volume for "dispensary near me" in my city?
You cannot, because that quantity does not exist. "Dispensary near me" is a single query string that every person types identically, and Google resolves it differently for each of them: its documentation describes distance as how far each business is from the customer who is searching, and says that when a customer does not share their location Google uses what it knows about it. A national volume figure for the phrase describes a national aggregate. It tells you nothing about your trade area, and no tool can split it by city because the split does not happen in the query.
How should a dispensary choose what to write about?
Start from questions a customer asks before choosing a store, then check whether you can answer them better than the current results. Google publishes the warning signs for the opposite approach: content made primarily to attract search visits, lots of content on many topics hoping some performs, extensive automation across many topics, and content that leaves readers needing to search again. If a page would not exist were it not for a keyword, that is the signal to stop.
Can I use AI to write dispensary content from a keyword list?
Google's policies are about output rather than tooling, so the answer depends entirely on what comes out. Scaled content abuse covers many pages generated primarily to manipulate rankings, creating large amounts of unoriginal content with little value, and Google says explicitly that this applies no matter how it is created. Google also asks whether the use of automation is self-evident to visitors through disclosure, and why automation was useful here. A keyword list fed to a generator is the shape of thing those policies describe.
What is a realistic keyword target for a new dispensary site?
Long-tail questions where the current results are thin, rather than the head terms. Our own site is the example we can show you: ten weeks after relaunch, our commercial head terms drew two impressions or none at all, while sourced data questions and service-style geographic queries sat between positions 5 and 12. The pattern to expect is that specific, answerable queries arrive first and the words you most want arrive last, if at all.
Should I target the same keywords as the dispensary down the road?
Only after checking whether they can be reached from where your customers are. In local results, distance is measured from the person searching, so a competitor ranking for a term two neighborhoods over is not necessarily beating you on a query you could win. Map position across your trade area answers that question and a keyword list does not.
How do I know if my keyword work is doing anything?
Compare exports rather than checking positions. Pull the same Search Console report each month, split queries into families you defined in advance, and look at whether the families you are trying to grow are gaining impressions and moving up, separately from families you no longer sell. That last part matters: on our own site, 62 percent of impressions in the last export landed on legacy topics we do not serve, and folding them into one total would have made a flat quarter look like a good one.
