Dispenza
Book demo

How we measure the rankings

Every number on these pages is a live Google Maps reading, dated and counted point by point.

What a scan is, what a round is, what we remove, what we tag and how a share is worked out. Short, because the numbers should do the talking.


What a scan is

A scan reads Google Maps at 121 points laid out as an 11×11 grid, centred on the store's own Google pin. The grid reaches 2 miles out from the pin in New Jersey, Rhode Island and Washington DC; in Nevada, 2 miles in Clark County and the Reno area, 5 miles elsewhere; in California, 2 miles in the 12 metro counties of Greater Los Angeles, San Diego, the Bay Area and Sacramento, 5 miles elsewhere. Everywhere else we scan, it reaches 2 miles in the 234 core counties of the New York City, Buffalo, Rochester, Boston, Worcester, Providence, Detroit, Grand Rapids, Chicago, St. Louis, Seattle, Portland, San Juan, Hartford, Philadelphia, Baltimore, Washington DC, Pittsburgh, Richmond, Virginia Beach, Miami, Tampa, Orlando, Jacksonville, Columbus, Cleveland, Cincinnati, Charlotte, Raleigh, Memphis, New Orleans, Kansas City, Minneapolis, Oklahoma City, Tulsa, Dallas-Fort Worth, Houston, San Antonio, Austin, Denver, Salt Lake City, Phoenix, Tucson, Honolulu, Atlanta, Louisville and Birmingham metro areas and 5 miles elsewhere, all of Maine, Vermont, West Virginia, Iowa, North Dakota, South Dakota, New Mexico, Montana and Alaska included. Each store is scanned for “dispensary near me”, “cannabis near me” and “weed near me”, one grid per search.

At every point we record the order of the listings a shopper standing there is shown. Every point reads the first 20; the centre point reads the whole list, so a store's position at its own door is known however deep it sits.

Lists for “weed near me” ran about 8 listings a point (median), against about 20 for “dispensary near me” and “cannabis near me”; the centre point, which reads Google's whole list, holds about 8 too. Every scan runs the same settings, so the short lists are Google's own, and a store page says how many listings Google showed rather than “not in the top 20”.

What a round is

A round is one pass over every store on our list. A store's scan opens a new round when it runs at least 7 days after its last one, and each round is compared with the one before it, like for like: only the map squares both rounds measured count in a change, so a round still being scanned never reads as a fall.

1 round so far; the latest, October 24, 2026, is still being scanned (14,854 of 30,891 store searches read).

What we remove

Google mixes in businesses because of where the scanning server sits, in Council Bluffs, Iowa: shops in and around Omaha, and a few others nowhere near the grid. Nobody searching anywhere we scan is shown them, so we remove them and renumber the positions that remain. They are the only listings we ever remove: a far-off store that Google really shows a shopper there stays on the board like any other. Left in, the removed ones would have held a top-three spot 395,793 times in the scans behind these pages.

A grid near a state line reaches across it (some Hudson County grids reach Manhattan). Those points stay on the store's own grid, in a dashed ring, and count in its own numbers, but never in a county, city or state number. A listing from across the line that ranks inside the state is on the boards like any other.

What we tag

We tag a listing lead-gen when its name is spelled with lookalike letters; when the same dispensary-style name sits on three or more listings at different addresses that carry next to no reviews; or when a dispensary-style listing that is not on our store list borrows another company’s brand, or has next to no reviews and a short link for its website or a tagged network’s name. Real chains are excepted, and a check by hand can add the tag to a listing or take it off.

Tagged listings stay on every board in the position Google gives them, because shoppers see them there. The tag is reported as a number too: the share of top-three slots tagged listings hold in each area. 484 listings are tagged so far. Where the evidence is not clear, a listing is left untagged.

What the numbers mean

Map square.

Every reading is placed in a square about a quarter mile on a side. Neighbouring stores’ grids overlap; a square is counted once, however many grids cross it.

Share of the top three.

The part of an area’s measured squares where a listing shows in the top three, counting a square in part when it holds the spot in only some of the readings there. The combined board averages the three searches.

Share of Local Voice.

On a store’s own grid: the share of its 121 points where it holds the top three, for one search.

Average position.

The listing’s mean position where it shows within the first 20, so the deep list at the centre point does not tip it.

Gaining and losing.

The change in a listing’s share between two rounds, in percentage points, over the squares both rounds measured.

Cities and the directory

A store's city is Google's postal city for its address (the licensed-store roster's where Google gives none). A city's market is every map square its stores' grids read, whichever scan made the reading; a city sits in the county where most of its stores are.

Each store's listing gives its name, address and website, Google's where its listing has them, else the roster's, and links to its Google Maps listing and directions.

Websites and the firms behind them

Each month, every store website on our list is loaded once by Google's PageSpeed Insights as a mobile visit. The load time shown is its Largest Contentful Paint, the seconds before the main content appears: real Chrome visits first (Google's Chrome UX Report, 75th percentile), where Google has enough of them, and beside it the lab test, which loads the page on a simulated slow phone and reads slower than most real visits. A lab reading the test could not repeat is left out, and an address that answers with an error page is not timed. Google's Core Web Vitals call 2.5 seconds or less good and more than 4 seconds poor.

Detection reads what a store’s homepage says in public: a credit in its footer (“Website by …”, “Marketing by …”) or a vendor’s signature in its code. The credit’s own words set the role: a site, design or “powered by” credit is the website; only a marketing or SEO credit is marketing. A firm that credits no one does not show up, so a firm can be missing from these numbers. A website credit says who built the site; it does not show who does the store’s Google Maps work. The firms are compared on the agency insights, by the share of their stores’ own grids in the top three, because county shares are not comparable across counties of different size.

Dates and notes

Scans behind the latest round: Arizona: October 25 to 27, 2026; California: October 24 to 27, 2026; Connecticut: October 26 to 28, 2026; Washington DC: October 28, 2026; Delaware: October 25 to 28, 2026; Florida: October 28, 2026; Massachusetts: October 26 to 28, 2026; Maryland: October 28, 2026; Maine: October 28, 2026; Michigan: October 27, 2026; North Carolina: October 28, 2026; New Hampshire: October 26 to 28, 2026; New Jersey: October 24 to 28, 2026; Nevada: October 24 to 27, 2026; New York: October 24 to 28, 2026; Ohio: October 28, 2026; Oregon: October 27, 2026; Pennsylvania: October 24 to 28, 2026; Puerto Rico: October 28, 2026; Rhode Island: October 27 to 28, 2026; Utah: October 25, 2026; Virginia: October 28, 2026; Vermont: October 27 to 28, 2026; Washington: October 27, 2026; West Virginia: October 28, 2026. Each page carries its own dates, and every page is rebuilt as new scans land. The scanning tool attaches a note to a scan when it leaves points out; none so far.