Craft cannabis retail

The number that matters is the last one. Anyone can rank for their own name. Clicks from people who were not searching the brand grew 70%, and the brand’s share of its own top queries fell from 84% to 73%, which is what it looks like when search starts bringing in strangers instead of regulars.

The brand arrived late to a crowded category. Established competitors already held the obvious searches, and going straight at them with the main site would have meant years of losing to sites with a head start. The plan inverted the usual order: optimise the product pages first, before the site as a whole, and compete on the specific product searches nobody else had bothered to claim. Those queries convert harder because someone naming a product is already close to buying it, and they were winnable. The target was the top five. By April, 48 of the 50 tracked queries sat on the first page and the average position across them was 10.7, so the first-page goal was met broadly while the top-five target was not.

Website analytics figures are deliberately left off this page: that property was installed part-way through January 2026, so any growth figure drawn from it would measure the install date rather than the work.

All three name it, and one of the three puts it first. Asked for the best craft dispensary in Vancouver, ChatGPT answers with the brand outright, ahead of three alternatives it also names. Perplexity recommends it as a top pick for craft cannabis in BC and cites the client’s own site to get there. Gemini lists it among the city’s top craft dispensaries, second of its set rather than first, which is said that way here and shown that way in the screenshot above.

The questions it lost are in the report too. Delivery, mail order and product lookups returned competitors and grey-market sites, and the education questions were owned by American blogs. That list of misses is the content plan for the following month.

The June run also established that Google was withholding AI Overviews from cannabis queries entirely, across all fifteen. Knowing a channel is closed is worth as much as a ranking in one that is open. That is the same month ChatGPT could not be scored at all, because the unattended run met a login wall. Both are recorded as misses rather than quietly dropped, and the August captures are what closed the ChatGPT gap.

Answer engines are sampled, not guaranteed: results move month to month, which is the reason for tracking them monthly.

The mass-interest communities looked like the win and were not. One general-interest community absorbed the largest share of impressions in the opening month and returned the weakest click-through on the board. The small BC communities beside it, a fraction of the size, pulled better than twice its rate. Four months of that pattern is not noise, and the buying changed to match it.

By April the five strongest communities in the campaign were all local, and the reach behind them had grown nine-fold since January.

Figures cover the top communities reported each month. Click-through rates and ratios only: budgets and cost figures stay with the client.

Two of the three grew, and the flagship did not. The Kingsway store slipped over this window on Apple while it rose on Google. That is in the report rather than out of it: a channel that only ever produces good news is not being measured, it is being decorated.

The category-versus-brand split is the useful part. Three in four people who reached the stores through Apple were searching for a kind of shop, not a name they already knew, which is the definition of a listing doing acquisition rather than just catching people who were already looking.

Apple Business Connect reporting, January against May 2026.

Rebuilt from zero, and past where it started. Five months after the takedown the account was reaching three times as many people per month as it did in its last full month before it, on roughly a third of the followers. Against the previous full year’s monthly average it was reaching close to six times as many.

The reach also changed character. Before the takedown, most of it came from people already following. After the rebuild, two thirds came from people who were not, which is the difference between an audience being reminded and an audience being found.

An audience built up post by post over years can be gone inside a day, and on this platform that risk never goes away. What made the difference is that the strategy did not change. It had worked to build the account the first time, so it was run again exactly as it was, and it rebuilt the presence in a fraction of the time the original took. A strategy is only proven when it has to work twice.

Instagram account reporting. Follower counts, reach and impressions only.

Most print is spent on faith. A flyer goes out, sales move or they do not, and nobody can say which piece did the work. Here every piece had its own code, so a placement could be judged on its own and the weak ones stopped.

The scan volumes are modest next to the digital channels, and are shown that way rather than inflated. The point is not the size of the numbers, it is that the numbers exist at all: one placement produced two thirds of its link’s entire year inside a single month, which is only visible because it was being counted.

Link and QR scan reporting, 2023 through 2026.

Timing beat copy, decisively. The same abandoned cart, chased one hour later, converted at 11.9%. Chased the next day instead, 4.1%. Nothing about the message changed. That single finding is the argument for measuring flows separately rather than reporting an email programme as one number.

Overall conversion across every flow ran at 7.4% against a category benchmark of five to eight, and the weakest flow was identified and named in the same report rather than buried in the average.

Cart recovery alone brought back thousands of dollars of orders the business would otherwise have lost, from three flows that ran without anyone touching them.

Conversion rates and run counts only: order values and revenue attribution stay with the client.