The Baymard Institute puts the average documented online shopping cart abandonment rate at 70.22%, calculated across 50 separate studies published between 2006 and 2025. That is an all-ecommerce figure, not a cannabis-specific one, and it should not be read as a benchmark for your store. What it does establish is that leaving a cart is ordinary behaviour across online retail rather than a symptom of something broken.
So the question is not how to stop it. It is what happens next. Follow one shopper through and the answer gets concrete.
Tuesday, 4:40pm
She has been browsing for eleven minutes on her phone. Three items are in the cart. She is deciding between two of them, gets a call, and closes the tab. Nothing about this is a failure. She was interrupted.
In most retail categories, what happens next is an ad. She sees those three items in a display unit an hour later, then again the following morning, and one of those impressions brings her back. That entire mechanism is unavailable here. Google, Facebook, Instagram, X and TikTok all prohibit cannabis advertising, and broadcast is closed under FCC rules, so the standard retargeting loop simply does not exist for a dispensary.
Which means everything that happens from here has to happen on channels the store owns.
Beat one: she comes back on her own
Later that evening she opens the site again, this time to look at something else. If nothing is done, she lands on the homepage and her earlier session is gone. She is not going to reconstruct an eleven-minute browse from memory.
What should happen instead is that the store recognises the unfinished session and renders it back to her: the exact three items, at the current price, with their current availability, ready to continue. Not a message saying she has items waiting. The items themselves, on the page, in front of her.
The distinction matters because the friction being removed is memory. She does not have to recall what she was comparing. It is already there.
Beat two: the notification, after a delay
Suppose she does not come back. A push notification fires after a set delay.
The delay is the design decision. Immediately is wrong, because she was mid-decision four minutes ago and a nudge reads as surveillance. Three days is also wrong, because the decision is stale and the item may not be on the shelf. Somewhere in between is a reminder rather than a pester, and the right interval is a thing to test against your own store rather than to take from a template.
Push works here specifically because of what it does not require. No email address, no phone number, no consent flow beyond one tap taken earlier. For a shopper who browsed but never registered, it is often the only line that exists at all.
Beat three: the email that knows what she looked at
If she is a known customer, an email follows, and its content is assembled from that session rather than written in advance.
The contrast is stark. A generic email says the store misses her and links to the homepage, which drops her at the top of a catalogue and asks her to start over. A dynamic one shows the three products she actually left, with the price as it stands now and an honest availability state, linking to each item.
Honest is the operative word. If one of the three sold out, saying so is better than quietly omitting it, because the alternative is a shopper who returns for something that is not there. An out-of-stock item is also an opening for a relevant alternative, which is a genuinely useful message rather than a nag.
Why three connected beats beat three separate tools
Most stores have versions of all three of these. They usually do not have them connected, and the difference shows up in the seams.
Disconnected, each tool holds its own copy of what happened. The recovery page knows one thing, the notification tool knows another, the email platform knows a third, and none of them know what the others did. So she gets the push, comes back, buys, and then receives the email an hour later about the cart she already completed. Or the email shows a price the page no longer honours. Or all three fire and she reads it as three stores shouting rather than one store being helpful.
Connected, there is a single record of the session. The page, the notification and the email all read from it, all show the same items at the same prices, and any one of them completing the purchase stops the others. The shopper experiences one coherent thread instead of three uncoordinated ones, which is the actual difference in outcome, not the number of touchpoints.
Why this carries more weight here
In most verticals, owned recovery is one lane among several, running alongside paid retargeting that catches the shoppers it misses. In cannabis there is no alongside. The ad platforms are closed, direct mail is closed for cannabis marketing, and broadcast is closed. Whatever share of abandoned sessions gets recovered is recovered here or not at all.
That is an argument for building the loop properly rather than for messaging harder. See checkout and cart recovery for how the three beats are wired together, CRM and marketing automation for the sending side, and personalization for how the session informs what she sees next time.