A Shopify checkout bot does not shop. It picks one item, usually one of the cheapest things the store sells, puts a single unit in the cart, and goes straight to checkout. The next bot does the same with the same item.
CartWatch measured this on 372 Shopify stores over the 90 days ending 26 October 2026. We compared 117,349 checkouts that CartWatch flagged as bots with 77,505 checkouts it did not flag. Where the cart was in US dollars, the median bot cart was $3.50. The median shopper cart was $79.94.
Three terms, so the rest reads plainly:
- A bot checkout is one CartWatch scored as a bot. A shopper checkout is one it did not flag. That means scored under the bot threshold, not verified human.
- A variant is one product in one size, colour, or option. It is what a Shopify cart actually holds.
- The typical store is the median store among the 164 stores that had at least 50 bot checkouts and 20 shopper checkouts to compare.
Key findings
- 97% of bot carts held one unit of one item. For shoppers it was 43%.
- 79% of bot carts were under $10. For shoppers it was 8%.
- At the typical store, 87% of bot checkouts held an item from the cheapest tenth of the store’s price range. For shoppers it was 15%.
- At the typical store, one variant took 54% of all bot checkouts. Shoppers’ most popular variant took 12.5%.
- On half the stores, the variant bots hit most did not appear in a single shopper checkout in 90 days.
- When a store’s most-hit variant changed from one fortnight to the next, the new target was still in the cheapest quarter of the store’s range 80% of the time.
One unit of one item
Bot carts are the same shape everywhere. Shoppers mix things.
| US-dollar carts | Bots | Shoppers |
|---|---|---|
| One item in the cart | 99.4% | 52.5% |
| One item, quantity one | 97.2% | 43.0% |
| Two to four units | 0.6% | 36.1% |
| Five or more units | 0.1% | 20.7% |
| Ten or more units | 0.0% | 9.4% |
This holds store by store. On nine stores in ten, at least 99% of bot checkouts were a single item. Nobody is adding three of something. A single unit of a single variant is what a cart permalink produces when a script opens /cart/{variant}:1 and skips the storefront.
Selling out does not stop them. 1,817 bot carts held only an item with quantity zero, which is what Shopify records when the variant is unavailable. The script kept opening the same link after the product was gone.
What the cart costs
| US-dollar cart value | Bots | Shoppers |
|---|---|---|
| $0 | 2.8% | 4.5% |
| $0.01 to $4.99 | 57.6% | 1.2% |
| $5 to $9.99 | 18.3% | 2.6% |
| $10 to $24.99 | 12.4% | 9.7% |
| $25 to $49.99 | 2.4% | 18.7% |
| $50 to $99.99 | 2.3% | 19.9% |
| $100 to $249.99 | 1.1% | 24.1% |
| $250 and up | 3.1% | 19.3% |
Half of bot carts were under $3.50 and nine in ten were under $25. Shoppers’ carts ran the other way: half were over $79.94.
The $0 row is gift cards, fully discounted samples, and free items. Shoppers had more of them than bots did. Free-product abuse is mostly a human problem, covered in Free Sample Abuse on Shopify.
Totals like these lean toward the biggest bot farms, so the typical store is the better guide. At the typical US-dollar store, the median bot cart was $3.99 and the median shopper cart was $60. On six US-dollar stores in ten, the median bot cart was under $5.
The cheapest thing in the store
A $3.50 cart could just mean a cheap store. It does not.
For each store, we took every variant that appeared in any checkout during the 90 days and ranked them by price. The typical store had about 140 variants on that list. Then we asked where each cart’s cheapest item sat on it.
| Typical store | Bots | Shoppers |
|---|---|---|
| Cart holds an item from the cheapest tenth of the list | 87% | 15% |
| Cart holds an item from the cheapest quarter | 95% | 33% |
Bots sit at the bottom of every store’s list. At the typical store, the bot’s item cost a quarter of what the shopper’s cheapest item cost. On almost every store the median bot cart was below the median shopper cart, and on nine in ten it was below half of it.
Look at the single variant bots hit most, and it is sharper still. On almost half the stores, that variant was the cheapest anyone checked out with. On six in ten it was one of the three cheapest, and on eight in ten it cost under 10 in the store’s currency.
Often it is a variant shoppers do not buy at all. On half the stores, the most-hit bot variant did not appear in a single shopper checkout in 90 days. Across those stores, 57% of bot checkouts sat on a variant no shopper used.
One variant takes the hits
Bots do not spread across the catalogue. They pile onto one variant.
| Typical store | Bots | Shoppers |
|---|---|---|
| Share on the single most-hit variant | 54% | 12.5% |
| Share on the three most-hit variants | 90% | 30% |
| Variants needed to cover half of checkouts | 1 | 8 |
| Distinct variants seen | 16 | 115 |
On more than half the stores, one variant took at least half of all bot checkouts. Only a handful of stores saw the most-hit variant take less than a tenth, and those are the stores where a farm walks the whole catalogue.
The biggest single target in the data was a $0.98 variant on one store, hit 2,442 times in 90 days. Almost all of the 25 most-hit variants across all stores were priced under 10 in the store’s currency.
Merchants write in and say “every abandoned checkout is the same $3 item.” That is not an outlier. It is the normal case.
Taking the product down moves the bot, not the attack
The first instinct is to hide or delete the targeted product. The data shows what happens next.
We split each store’s bot checkouts into two-week periods and found the most-hit variant in each. 104 stores had enough bot checkouts to compare across at least three periods.
- On 26 stores, the same variant was the top target in every period.
- On 21 stores, the top variant was different in every period.
- On the other 57, it changed at least once.
The top variant changed 169 times in all. 80% of the time, the new target was in the cheapest quarter of the store’s price list. The bots do not care about the product. They care about the price, and when one cheap variant goes, they take the next.
Overall, nothing moved. The median bot cart stayed between $1.96 and $4.99 in every two-week period of the window. The median shopper cart stayed between $66 and $109.
The expensive exception
Not every bot wants the cheap thing. About 3% of US-dollar bot carts were $250 or more. They came from 70 stores, but five stores had two thirds of them. The biggest had 810 bot carts with a median value of $3,382 and not one payment attempt.
These carts behave differently. On stores where CartWatch was not refusing payments, 11% of bot carts at $250 or more attempted payment. For bot carts under $5 it was 75%.
Why cheap
The price of the cart is the clearest sign of what the bot is for.
On the US-dollar stores where CartWatch refused no payments during the window:
| Bot cart value | Checkouts | Attempted payment |
|---|---|---|
| $0.01 to $4.99 | 22,923 | 75% |
| $5 to $9.99 | 8,188 | 63% |
| $10 to $24.99 | 7,953 | 49% |
| $25 to $49.99 | 1,234 | 31% |
| $50 to $99.99 | 927 | 26% |
| $100 to $249.99 | 528 | 30% |
| $250 and up | 1,543 | 11% |
Seven in ten bot carts under $10 attempted payment. Two in ten over $25 did. The fraudster has a stolen card and wants to know whether it still works before spending it on something they actually want. So they run a charge small enough that the cardholder will not notice it on a statement. A $2 approval tells them the card is live, and the card stays live because nobody reports or cancels it. A $200 charge on your store would get the card frozen before they could use it anywhere else. Your cheapest product is the test, not the prize. That is card testing, and it is why the cheapest variant on the store is the one that gets hit. How many of these attempts there are, and how few become orders, is in How Many Shopify Bots Are Actually Testing Credit Cards?.
The bots that never attempt payment pick cheap items for a duller reason. A script that only exists to create checkout records needs any variant it can add to a cart. The first result in a cheapest-first sort, or the first tile on a collection page, is what it finds. The expensive carts that never attempt payment look like the same kind of script pointed at a different page.
What this means for a merchant
Abandoned checkout recovery. For every shopper cart under $5 in this data there were 77 bot carts. A recovery email to a $3.50 cart is almost always going to a bot, and sending it creates a Klaviyo or Omnisend profile that bills you. If your recovery flow has a minimum cart value, this is the data for setting it. How much those profiles cost is in how bot profiles raise your Klaviyo bill.
Analytics. A cheap single-unit cart that never converts drags down average cart value, conversion rate, and the abandonment rate of the product it lands on. The product looks like it has a checkout problem. It does not. More in Fake checkouts hide your real abandonment rate.
Card testing. If your cheapest product is a $2 add-on, a sticker, or a sample, it is your card-testing surface. Every approval there confirms a stolen card for use elsewhere. Raising the floor price helps less than it seems, because the bots moved to the next cheapest variant 80% of the time.
The product itself. Hiding or deleting the targeted product gives you a quiet week on that product and the same volume on the next one. It also breaks real links. On the half of stores where no shopper bought the targeted variant anyway, leaving it up costs nothing. The fix is to score the checkout, not the catalogue.
Selling out. 1,817 bot carts hit a variant after it was unavailable. Running out of the targeted product does not end the attack.
CartWatch scores every checkout as it arrives, on the whole record. The combination of one cheap unit, no storefront visit, and a form filled in seconds is covered in How Shopify Checkout Bots Actually Behave. On the Standard plan, payment blocking refuses qualifying bot payment attempts, so the $2 test never reaches the card network. Shopify still creates the abandoned checkout.
How we counted
- Window and stores. Checkouts created from 28 July through 26 October 2026 on 372 stores running CartWatch. Banned shops, CartWatch’s own development stores, point-of-sale checkouts, and CartWatch’s own audit checkouts are left out.
- Records. 429,317 checkout records were read. 31,913 had their details wiped when the customer record was later deleted, 191,341 had no score (usually because they had no email), and 655 had no line items. All are left out. That leaves 117,349 bot checkouts and 77,505 shopper checkouts.
- Samples. 272 stores were read in full. The 100 busiest were sampled: about 3,000 checkouts each, spread across every day of the window. Counts are counts of records in the sample, not estimates of volume.
- Money. Dollar figures use the cart’s merchandise subtotal and only checkouts whose store currency is US dollars: 86,148 bot and 55,669 shopper checkouts. Carts in 17 other currencies are included in every percentage that does not quote a dollar amount. Per-store price ranks are in the store’s own currency.
- Price list. Each store’s list is every variant that appeared in any checkout on that store during the window, ranked by the lowest unit price seen for it. Free items are excluded. “Cheapest tenth” is the bottom 10% of those variants by price. Ranking only the variants shoppers used, so that bots could not define the floor themselves, gave the same answer: 78% of bot checkouts held an item from the cheapest tenth of what shoppers checked out with.
- Typical store. The median store among 164 stores with at least 50 bot checkouts and at least 20 shopper checkouts in the sample. The median such store had 368 bot checkouts and 171 shopper checkouts. 119 of the 164 are US-dollar stores. Where a figure is a count of stores, the denominator is those 164, or 163 where one store had no priced variant to rank.
- Payment attempts. “Attempted payment” means Shopify recorded a payment method on the checkout. The attempt rates by cart value use only the 121 US-dollar stores (44,363 bot checkouts) where CartWatch refused no payments at any point in the window, because a refused payment is not recorded.
- Over time. Two-week periods from the start of the window. A store contributes if it had three or more periods with at least ten bot checkouts; 104 did. Store counts in the body (“half the stores”, “nine in ten”) are out of the 164 typical-store set: 83 of 164 had a most-hit variant no shopper used, 148 of 164 were at least 99% single-item, 159 of 164 had a median bot cart below the median shopper cart (147 below half), 93 of 164 had one variant taking at least half of bot checkouts, and the most-hit variant was the cheapest on 72 of 163, among the three cheapest on 100, and under 10 in the store’s currency on 131.
These figures are stores running CartWatch, not Shopify as a whole. Stores usually install CartWatch because the abandoned checkout list already looks wrong, so the share of cheap single-unit carts here is higher than it would be on a random store. The price list is the catalogue as checkouts reveal it, not the full product list, so a store’s true cheapest product may never appear on it. Bots change behaviour, and the expensive-cart farms above are a second pattern already in the data.
Source: CartWatch Bot Intelligence. Analysis based on anonymized checkout activity processed by CartWatch across participating Shopify stores.
FAQ
Why do bots keep adding the cheapest product to my Shopify checkout? Because a card test wants the smallest charge that returns an approval and that the cardholder will not notice, so the card stays open for the purchase the fraudster actually wants. A script that only creates checkout records simply takes the first variant it finds. In 90 days of CartWatch data, 87% of bot checkouts at the typical store held an item from the cheapest tenth of the store’s range, and on almost half the stores the most-hit variant was the single cheapest one.
What does a Shopify bot cart look like? One unit of one item. 97% of bot carts in US dollars were a single item with quantity one, against 43% of shopper carts. The median bot cart was $3.50 and the median shopper cart was $79.94.
Will hiding the product the bots target stop them? Not for long. When a store’s most-hit variant changed from one two-week period to the next, the new target was in the cheapest quarter of the store’s range 80% of the time. The attack follows the price, not the product.
Do Shopify bots ever add expensive items? A few do. 3.1% of bot carts in US dollars were $250 or more, but five stores held two thirds of them, and only 11% of those carts attempted payment, against 75% of carts under $5.
Is a cheap one-item abandoned checkout always a bot? No. 1.2% of shopper carts were under $5 and 43% were a single unit. The combination is the tell: one cheap unit, no storefront visit, and contact details that arrive within seconds. CartWatch scores the whole record rather than the price alone.
CartWatch scores Shopify checkouts as they arrive. On the Basic plan and above, it can tag the customer and keep the address out of Klaviyo, Omnisend, Mailchimp, Drip, Brevo, and Privy. On the Standard plan, payment blocking blocks bots from attempting payment. Install CartWatch from the Shopify App Store.