fraud · analytics · email · checkout
Fake Abandoned Checkouts: Clean Your Data and Email List
Spot fake abandoned checkouts, clean up your store analytics, and keep bot signups out of your email list so real customers still get your cart reminders.
By cartt.io Team · November 20, 2026 · 11 min read

You open your abandoned-cart report and see the same street address on checkout after checkout. Your conversion rate has dropped overnight, and yesterday's cart reminders are bouncing. If you suspect fake abandoned checkouts, pause the affected reminders, check whether paid orders actually fell, and tighten who qualifies for a recovery email before you resume sending.
What are fake abandoned checkouts and why do they happen?
A fake abandoned checkout is an unfinished checkout created by an automated script or test that had no intention of buying. A scraper can start checkout to read shipping or pricing details. A card tester can use it to reach the payment step. A crawler may fill in a form while checking whether your advertised item is available. All three can leave behind records with placeholder names or example email addresses.
When you see what looks like a batch of fake orders, open one record and check its payment status. Shopify records most failed payment attempts in the abandoned checkout timeline. A paid order will have a payment record you can inspect. That distinction tells you whether to start with your reminder flow and reports or investigate actual charges.
How do you spot bot traffic in your store analytics?
Start with the hour the spike began. Pull checkout starts and paid orders for that window, then compare them with the same time on a comparable day. If checkout starts are up sharply and paid orders are steady, sample the new checkouts. Look for the same shipping address with changing names, email addresses built from the same pattern, or several submissions at unusually regular intervals. Check whether messages tied to that batch started bouncing around the same time.
In GA4, break out the spike by geography and browser. If your host exposes request logs, check whether the checkout requests cluster around a hosting provider or a small set of IPs. Geography is a clue. A new campaign can bring a wave of shoppers from one city, and a real shipping or payment error can leave you with a pile of unfinished checkouts. Check the campaign launch time and the payment timeline before treating the whole spike as automation.
| What you see | What to check next |
|---|---|
| More checkout starts, steady paid orders | Sample the new records for repeated fields and group them by time. |
| Fewer paid orders, familiar-looking checkout records | Try the checkout yourself and look for a payment or shipping error. |
| More recovery email bounces from one batch | Open the email provider's bounce details and check which errors are permanent. |
You can start your bot traffic analytics in a spreadsheet. Put checkout starts, paid orders, and reminder bounces side by side, with one row per hour. Record when the related reminders actually sent, since a delay in your flow can put the bounce spike hours after the suspicious checkouts.
How do fake checkouts hurt your analytics and ad numbers?
Say that, on a normal day, 30 of 120 checkout starts result in a paid order. In this hypothetical example, you see the same buying activity the next day, plus 180 extra starts from scripts. Your checkout completion rate falls from 25% to 10%. If those scripts also register site sessions, they can pull down your session-based conversion rate as well. Before you change shipping prices or rewrite the page, check what happened to the count of paid orders.
For your checkout funnel, count unique checkouts started in the same window and see how many eventually paid. Give both groups the same time to convert. Pull payment status from your store or processor. In Stripe Checkout, inspect the session's payment_status. A session can complete while a delayed payment is still pending. If you send a GA4 purchase event, use the order's unique transaction ID so repeat events for the same web transaction can be deduplicated.
Then open your ad account. If checkout starts count as the campaign's purchase goal, fake starts can show up as successes and feed automated bidding. In Google Ads, you can make a payment-confirmed purchase the primary conversion and keep checkout starts secondary for observation. Check the campaign's goal settings too. A secondary action included in a custom goal can still be used for bidding. For a period you suspect is polluted, compare campaign results with paid-order totals you can verify. You may be able to reconcile the sales count even when the original ad source is unclear.
GA4 automatically excludes known bots, and Google doesn't show you how much it excluded. Scripts it doesn't recognize can still show up. The internal-traffic filter is for IP addresses you identify as your own. For past traffic, use a report filter or exploration and keep the original for comparison. If your property has Google's Custom condition traffic filter, start in Testing and check whether it catches legitimate purchases. Activating it permanently drops future matching events. The feature is rolling out, so it may be absent from your property.
Use cart abandonment statistics for context once you trust the traffic you're comparing. A storewide benchmark can't tell you what happened in a single burst of suspicious traffic.
How do they damage your abandoned-cart emails?
When someone enters an address that doesn't exist, a reminder can hard-bounce. Someone whose real address was entered by a script might receive a cart they never started and report it as spam. Both can hurt your email sender reputation. Open the recovery flow's own report so a healthy newsletter average doesn't hide a spike in the abandoned cart email bounce rate.
If the spike is still coming in, pause the recovery automation. In Klaviyo, switch all actions in that flow to Draft to stop entry and sending, or use Manual if you can review recipients individually. Manual messages wait for approval. Review any queued recipients in your email tool before restarting, and leave the suspect batch out of any catch-up send. If both your store and email provider send reminders, check both automations.
For recovery eligibility, use addresses already confirmed through a sign-in code or a confirmed subscription, where your store can pass that signal to the email tool and you have permission to send the reminder. Basic form validation catches malformed addresses. Mailbox confirmation shows the recipient could open a message at that address. An existing confirmed address can still be submitted by someone else, so an authenticated checkout gives you a stronger link to the shopper. If your integration cannot provide a usable verification signal, keep the suspect batch out and test the revised flow with known recipients first.
Suppress permanent failures. Klaviyo automatically suppresses profiles after a hard bounce, and Yahoo recommends removing invalid recipients promptly. Check the error before treating every failure as a bad address. A full inbox or temporary deferral calls for different handling. Before restarting, also check that the flow skips shoppers who have since placed an order. Klaviyo's flow filter can do this when your store's order event is syncing. Once those checks pass, review the timing of your abandonment recovery emails.
Google advises keeping user-reported Gmail spam below 0.1% in Postmaster Tools and avoiding 0.3% or higher. Yahoo requires a spam rate below 0.3%. Those figures measure complaints. Use your email provider's bounce report separately to see which messages couldn't be delivered.
How do you keep fake signups out of your newsletter?
Enable double opt-in on the newsletter list. Klaviyo sends a confirmation message and adds the subscriber to the list after they confirm. Trigger your welcome email from that confirmed subscription. A signup form can still generate unwanted confirmation messages during a flood, so protect the form itself too.
For a newsletter or account creation form you control, Cloudflare Turnstile, hCaptcha, and Google reCAPTCHA all let your server evaluate a submission before you accept it. Turnstile tokens expire and are intended for one-time use. hCaptcha requires server-side verification, and a scored reCAPTCHA integration requires you to decide how to handle risky scores and stay within your assessment quota. Check which protection your hosted form already provides. Klaviyo, for example, challenges suspicious visitors on its eligible signup forms automatically.
A honeypot is another option for a form you control. It's a hidden field a simple script may fill in along with everything else. Gravity Forms can mark those entries as spam before running the form's notifications. Review flagged entries when you turn it on. Scripts that recognize the trap can skip it, and a filled field by itself gives you limited evidence.
For the contacts already on the list, build a segment from the suspect form and submission window, then review it against confirmed subscriptions and purchase history before suppressing it. Klaviyo's list-bombing guidance recommends reviewing that segment before acting. A shared email domain or a timestamp alone can sweep up customers who signed up during the same promotion.
Which "bots" should you NOT block?
Google Merchant Center says its StoreBot can open cart and checkout pages and fill in fields such as the shipping address while checking product and checkout details. Google Ads also needs AdsBot to be able to crawl your ad destination. A placeholder-looking name is weak evidence for a storefront block. Google documents how to verify crawler requests against its published IP ranges or with reverse and forward DNS checks. A user-agent string alone can be copied.
Check your uptime tools before you block a burst that arrives on a schedule. An HTTP monitor may just request a URL, while a scripted browser check can click through checkout if you configured it that way. Match the timing and path to your monitor's run history. If a test creates checkout records, give those test checkouts a consistent identifier you can recognize in reports and exclude from recovery sends.
How do you reduce fake checkouts at the source?
If you want to keep guest checkout, rate-limit the checkout creation route where you control it and challenge suspicious submissions where your platform allows it. Cloudflare's rate-limiting rules can target a path on traffic it proxies, with counting options that vary by plan. An IP limit can catch a burst from one source. It can also catch buyers sharing a network, and distributed traffic can come in under the limit. Check every public hostname that can create a checkout. A rule on your custom domain has no effect on a separate platform hostname outside that proxy.
If you're considering a country block, start with whether you accept orders from that market. Where you have customers, use a narrower rule. Even in a market you don't serve, a geographic block only covers requests that appear to originate there, and the next burst can come through another region.
Requiring sign-in or email verification before checkout makes sense when abuse keeps coming through and you can afford an extra step for every guest. Shopify lets you require customer sign-in, which also hides some accelerated checkout options in the cart. Compare paid orders among new visitors before and after the change. If you need to keep guest checkout, you can require verification for the recovery flow alone and leave first-time buyers free to pay without checking their inbox. Weigh the extra step against the rest of your checkout page optimization. If the records show repeated payment attempts, use the separate guide to a card testing attack.
Frequently asked questions
Do fake checkouts cost me money?
An unfinished checkout with no payment taken has no completed sale to refund. The indirect costs can include wasted email sends, a higher contact count on your email plan, and ad spending steered by a misleading conversion signal. If you see a captured payment, check that transaction and the processor's applicable fees separately.
Should I delete bot data from my analytics?
Keep the original counts, mark the affected period, and use a filtered report for decisions. If you can only identify part of the suspicious traffic, label the adjusted rate as an estimate. In GA4, a new data filter applies to incoming events and can't clean up an old spike.
How many bounces are too many?
Klaviyo's campaign and flow deliverability report calls a bounce rate of 1% or less healthy and flags rates above 2% as needing attention. Use your provider's guidance and your flow's usual rate. A sudden cluster of permanent failures from one batch is worth checking even when your account-wide average is still low.
Does double opt-in lower my list growth?
It can lower the number of new subscribers you count because some real people won't take the second step. Make the confirmation instruction visible after signup and compare confirmed subscribers by form. If the form was flooded, the old submission count overstates how many people you had permission to email.