Most merchants track fraud losses closely. Far fewer track the opposite mistake: Ecommerce false declines, when real customers are turned away even though they posed no fraud risk. The cost is largely invisible because it rarely appears on a fraud report. Yet for many brands, it exceeds the fraud they’re preventing.
Understanding where false declines occur — and how to measure and reduce them — can help you recover lost revenue, improve customer lifetime value (CLTV) and strike a better balance between fraud prevention and conversion.
TL;DR
- The term “false decline” (or “false positive”) indicates that an order was erroneously flagged as fraudulent by a merchant or issuing bank.
- The global average false declines rate in ecommerce: 1.51% of sales (Datos Insights).
- Projected global merchant losses from false declines exceed $231 billion in 2026, rising to almost $265 billion in 2027.
- Merchants can measure false declines by tracking successful retry purchases and customer complaints after declines.
- Five steps to reduce false declines: (1) Analyze and categorize declines, (2) retune risk thresholds, (3) use accuracy-built fraud technology, (4) take calculated risks safely, (5) account for AI agents at checkout.
What is a false decline in ecommerce?
False declines are valid orders that a merchant or a bank declines for fear of fraud. Some merchants refer to these orders as false positives — the order tested “positive” for fraud when it wasn’t — or “customer insults,” a nod to the damage they do to customer relationships.
What is the average false decline rate?
The global average false decline rate (or average of false positives) is 1.51% of ecommerce sales, per Datos Insights (formerly the Aite-Novarica Group).
Additionally, the Merchant Risk Council’s 2026 Global Payments and Fraud Report found that most merchants’ own averages run as high as 10% of orders: 38% of merchants put their false decline rate between 2% and 5% of orders while two-thirds of the total merchants surveyed said their false decline rate falls somewhere between 2% and 10% of orders. Notably, the false positive rate tends to climb as merchant size increases.
This raises the question every merchant eventually asks: How much revenue am I losing to false declines, and how do I measure it?
How much money do merchants lose to false declines?
Globally, false declines are projected to cost ecommerce brands over $231 billion in 2026, rising to almost $265 billion by 2027.
But those are global forecasts… What could false declines cost you? Going back to the average false decline rate of 1.51% of orders, say you’re doing $10 million in annual ecommerce sales. At that rate, roughly $151,000 worth of legitimate orders would be wrongly declined in a single year.
Unfortunately, the costs don’t stop there. Among loyal customers — those who have previously had at least three orders approved — Signifyd found that a false decline is followed by a 65% decline in the number of orders placed by that customer and a 16% decline in their average order value. Moreover, 27% do not return to the retailer at all — meaning you don’t just lose that one sale, but a lifetime of potential sales from a repeat customer.
How do you measure false declines?
It’s tricky to get an accurate false decline rate, partly because only 74% of U.S. merchants and 61% of U.K. merchants track it as a key performance metric at all, according to Datos Insights data. The core problem is that a false decline is a good order you rejected, so it never became a sale you can observe directly. This means the number is always an estimate rather than an exact count.
Teams that do track false declines typically use the following strategies to identify which declines were actually legitimate:
- An order that goes through on a second attempt and doesn’t result in a chargeback
- A blocked order followed by an inbound complaint to customer service
Some also purposely approve a small number of orders that appear fraudulent to test if their systems are correctly identifying fraud.
What causes false declines?
False declines can happen during either bank authorization or merchant approval. Some legitimate transactions are declined by the issuing bank before you ever see them. Others are authorized by the bank but not approved by the merchant’s fraud system because the order is mistakenly identified as risky.
Common causes include:
- Ruleset bloat: Legacy rules-based systems become increasingly restrictive as new fraud rules are added, increasing false positives over time.
- Human error in manual review: Manual review teams under high volume or with limited context can mislabel legitimate orders as risky.
- Missing context: Single-factor signals (billing/shipping mismatch, new shipping address, rapid reorders) without broader behavioral and historical context can be misleading.
- New behavior channels: Authorized AI shopping agents or new legitimate behaviors can resemble fraud to outdated systems.
What are the problems with rules-based systems?
Legacy, rules-based systems automatically decline an order whenever it matches a preset pattern of red flags. The trouble is that these rulesets keep growing: Every new fraud pattern adds another rule, an issue known as fraud ruleset bloat. Over time the system turns more restrictive, approving fewer and fewer orders and generating false declines whenever it errs on the side of caution. On top of that, merchants relying on rules and manual review are stuck constantly retuning them to keep pace with fraudsters.
How are AI-driven systems different?
AI-based systems and machine learning models for fraud prevention avoid the issues that accompany rules-based systems by making ship-or-don’t-ship calls from vast amounts of transaction, behavioral and historical data.
“There are over 1,000 features that the model takes into consideration and thousands of data elements within those features,” — Jasal Motiram, manager, enterprise customer success at Signifyd.
No single factor decides an order; risk comes from the combination, for example whether the item is a frequent fraud target, the account or shipping address is brand new, the CVV doesn’t match or an order is retried within seconds, faster than a human could retype the details.
This gives the model far more context per order than any ruleset. But fraudsters know many of these signals and adapt constantly, which is why human expertise, guiding how the models are designed and trained, remains essential to keeping them effective.
How to reduce ecommerce false declines
To lower ecommerce false declines, you need to do five key things:
- Understand your declines
- Retune your system
- Lean on better technology
- Take calculated risks safely
- Account for AI agents at checkout
1. Analyze and categorize your declines
You can’t fix what you don’t measure, and simply tracking false declines is more than many merchants do. Pull a sample of declined orders (say, everything over $500), review how many were actually legitimate and categorize your insults.
Did the customer call in? Did your risk team later judge it a good order? Understanding where you’re committing insults shows you exactly what to fix, whether it’s over-scrutiny on high-value items or too much weight on a single factor like a billing-shipping address mismatch.
2. Retune your risk thresholds
Fraud detection is a balance: Screen too loosely and fraud gets through; too tightly and you turn away good customers. If your review shows a category being wrongly declined, like those over-$500 orders, adjust how the system weighs it. High-value orders carry risk, but it’s often worth accepting a little more to avoid sending a high-spending, legitimate customer away empty-handed.
3. Use technology built for accuracy
Manual review, verification calls, and third-party card checks all reduce declines but frustrate customers and add delay. AI-driven platforms like Signifyd’s Commerce Protection Platform instead draw on vast transaction, historical and behavioral data to read the identity and intent behind each order, sorting good from bad instantly without insulting legitimate buyers.
Philips saw this firsthand: When they began selling direct to consumers online, only about 40% of orders were converting, partly because good ones were being turned away. After adopting Signifyd, their conversion rate rose to 75%, with a goal to climb over 90%.
4. Take calculated risks
The most systematic approach is to deliberately approve a few orders you’d normally decline, then watch for chargebacks. No chargebacks means you can safely loosen those controls. It’s what vendors like Signifyd do, except Signifyd absorbs the financial risk, paying the full cost of any approved order that turns out fraudulent.
“We take calculated risks, and we continue to refine the system. If there are no chargebacks, then we adjust the threshold so that we can open up for more approvals,” Motiram said. Doing this alone is expensive and risky, which is why many top retailers hand it to a partner that will reimburse the loss.
5. Account for AI agents at checkout
A newer source of false declines is agentic commerce, where AI agents shop and check out on behalf of customers. The problem: An agent’s behavior (rapid, sequential, cross-category orders) can look identical to a compromised account to a rules-based system, so legitimate agent-led or, in some cases, agent-placed orders get wrongly declined.
To avoid turning away this growing channel, pass agent metadata (which agent, what permissions, what session) through to your fraud tools, and verify both that the agent is authorized and that the human behind it is legitimate. Fraud tools that score behavior rather than rely on static rules are best positioned to tell an authorized agent from an adversarial bot.
Turn false declines into recovered ecommerce revenue
The instinct when fraud ticks up is to tighten the filters, but every notch tighter swaps a fraud loss you can see for a false decline you can’t, and the false decline usually costs more.
Turning those losses back into revenue means decisioning smart enough to tell a good order from a bad one, not just a stricter rule. Signifyd’s Commerce Protection Platform makes that call in real time and guarantees the outcome, so approving more good orders never means absorbing more risk — learn how it works here.
Want to dig deeper into how you can reduce false positives, safely increase approval rates, optimize manual review queues and recover millions in falsely declined valid orders? Then check out our free video masterclass on false declines and revenue optimization.
