What to fix first when your chargeback ratio is already high
Your ratio is already over the line. You know the number, you have read what happens if it stays there, and none of that tells you what to fix on Monday morning.
Most of what ranks for this query explains the threshold you already crossed: a basis-point line, a warning email, a reserve on your payouts. Useful if you did not know the number existed. Not useful once you are looking at your own dispute list and wondering where to start.
The honest answer is that a high ratio is not one problem. It is however many different problems your disputes are actually made of, and the fix depends entirely on which ones are driving your count. The only way to find that out is to sort last quarter's disputes three ways before you change anything.
Sort last quarter's disputes three ways before you fix anything
Pull the raw list first. Every dispute from the last full quarter, won and lost, every reason code.
Reason code and days to file
Add two columns: the reason code your processor assigned, and the number of days between the order date and the date the dispute was filed. The code tells you what the customer claimed. The clock tells you whether to believe it.
Order value
Add the order value next to each row. A fix that makes sense on a $150 order can be the wrong fix on a $12 one, and sorting by count alone hides that difference.
Likely outcome
Mark whether you won, lost, or never fought each one. This is what separates a filing problem from a fighting problem, and most stores have never looked at all three columns together.
What the reason code and the clock are telling you
Two clusters show up in almost every store's data, and they point at completely different fixes.
A cluster coded fraudulent or unrecognized, filed inside about a week of the transaction, is rarely actual fraud. It is a cardholder who did not recognize the name on their bank statement. Fix the billing descriptor and that cluster shrinks within a billing cycle or two, once new orders start posting under the corrected name.
A cluster coded not received, filed 15 to 30 days after the order, is rarely a lost package. It is a customer who stopped waiting and called their bank instead of calling you. Fix delivery visibility, tracking on every order and a message before the customer has to ask, and that cluster shrinks too.
The table below is the shorthand version. Match your own clusters to it before assuming you already know which fix applies.
| Symptom (reason code, timing) | Likely cause | First fix |
|---|---|---|
| Fraudulent or unrecognized, filed under 7 days | Billing descriptor does not match the store name | Correct the statement descriptor in payment settings |
| Not received, filed 15 to 30 days out | Delivery visibility gap, no proactive update before the customer asks | Tracking on every order, delay messaging sent before the complaint |
| Not as described or defective, filed within days of delivery | Listing, sizing, or quality mismatch | Fix the listing copy, tighten QC on the affected SKU |
| Subscription canceled, clustered around a renewal date | Weak consent at signup or no findable cancellation path | Renewal reminder before each charge, self-serve cancellation |
| Any code, filed the same week as a support ticket or refund request | Customer asked first and escalated before you answered | Faster refund turnaround, pre-filing deflection |
| Any code, won on representment, recurring month after month | Recovery is working, the filing rate is not changing | Representment keeps the money, it will not move the count |
The fixes themselves, the descriptor, the tracking, the reminder, are the ordinary work covered in how to prevent Shopify chargebacks before the bank gets involved. What matters at this stage is matching the fix to the cluster you actually have, not the one you assumed you had.
Only two of these fixes move the ratio itself
Look at the last column in that table again. Five of the six rows change how many disputes get filed. One does not.
That distinction is the whole argument. Visa's acquirer monitoring fact sheet defines the ratio as a count: fraud reports plus disputes, divided by settled transactions. Its two stated exclusions are disputes resolved through a pre-dispute solution and fraud that qualifies for Compelling Evidence 3.0. A dispute you fought and won is not on that list. The full formula and the current thresholds are worth reading in full if you have not already, because that exclusion list is effectively the instruction manual.
That is why the descriptor fix, the tracking fix, and the renewal reminder all belong on the ratio side of the ledger. Preventing the avoidable ones is the direct version of that work. For the disputes that still start, a share can be caught before they are formally filed, in the window when the card networks push an early signal to the merchant ahead of the chargeback itself. Deflecting a dispute before it is filed is built on exactly that window, and it is the only intervention available after the customer has already called their bank that still keeps the dispute out of the count.
Representment sits in a different column entirely. Winning it recovers the transaction amount. It does not touch the ratio, which is the whole argument in why winning a chargeback does not fix your dispute ratio. If last quarter's data shows a healthy win rate next to a climbing ratio, you do not have a fighting problem. You have a filing problem, and the fixes for that live earlier in this list.
Weigh it by dispute value before you commit to a fix
Reason code tells you what to fix. Value tells you what to fix first, and how much effort it earns.
Deflecting a dispute means refunding the order in full, before a chargeback is ever filed. On a low-ticket item you have already shipped, that refund, on top of the fulfillment cost already spent, can add up to more than the fee you were trying to avoid. On a high-ticket order the arithmetic runs hard the other way: the fee is trivial next to what you keep by not losing the whole sale to a filed dispute you might not have won.
That means the fix with the best return is rarely the one generating the most disputes by count. It is the one generating the most disputed dollars, weighted by how likely you were to win if you had fought instead of deflected. Sort last quarter's list by dollar value, not row count, before deciding where deflection earns its keep and where representment is still the better tool for that particular order.
The honest part: a ratio does not move this month
None of this shows up quickly, and I would rather say that plainly than let you find out in September.
Shopify's own chargeback rate looks at a rolling 90-day window, and a threshold breach only clears once the rate falls below 1.00% and stays there for a further 30 days. Visa's ratio resets monthly, but one clean month after a bad quarter does not clear you either, since the trailing months are still in view. A descriptor fixed today changes what gets filed tomorrow. It does not change what already got filed last month, and that is still sitting in the window your acquirer is watching.
Plan on a full quarter before the number moves in a way anyone believes. If it moves faster, good. If you are already close to a threshold, that timeline is also the argument for calling your acquirer now rather than waiting to see whether the fix works first. Ask them, in writing, what they measure, over what window, and what their own internal action line is. It is often tighter than Visa's.
Frequently asked questions
What counts as a high chargeback ratio?
It depends which number you mean. Visa's Acquirer Monitoring Program sets the merchant excessive threshold at 150 basis points (1.5%) in most regions as of April 1, 2026, calculated as fraud reports plus disputes over settled transactions. Shopify separately tracks its own chargeback rate at a 1.00% threshold over a rolling 90 days. These are two different measurements with two different consequences, so check both rather than assuming one covers the other.
Does winning chargebacks lower a high ratio?
No. A dispute is counted the day it is filed. Visa's exclusions cover disputes resolved through a pre-dispute solution and fraud that qualifies for Compelling Evidence 3.0, not disputes you fought and won. A strong win rate protects revenue and says nothing about whether the ratio is improving.
How long does it take to lower a chargeback ratio?
Plan on a full quarter. Shopify's rate is a rolling 90-day window that only clears 30 days after it drops below 1.00%. Visa reassesses monthly, but a single good month following a bad quarter will not clear an acquirer's concern by itself, since the trailing months are still part of the picture.
Should I fix the highest-count reason code or the highest-value one first?
Value first, weighted by how likely you were to win each cluster. A large count of low-value, clearly avoidable disputes and a small count of high-value disputes you were likely to win on the merits require different responses, and prioritizing by row count alone tends to point at the wrong one.
The test I would run this week: pull last quarter's disputes, add reason code, days to file, and order value to each row, and sort by all three at once. If the under-a-week fraud cluster is the largest one, fix the descriptor first. If the 15-to-30-day cluster is larger, fix delivery communication first. Either way, expect the ratio to reflect it next quarter, not this one.