Payment Routing: How to Stop Losing Sales

somewhere between 5% and 15% of the cards presented at your checkout get declined, and a meaningful chunk of those are not real declines. they are issuer risk heuristics, velocity rules, and acquirer-side thresholds that would have approved the exact same card through a different path. payment routing is the infrastructure that gives the transaction that second path.

the second benefit is bigger than the first. routing across two live MIDs means no single processor decision takes your store offline, and no single account absorbs a sudden volume spike that trips a risk review. it is redundancy and load balancing at the same time.

the vocabulary, quickly

• —routing — a rule that decides which MID a given transaction goes to, based on card type, BIN, geography, ticket size, or simple percentage split
• —cascading — when a transaction soft-declines on MID A, the gateway automatically retries it on MID B before the customer sees a failure
• —load balancing — splitting volume by percentage across MIDs to keep any one account within its cap and baseline
• —soft decline — a recoverable decline (insufficient funds at that instant, issuer velocity rule, do-not-honor) that may succeed on retry
• —hard decline — stolen card, closed account, pickup card. never retry these. retrying hard declines is how you build a fraud profile.

that last distinction is the one that separates competent routing from a mess. cascading hard declines inflates your decline rate, annoys issuers, and drags your authorization rates down across every MID you own.

what routing requires

1. 1.two or more MIDs, each honestly underwritten and approved for your actual product by their acquiring bank
2. 2.an independent gateway that supports multiple MIDs and rule-based routing — proprietary processor gateways usually do not
3. 3.consistent billing descriptors across MIDs so customers see the same name regardless of path
4. 4.a rule set defining primary, secondary, and cascade conditions
5. 5.monitoring that shows authorization rate and decline reason codes per MID, not just in aggregate

the gateway decision is the constraint that traps most brands. if you launched on a processor-owned gateway, adding routing later means a checkout rebuild and, worse, losing stored tokens on every subscriber. choose an independent gateway before you need one.

a routing configuration that works

you do not need anything exotic. this simple structure covers most of the value.

1. 1.split baseline volume roughly 70/30 between primary and secondary so both accounts stay active and neither goes dormant
2. 2.cascade soft declines from primary to secondary once, immediately, with no customer-facing retry prompt
3. 3.never cascade hard declines — fail them cleanly and show a helpful error
4. 4.route high-ticket orders to whichever MID has the better authorization rate on large amounts, which you will only know by measuring
5. 5.define an outage rule: if the primary returns gateway errors or timeouts above a threshold, shift 100% to secondary automatically
6. 6.cap retries at one per transaction per MID to avoid duplicate authorizations

then measure. authorization rate by MID, by card brand, and by ticket band. the differences between two acquirers on the same traffic are often several percentage points, and several points of authorization rate is real revenue you were silently losing.

the line you do not cross

routing is legitimate infrastructure used across every serious ecommerce vertical. it becomes fraud the moment it is used to hide what you sell. the distinction is simple and absolute: every MID in your routing setup must be underwritten by a bank that knows you sell peptides, under your real business, with accurate descriptions.

• —legitimate: two MIDs, two banks, both approved for the same real business and product, split for redundancy and performance
• —fraud: routing peptide volume through a MID approved for coaching, apparel, or digital products
• —fraud: shell entities created to obtain MIDs that would be denied to the real business
• —fraud: splitting a single order across MIDs to stay under monitoring thresholds
• —fraud: continuing to process after a termination by moving volume to an account obtained through misrepresentation

the second list is transaction laundering. it is the fastest route to permanent MATCH placement and it carries real legal exposure. build the honest version — it works better anyway, because you can call your bank when something breaks. not legal advice; consult your attorney.

subscription routing is its own problem

recurring billing complicates routing because tokens are usually MID-specific. a card tokenized on MID A generally cannot be charged through MID B without re-tokenization. plan for it.

1. 1.use a gateway-level vault rather than processor-level tokens wherever possible, so tokens survive a MID change
2. 2.keep each subscriber's rebills on the MID that originally authorized them unless you have to migrate
3. 3.if you must migrate, batch re-tokenize through the gateway and notify customers before the next rebill
4. 4.keep descriptors identical across MIDs so a migrated subscriber does not see a new name and dispute it

the routing setup you never had to think about is the one that saves you at 2am when a processor's status page goes yellow.

what routing does not fix

routing recovers marginal declines and provides failover. it does not fix a genuinely bad account, a rising chargeback ratio, or a site full of claims that will get you dropped. if your fundamentals are broken, adding a second MID just means you now have two accounts heading toward termination.

• —fix chargebacks with fulfillment speed, clear descriptors, and easy refunds — not with routing
• —fix declines caused by aggressive fraud filters by tuning the filters, not by cascading
• —fix underwriting mismatch by correcting your site and telling your processor, not by moving volume

get the fundamentals right and routing becomes the multiplier it is supposed to be: a few points of recovered authorization rate, spike absorption, and an outage plan that runs itself.