The 2.5–3% you're paying that you don't think about

A Habesha restaurant owner in DC opens her processor statement on the first of the month. The sales total looks fine. Then she scrolls down to “fees.” It’s $5,400.

Her rent is $5,000.

The fees outpaced the rent again.

This post is for owner-operators who already know that number to the dollar — and for the customers who don’t yet realize that’s where their card swipe goes.

What’s actually inside that ~3%

The all-in cost of accepting cards is called MDR — merchant discount rate. For US small-business card volume, the industry average is somewhere between 2.5% and 3% (mostly depending on whether your traffic is in-person or card-not-present, and what kind of cards your customers carry).

That single MDR number breaks into four pieces:

Piece	Typical share	Goes to
Interchange	1.5–2%	The customer’s card-issuing bank
Network assessment	~0.13%	Visa or Mastercard
Processor margin	0.5–1%	Square, Toast, Stripe, your terminal vendor
Other fees	varies	Per-transaction, monthly, PCI compliance, chargeback admin

The largest piece — interchange — does not go to your processor. It goes to the bank that issued your customer’s card. That bank may be Chase, Bank of America, Capital One, or any of hundreds of others. It has nothing to do with the Habesha community, the corridor, or your business. It is the largest single line item in your fees, and it is the one most owners don’t realize they’re paying.

The processor margin — the slice your terminal company actually keeps — is smaller, but it is the piece you have any leverage on if you switch processors. In practice, switching processors usually moves the dial by a fraction of a percent. The networks set the floor.

What it actually costs at three real shop sizes

Numbers feel different when they’re attached to a specific business. Three illustrative shop sizes — names of the categories, not specific shops:

Small café — $30K/month revenue

Card share of revenue: ~80%
Card volume: ~$24,000/month
MDR @ 2.8%: ~$670/month
Annual: ~$8,000

That’s a kitchen oven. Or a month and a half of payroll for a part-time employee. Or your whole insurance bill for the year.

Mid-size restaurant — $100K/month revenue

Card share: ~80%
Card volume: ~$80,000/month
MDR @ 2.7%: ~$2,160/month
Annual: ~$26,000

That’s most of a car. Or — for many DC and Atlanta Habesha owners — the equivalent of three months of rent.

Busy grocery or large restaurant — $250K/month revenue

Card share: ~80%
Card volume: ~$200,000/month
MDR @ 2.7%: ~$5,400/month
Annual: ~$65,000

For a lot of US Habesha owner-operators, this is more than rent. It is comparable to a manager’s full salary. It is not a small line item. It is a major one, and it is a line item where the entire amount leaves the community.

Now multiply across the 347 Habesha businesses live on Mela today, the much larger universe of unaffiliated Habesha businesses, and the dollars start to look like a corridor of their own.

What that money buys you (and we are not pretending otherwise)

It is fair to be specific about what you’re getting in exchange for those fees, because it is not nothing:

Fraud-loss coverage. If a stolen card runs at your terminal and the customer disputes, the loss usually does not fall on you.
Dispute infrastructure. Chargebacks have a process, even if it doesn’t always go your way.
Cardholder rewards programs. A meaningful share of customers use cards because of points and miles. The interchange you pay funds those rewards.
Predictability. The terminal works. Funds settle. The system is mature.

So this isn’t “card networks are evil.” It is “card networks cost what they cost, and what they cost might be more than your rent.” Different framing.

The relevant question is whether all of your customer payments need to run on that rail.

What QR-to-USD-wallet replaces — and what it doesn’t

The Mela merchant rail is QR-based. Here is the mechanic:

The customer opens the Mela app and scans your QR (printed at the till, or generated dynamically per transaction).
The amount, the merchant ID, and the cashier (if you have multiple) are encoded.
The customer’s Mela USD wallet sends USD direct to your Mela merchant USD wallet.
There is no interchange. No assessment. No processor margin.
Funds settle to your USD wallet. From there you move to your bank, hold for payroll, or spend on supplies.

The honest framing: Mela is not a replacement for your card terminal. It is a parallel rail for the customers who choose to use it. You keep your existing processor for everyone else. Over time, as more A1 customers are on Mela in your metro, more of your traffic moves to the rail you don’t pay 3% on.

The math we are pitching is not “save 100% of card fees” — that would be a lie. The math is whatever percentage of your traffic moves to Mela costs you 0% in network fees. If it is 20% of your customers in year one, you save 20% of the line we calculated above. If it is 50% by year three, you save 50%.

Twenty percent of $65,000 a year is $13,000. That is not a marginal number for any small business.

What’s live in V1 — named honestly

If you talked to a vendor who pretended their V1 did everything, you should be skeptical. Here’s what Mela’s V1 actually does and doesn’t:

USD only. The customer pays in USD from their wallet. You receive USD. No multi-currency yet.
$5,000 cap per transaction. Fine for cafés, restaurants, groceries, salons. Flag this if you sell jewelry, premium clothing, electronics, or anything single-receipt above $5K. We will tell you the same thing on a discovery call.
Single-location merchant accounts. Most US Habesha businesses are single-location anyway, but if you operate two or three locations, the pilot has constraints we’ll walk through.
Refunds ship in V1.1. This is the most painful constraint and we say so up front. If your business sees frequent refunds (some retail categories), this matters more than it does in food service.
Customer must have a Mela account. Self-evidently. We’ll show you the live US user count in your metro at pitch time so you can decide whether the customer side is real for your shop yet.

If a constraint above is a deal-breaker for your business, we want to know on call number one, not call number five. Saying that in writing is part of our trust posture.

What “founder-led pilot in DMV” actually means

The strategy is to go DMV first, countrywide second. Specifically: founder + team phone outreach, no sales reps in 2026. The reasons:

Habesha business clusters are concentrated. The founder’s network in DMV is dense. Phone + warm-intro motion is more efficient than rep payroll at pilot scale.
Reps add a layer between the founder and the operator. The product is V1. The founder needs to hear, directly, what’s working and what’s broken.
After Phase 1 — once the math works in DMV — we re-evaluate and decide whether to add reps for countrywide.

Concretely: if you operate a US Habesha business and want to talk, you reach a person on the team or the founder, on a phone call or over coffee. There is no demo-environment funnel. There is a conversation.

That $5,400 statement line, again

Back to the restaurant owner from the opening. Half of that monthly fee, over time, can be a kitchen renovation. Or your cousin’s tuition. Or the down payment on the second location.

We are building the alternative. It is not perfect in V1. It is honest about its constraints. And it does not cost 3%.

If you run a US Habesha business and want to talk, hello@melafinance.com reaches the team directly.