How to Choose a Crypto Payment Gateway

What should you decide before looking at providers?

Two things, and both of them eliminate more candidates than any comparison table.

Your settlement currency. Taking crypto and keeping it is a treasury position. Taking crypto and converting to fiat on receipt is a payments decision. These are different products with different providers behind them, and deciding late means redoing the evaluation.

Your volume shape. A business taking two hundred payments of thirty dollars has almost nothing in common with one taking three payments of twenty thousand. Fixed per-transaction components dominate the first and are irrelevant to the second, and providers are priced for one or the other whether or not they say so.

What disqualifies a provider?

Run these before comparing anything, because a provider that fails one of them cannot be fixed by a better rate.

Does it support the assets your customers actually hold, on the networks they actually use? Stablecoin buyers overwhelmingly use the cheapest network available to them. A provider supporting Tether on Ethereum only, while your customers hold it on Tron, supports a product they will decline to use.

Will it accept your business? Providers differ sharply on verticals and several publish nothing at all about it. Ask before integrating rather than after.

Can it settle where you bank? Global asset lists are longer than the list live for merchants in any given country, and the gap is never advertised.

What should you ask, in writing?

Send the same questions to all three on the same day. Written answers are comparable and callable answers are not.

• Do you ever hold our funds, and if so where, and are they segregated?
• Which assets and networks are live for merchants in our specific countries?
• Does the person paying face an identity check or an account signup?
• What is the payout schedule, the reserve percentage and the hold period?
• What is the all-in cost including conversion spread and payout fees?
• What happens on underpayment, on late payment after expiry, and on a wrong-network send?
• Can one developer reach a test transaction in a day using public documentation only?

The last one you can answer yourself without asking, and you should.

Which test actually settles it?

A live transaction with each finalist, including a deliberate underpayment of about 2%.

It costs a few dollars in network fees and takes fifteen minutes. It tells you how the provider handles the most common real-world failure, whether the webhook fires when you expect, how long confirmation actually takes against what the documentation claims, and what the customer sees at each step.

Almost nobody does this before signing, and it produces more useful information than the entire paper evaluation preceding it. Underpayment in particular is the least documented behaviour in the whole category, and it happens inside your first hundred orders.

Where should you start?

The catalogue carries the same nineteen fields for every provider, including what each one declines to publish. If your requirement is specific, the shortlists filter that catalogue by job. If you disagree with the order, the weights are published and adjustable.

What the shortlist should look like on paper

Three providers, one page each, same fields. Rate, conversion, payout, assets, networks, verification, settlement currency, settlement schedule, reserve. Fill what they publish, mark what they do not, and send the gaps as questions.

That document is the actual deliverable of an evaluation, and it is what makes the decision defensible to somebody who was not in the conversations. It also makes the disclosure difference visible: a provider whose column is mostly blanks has told you how the relationship will go.

Reading the answers you get back

Watch for three patterns. A factual question answered with positioning is the most informative failure, and it recurs. A range where you asked for a number usually means the number depends on something they would rather discuss later. A commitment with no notice period attached is not a commitment.

None of these are disqualifying on their own. All three are worth noticing, because the way a provider handles procurement is a fair sample of how it handles everything else.

What to do with the winner

Integrate in test mode and run the live checks from the testing guide before switching traffic. Keep the runner-up’s account open and unintegrated for a while, because the cost of that is nothing and the cost of rebuilding under pressure is not.

Then diarise a re-check. Pricing pages in this category change without announcement, and the terms you agreed are worth confirming against the published ones a couple of times a year.

Where to go next

The catalogue is where the shortlist comes from, and the methodology lets you reorder it against weights you set rather than ours. The testing guide covers the checks to run before switching traffic, and the fees guide covers turning published figures into a monthly number you can put in front of a finance function.

The mistake that costs most

Evaluating on the criteria that are easy to compare. Processing rate is published, numeric and identical in meaning across providers, so it dominates spreadsheets regardless of how little it decides.

Settlement terms, reserve, review behaviour and vertical position are none of those things, and each of them can end the relationship or tie up more capital than the fee difference across a year. They are harder to compare precisely because they are not published, which is exactly why they deserve the time.

A useful discipline: before comparing any numbers, write down what would make you leave a provider after twelve months. Almost nothing on that list will be the processing rate, and the list is a better evaluation framework than any published table including this site’s.