How long business onboarding takes: what 92 fintechs, banks and payment platforms tell applicants

Sweat AI is an AI-native BPO for banks and fintechs, starting with back-office workflows like KYB, onboarding and fraud reviews. This study looks at one number every onboarding team publishes, whether it means to or not: how long a business applicant is told to wait for review.

On 2026-09-30 we read the public help centre, docs or FAQ pages of 92 companies across six segments: payment service providers, business-banking fintechs, traditional banks, crypto and stablecoin firms, marketplaces and platforms, and banking-as-a-service providers. We recorded what each one tells a business applicant about review or approval time, word for word, with the source URL. The result is 100 rows, because some companies publish more than one relevant time (Stripe has three relevant pages, Twilio four statements). Headline numbers count each company once.

We aimed for 100 companies. We could verify 92 from first-party pages, so n = 92.

Headline findings

• 61% of companies (56 of 92) state a review time. 39% (36) state none. Traditional banks are the quietest: 14 of the 18 banks we checked give no approval time on their business account pages.
• The median stated upper bound is 3 business days, across the 44 companies whose statement can be expressed in business days. A quarter promise 1 business day or less. Three go past a week: Twilio's manual brand review (7 business days or more), Wise's additional business review (up to 10 working days) and Meta's business verification (up to 14 business days).
• Crypto firms quote the longest waits and often in weeks. Bitpanda says 2 to 4 weeks, BitPay 1 to 2 weeks for its KYB and enhanced due diligence process, Bitfinex up to one week. Half the crypto firms we checked (8 of 16) state no time at all.
• Almost nobody talks about weekends. One company of 92, Holvi, says in plain words: "Please note that we operate on weekdays (excluding weekends and national holidays)." Another 32 state their time in business or working days, which implies weekends don't count without saying so.
• One firm publishes an overnight window. Unit's docs give a manual-review SLA of two hours between 8:00am and 8:00pm EST, and state that reviews landing between 8:00pm and 8:00am EST are "completed by 9:00am EST the following day." No other page in the sample says what happens to an application submitted at night.
• Most stated times are hedged. 36 of the 56 statements use words such as "usually", "most", "typically" or "can take". Only a minority read as a commitment.

Summary by segment

Upper bounds are normalised to business days only where the text allows it: business or working days as stated, hours divided by 24 and rounded up, minutes and "same day" as 0. Times stated in unqualified days or weeks are left out of the median and reported separately.

The bank median rests on three statements and describes account opening, not review: Bank of America says "In many cases, a business account can be opened the same day, especially when applying online", RBC says select accounts can be opened online "in less than 15 minutes", and Metro Bank's same-day opening is for walk-in customers in stores. The crypto median rests on three statements (Bitstamp, CoinJar and Paybis); the firms quoting weeks sit outside it.

What the pages actually say

The fastest statements come from business-banking fintechs that decide most applications automatically. Lili says "Most applications are reviewed and decided within minutes after you submit them." Tide "may approve your application within minutes if you pass our automatic eligibility checks." Countingup says "Most people receive a decision on their application within the same business day."

The middle of the distribution is 1 to 3 business days, and it is crowded. Mercury: "usually within 0–2 business days." Relay: "Most Relay applications are reviewed within 1–2 business days." Holvi: "In most cases this takes between 1–3 business days." Among payment providers, SumUp says verification "typically takes 1-3 business days", PayPal says the review "usually takes 3 business days", and Stripe asks applicants to "allow us 3-5 business days for Stripe to review your documents."

The long tail belongs to platforms and crypto. Meta: "A decision on your verification submission may take up to 14 business days." Google Ads: "It can take up to 5 business days to review your task. In rare cases, review time can take up to 30 days." Rapyd: "Our compliance review and approval can take up to 30 days." Bitpanda: "the review process can be completed within 2-4 weeks."

Some pages decline to give a number at all. Monzo tells business applicants "we can't tell you for sure how long that'll be." Gemini's article titled "How long does it take to get verified?" gives no time. Binance's entity verification guide, as served on the capture date, states no review time either, although search-engine snippets still quote an older version with a figure. That gap between cached and live text is one reason we fetched every page ourselves.

What delays a review, according to the companies

33 of the 92 companies name something that slows or extends review. The causes repeat across segments:

1. A request for more information. This is the most common. Mercury "may need additional documents or information before a decision can be made." Wise's up-to-10-working-day figure applies to that second review, which happens "if we can't confirm all your details during our initial review, or if your business operates in certain industries."
2. Document quality and mismatches. Relay lists four reasons an application runs past its standard timeline: a document "missing, blurry, or does not match your business records", owner identity that "could not be verified automatically", a beneficial owner who "has not accepted their invite", and an industry that "requires additional screening." SumUp asks applicants to make sure "the name on your SumUp account matches your bank account to avoid processing delays."
3. Stale or foreign-language documents. Bitpanda wants a company register extract "not older than 6 weeks". CoinJar requires certification "within the last 90 days". Binance recommends "a notarized translation into English to ensure a swift review process."
4. Registry lag. Squarespace warns that "The IRS can take up to two weeks to register new Employer ID Number (EIN) and Tax ID Numbers (TIN) in their permanent records", so a brand-new company can fail an automated EIN match through no fault of its own.
5. Business model questions. Stripe may need "more details on the goods and services you sell or provide."

Each of these shows up in a review queue as an open request for information (RFI) and a paused clock. See the glossary entries on requests for information and KYB manual review queues.

Method

• Sample. We started from candidate lists per segment (well-known providers in the US, UK, EU, Canada, Australia and Singapore) and kept every company for which we found a first-party page about business or merchant onboarding review. A company whose relevant page states no time is kept, with an empty quote, so the no-time share is measured rather than hidden. Companies for which we found no relevant first-party page at all were skipped and are not in the denominator.
• Sources. Only the company's own domains: help centres, developer docs, FAQ and product pages. Search engines were used to find pages, never as a source.
• Capture. All pages were fetched on 2026-09-30 with curl, with a headless Chromium browser for pages that render in JavaScript (including collapsed FAQ sections), or through the help centre's own public article API when the HTML sat behind a bot wall. One page (Helcim) could only be read through a web-fetch tool, and the dataset flags it as lower confidence.
• Verbatim check. Every quote in the dataset was checked by script against the saved text of the page it came from. 99 of 100 row quotes matched exactly; the Helcim quote could not be checked this way.
• Normalisation. Business, working and banking days are copied as stated. Hours are divided by 24 and rounded up. Minutes, "instant" and "same day" count as 0. Unqualified days and weeks are not converted, because we cannot know whether they include weekends. For open-ended statements ("seven business days or more") the stated floor is used.
• One company, one vote. Where a company has several relevant pages, we mark one row as primary for the headline statistics and keep the rest in the dataset.

The dataset has one row per statement with the company, segment, product, country, verbatim quote, raw and normalised values, weekend language, delay causes with their own quote, source URL, fetch method and capture date. It is free to reuse under CC BY 4.0.

Limitations

• Stated is not actual. These are the numbers companies publish to applicants. They say nothing about measured turnaround, and most are hedged. A firm that states no time may be faster than one that states two days.
• Different clocks. Some times start at submission, some after all documents arrive, and some cover only a step (Apple's two business days cover D-U-N-S data reaching Apple, not the enrollment review, so we excluded it from the median). The quote column shows exactly what each page says.
• Selection. The sample leans toward companies with public, crawlable help centres. Banks that explain onboarding only in branch or by phone look quieter than they may be in practice.
• Small cells. Segment medians for crypto, BaaS and marketplaces rest on 3 to 5 normalisable statements. Treat them as indicative.
• Point in time. Help pages change without notice. Every row carries its capture date.

What it means for onboarding teams

The benchmark applicants see is 1 to 3 business days. That is what most fintechs and payment providers publish, and a business that has applied to several providers will compare you against it. If your real cycle is longer, stating it honestly on a help page beats a silent queue, and Monzo and Gemini show the cost of not stating one: the question gets asked anyway, through support.

Business days hide the weekend. Nearly every stated time excludes weekends by definition, and only Holvi says so. An application submitted on Friday evening against a "1 to 2 business day" promise can wait four calendar days before anyone opens it. Unit's overnight rule is the only public example in our sample of a provider saying what happens outside office hours. If your queue stops on Friday night, say so; if it doesn't, that is worth saying too. Our 24/7 onboarding review page describes how outside coverage for nights and weekends works.

Most delay is waiting on the applicant. The causes companies name are nearly all requests for something missing, unclear, stale or unverifiable. Asking for everything at once, in one consolidated list, is the cheapest speed-up available. Publishing the document freshness rules up front (6 weeks, 90 days, a certified translation) prevents a second round trip.

Define your own clock before you publish one. Decide whether the time starts at submission or at a complete file, whether it pauses during an RFI, and whether it counts calendar or business days. The onboarding SLA entry covers how to write that definition, and the onboarding backlog calculator shows what a given volume and staffing level does to wait times.

If your team wants applications reviewed while it is offline, with the evidence and a recommendation ready for your decision when it logs in, Sweat AI works KYB and onboarding queues around the clock. Your team keeps the final approval.

Download the dataset: Stated business onboarding review times, 92 companies (CSV) · 100 rows · CC BY 4.0