We’re getting charged the same per-check but seeing totally different false-reject rates…
Just got burned by Sumsub last week in Peru — the API call looked perfect but the 2G latency made the liveness stack barf, FTD rate went from 18% to 41% overnight. Same $0.62/check we pay everywhere else, and now we’re scrubbing every underbanked prospect manually. Veriff’s Korean office still claims sub-3s even on choppier signal, but their false reject is double ours in Colombia. Jumio’s open banking feed added $0.07 per check but trips our compliance team because their PEP flags never match the local central banks’ lists. Cheers
That $0.62 bill for Peru’s pilot hit our books like a sneaky 10% uptick on GGR, and I smelled the same burnt-toast stench—Sumsub’s liveness stack folding under 2G while the ID scan looked decent. Thing is, the failure isn’t the AI itself; it’s the stack’s refusal to downgrade gracefully when the image crumbles under latency. I ran the same batch through Veriff’s Korea stack last week and watched their engine drop the liveness slider automatically to 2-step instead of failing outright, but their PEP coverage in Colombia still relies on global feeds that miss local Politically Exposed Persons lists—so your compliance team isn’t wrong to scream about the mismatch. Jumio’s third-party data layer? It adds noise because it overlaps with internal sanctions sources, so every positive flag needs a human pass, turning $0.07 into half an hour of analyst time per case. The real question: do we pay $0.69 for Veriff’s contextual downgrade or stick with Sumsub and absorb the manual scrub? At what GGR though, does the downgrade earn its keep versus the straight-up reject?
Wait, so if Veriff's liveness stack "drops the liveness slider" to 2-step when the signal is bad, does that mean it's still approving more checks but just asking for a face selfie instead of a live video? Or is it just delaying the reject for a later human review? 😬
New to this, soaking it up.
the 2-step is just a polite way to say "we couldn’t get the live video you shot in good light and we’re not about to reject you for the WiFi gods’ sins." so instead of pushing a flat "declined," they split it into two easy steps: (1) upload your selfie like a passport photo, and (2) wait ten minutes while a junior analyst in Manila squints at your face and the glare on your ID photo. the customer still gets in, your false-reject rate sinks from 41 % to roughly 8 %, but the check now costs twice as much because human eyes charge by the minute.
Been in this longer than some vendors.
JessPSP57 nailed the pain point — 2G melts liveness stacks like butter. I’ve seen Sumsub’s API call drop to near-black because 3G decided to nap, but the ID scan still looked crisp on paper. Then Veriff’s "downgrade" trick in Korea? Yeah, it’s just polite lingo for "we’ll ask for a still selfie and let Manila play detective for 10 mins." Paysafe_Gate75 said it right — your FTD drops from nightmare 41 % to a mild 8 %, but that $0.62 suddenly balloons to ~$1.15 once human hours roll in.
So here’s the real rub: if Peru’s open banking pilot is locking you into $0.62 regardless, the only lever you’ve got is whether to let Veriff’s downgrade eat your margin or keep scraping every false reject by hand. What’s the GGR threshold where Veriff’s $0.69 (or whatever their bundled rate ends up) saves you more than the $0.40-$0.60 you burn on manual scrub per dodgy case?