Can anyone show real A/B data proving Praxis’ cascading/smart-routing ROI vs PaymentIQ?

Ever seen a rollback where the “smart” math melts into “where did those 12 bps vanish”? That’s exactly what happened to us on our CZ wallet when we forced Praxis into failover-only mode last August—after their sales deck swore we’d see sub-0.60 % blended in a week. We watched 94.7 % approval at 0.87 % blended flip to 92.4 % approval and 0.75 % blended inside five days. The delta wasn’t theory; it was Visa’s cascade feed artificially raising first-pass decline masks to create the “smarter” numbers. Their routing table pushes anything with even a 0.3 % auth rate straight to fallback PSPs, which look expensive because they’re tier-1 incumbents. Meanwhile, PaymentIQ’s sub-MID clusters in CZ catch the same transaction at half the surcharge and ten points higher on auth, but Praxis’ engine can’t differentiate the MID profile from the BIN profile, so it routes blindly. Our finance sheet told the story in black and white: GGR dropped 1.6 % the week we onboarded Praxis because the rollback surged—high-risk declines don’t count toward volume-based discount tiers, so the blended rate shrank only on paper. When we peeled back the dashboard, we saw 11 % of every approved first-pass attempt now hits a 2 % fee tier from Worldline, whereas PaymentIQ’s gateway tiers sit at 0.9 % for the same card bin. That one tier jump alone eats every bps they claim. Add the rolling reserve: CZ gives us 1 % rolling after 90 days, same as Malta, so the promised “reserve arbitrage” arbitrage LeeCrypto is chasing is already capped. Unless you’re routing into crypto PSPs—where Praxis touts its “smart”—the fixed-cost base doesn’t move. For our 12 M/month CZ wallet, we flipped the Praxis toggle off yesterday; approval went from 92.4 % back to 94.2 %, blended from 0.75 % back to 0.88 %, and the delta wiped out the promised savings. So much for smart.

What does "first-pass decline masks" even mean? Like, is that Praxis pretending some declines never happened by routing them straight away to fallback PSPs so the approval rate looks higher in their dashboard? 😬

rolled his eyes when Praxis dumped me into their “cascade failover” in november because sales kept barking about “first-pass decline masks” like it was some holy rogerian maneuver. think of it like this: you fire 1,000 visa transactions at a skin; the gateway immediately sees 200 that smell funny—soft decline, velocity flag, whatever—so it coughs out a 403 straight back to the player. now picture Praxis taking those 200 soft declines and shovelling them straight to some tier-1 fallback PSP instead of letting the gateway hit the bucket with an outright deny. the psps approve maybe 120 of those 200, but Praxis sweeps the messy 80 under the rug and slaps a “first-pass approval” sticker on the remaining 880 in their dashboard. the customer thinks he’s playing, you think the approval meter just jumped 8 %, but all Praxis really did was postpone the failure to the next layer—where the fee is 2 % instead of the planned 0.9 %. that’s the mask: they hide the soft decline in the routing table so the top-line number stays pretty while the surcharge balloons later.

Yeah I’m literally staring at our Praxis dashboard for the CZ skin right now and it’s doing the exact same sleight-of-hand. First-pass approvals printed 94.5 % on Friday, but when you filter for “actual gateway accepts before routing,” it’s only 86 %. The extra 8.5 % are those soft declines Praxis is kicking down the chain to tier-1 PSPs that charge 2 % rolling reserve at 1 %—so the blended fee drops from 0.98 % to 0.71 % on paper, but the moment we pull the full PSP-level report the 1.3 % surcharge on every hidden decline wipes out every basis point they promised. ClassicGuy’s right: they’re masking, not routing. Anyone here actually running the paid mid-month reconcile on Praxis? How do you stop the cascade table from artificially inflating approval numbers while the finance team still sees the real blended fee?