Two directions in the same month
In July UPI carried 23.66 billion transactions worth 29.88 trillion rupees; volume rose 4.1 percent and value 3.3 percent on the month, with annual gains of 22 percent and 19 percent. IMPS handled 364 million transactions worth 7.12 trillion rupees; volume rose 3 percent and value 5 percent on the month, while annual volume fell 24 percent and annual value rose 13 percent.[1]
The two series moving in opposite directions on annual volume, read together with value rising on both, is consistent with small-value traffic shifting from IMPS to UPI. Average transaction sizes derived from the published figures make that measurable: about 1,263 rupees per UPI transaction against about 19,560 rupees on IMPS. A mechanism other than migration delivers the same fall: banks reclassifying flows, or a large aggregator moving batch processing to another system, would pull IMPS volume down without any household changing rail. Monthly totals are not enough to separate the two possibilities.[1]
How much has to keep working
UPI daily averages were 763 million transactions and 963.83 billion rupees. In the same month FASTag volume fell 4 percent to 347 million and AePS volume rose 3 percent to 100 million; the other rails inside the system stay one to two orders of magnitude below the size needed to take displaced retail traffic. There is no comparable backup under the same roof if there is an outage. The counter-reading matters too: cash and card networks outside NPCI's data could absorb part of a disruption, and availability and downtime are not given alongside these monthly totals.[1]
The threshold that would test this is clear. If UPI's transaction value exceeds 29.88 trillion rupees in the August data while IMPS's annual volume keeps falling, the shift is confirmed as continuing displacement. The critical data are not published: unless NPCI adds availability and downtime alongside the monthly totals, the public has nothing measurable on the resilience of a rail at this scale.[1]