I began the review with a strict rule that every statement about fees, recovery, and reporting had to produce a number I could check against our own records. Our current process left gaps in recovered payment tracking and in the visibility of costs across processors. Clear transaction histories, predictable monthly costs, and controls that would hold up under audit formed the baseline. I compared the available documentation against three months of our historical volume before any further discussion.
Fee Structure and Cost Predictability
The stated rate is $99 per month plus 0.65% of successfully processed volume. I took one full month of our cleared volume and calculated what paymentkit would add to our existing payment processing expenses. Since paymentkit’s fees are separate from the charges imposed by individual processors, I also considered whether better routing could reduce our blended processing costs enough to offset the additional expense. I asked for the precise line items that appear on a sample invoice and confirmed that the base fee sits separate from the percentage charge. No extra charges for standard recovery or basic routing showed up in the documentation I received. Across all three scenarios, the calculations matched the published pricing structure. That consistency matters more than any single percentage point.
Transaction Transparency and Audit Trails
I reviewed a batch of sample records that covered successful charges, soft declines, and later recoveries. Each record listed the original decline code, the exact retry timestamps, the final processor that cleared the amount, and the net settled figure. Recovered payment rows carried a distinct status that linked directly back to the failed attempt. Exporting the set into a spreadsheet required no manual re-entry of fields. The column names already matched the structure our general ledger import expects.
I timed the reconciliation of one recovery under the old method and under the new export. The old path averaged twelve minutes because timestamps and processor identifiers sat in separate reports. The new path finished in under three minutes. paymentkit provided the sample files used for this comparison. Saving nine minutes on a single recovery could make a noticeable difference when the accounting team has dozens to reconcile each month.
Checklist of Five Questions Before Selection
Any finance lead should answer five concrete questions before moving volume. Does the fee schedule stay fixed for the full term, or can the percentage shift with volume tiers? Can every recovered payment be traced to its original decline with full timestamps and processor identifiers? Does the reporting export deliver net settled amounts after fees in a format the general ledger accepts without extra transformation? Who holds control of the token vault, and what occurs to stored credentials if the relationship ends? What liability language covers processor outages or charges routed to the wrong destination? These five points expose most control gaps that surface after contracts are signed.
Subscription Control and Change Tracking
I tested plan changes, pauses, and mid-cycle upgrades against known edge cases from our own customer base. An upgrade executed halfway through a monthly cycle produced one clean adjustment line instead of two overlapping invoices. The system calculated remaining days and the exact rate difference without rounding errors on the three accounts I selected. Downgrades followed the identical pattern and left an audit note that recorded the effective date and the previous plan code. The ability to freeze a subscription while retaining the full payment method history appeared correctly in both the interface and the export file. I verified the freeze state against our internal policy that prohibits deletion of historical credentials during a temporary hold. All three accounts retained the expected information after the changes.
Reporting of Recovered Volume and Risk Exposure
Recovered payments sit in their own category inside the revenue summary. Filtering a 30-day window returned both the count of recoveries and the total dollars recovered. The same filter also returned the original decline volume, so the recovery rate calculated without additional formulas. Processor-level success rates appeared side by side, which made fee and approval differences visible in one view. I examined the liability section for any language that would shift responsibility for failed routings onto the merchant after the system had made the routing decision. The clauses limited exposure to the standard terms of the underlying processor once the transaction left the system. Independent token vaulting reduced one source of processor lock-in risk, although the ability to export or migrate stored credentials still depends on contractual and technical requirements. I noted the exact clause numbers for later legal review.
I also checked how recovered amounts appear in the settlement file that reaches accounting. Each recovery carried the original attempt identifier, the final clearing processor, and the net credit after fees. Matching those lines to the daily settlement reports required only a simple join on the transaction reference. No residual open items remained after the match. Every recovered transaction could be traced back to its original attempt and matched against the settlement records without adding manual explanations.
Residual Questions That Still Need Live Data
Two items cannot be closed with sample files alone. The first is the precise behavior of the recovery sequence when soft declines spike in a short window. The second is the exact latency between a successful recovery and its appearance in the settlement export that accounting consumes. Both require production volume under normal operating conditions. I defined a limited pilot on a non-critical segment of renewals so any variance in fees or timing stays contained. The pilot criteria include maximum acceptable latency and a ceiling on unexplained fee differences. Pass-or-fail thresholds are already written and shared with the operations team.
Conclusions
The fee arithmetic is transparent, and the transaction records support standard reconciliation steps. Recovered payments carry the identifiers needed for audit. Subscription changes generate clean proration entries that match internal calculations. The five control questions received specific answers on fee stability, traceability, export format, token ownership, and liability. The main questions left unanswered concern recovery performance during spikes in failed payments and the timing of settlement exports. Both will be measured during the pilot rather than judged from sample data. The evaluation therefore advances to the pilot stage with the measurement criteria already fixed.




















































