The effective rate, and why it is not what you were quoted
The effective rate is the simplest arithmetic in this series and the most commonly got wrong, almost always by leaving things out of the numerator.
Key takeaways
Total fees over total card turnover. Everything the provider charged goes on top.
Include terminal rental, PCI, gateway and the monthly minimum.
Include authorisation fees on declined transactions. They have no sale to sit against.
Report the margin separately. It is the only number you can move.
Never compute a rate on a partial month.
What goes in the numerator
Fig 1. The left bar is the quote. The middle bar is what most people compute. The right bar is what you actually pay, and the gap between the middle and the right is the part that gets left out.
The things most often left out
Terminal rental and PCI charges. They are fixed monthly costs of taking cards. On a quiet month they can be a third of the bill.
The monthly minimum service charge. If turnover falls below the threshold you pay the minimum, and the effective rate that month is far higher than any quoted figure.
Authorisation fees on declines. Charged per attempt, whatever the outcome, and they sit against no sale at all. On a subscription or a hotel book this is material.
Chargeback and retrieval fees. Small, occasional, and they belong in the total because you paid them to take cards.
The two numbers to carry
Every conversation about switching should start from two figures: the effective rate, and the acquirer margin as a share of it. The first tells you what card acceptance costs your business. The second tells you how much of that any provider could possibly change.
In the example above the effective rate is 2.31% and the margin is 0.42 points of it. A provider who halves their margin saves you 0.21 points — about eighty pounds a month on thirty-nine thousand of turnover. That is worth having and it is not the thousand pounds a switching pitch will imply.
Why partial months lie
Fig 2. This is the single most common way a fee analysis embarrasses itself, and the fix is to refuse to report until the month is closed.
The report therefore fires on the monthly close and not on upload. It is a small constraint that removes an entire class of false alarm, and it is the reason the scheduling in part seven looks the way it does.