There is a question a pricing page is built not to answer, and it is what you will pay in the second year. The figure on the card is frequently an introductory rate, and it lives beside a standing rate you will meet later, at a date the page may not mention. Eight vendors in our record advertise on that structure. The gap between the two numbers is not small in any of them, and in one case it is larger than the sticker itself.
NordVPN is the clearest instance, because it does not hide the mechanism at all. Its own pricing page carries the line 'Savings: intro vs. renewal prices', which is an admission most vendors would never print: the discount is measured against what you will pay on renewal, not against some earlier price. The entry plan advertises $3.49 a month on the term rate. Our note against that record, taken from the same page, puts the renewal at more than three times that rate per month. Nothing there is concealed. It is simply written where nobody reads, in the same typeface as the guarantee and the VAT line.
The first tell that a price is introductory is that it sits on a round fraction of another number. FreshBooks advertises its whole ladder at exactly a tenth of the standing price, every tier: the entry plan shows $2.3 against a standing $23, and the top of the ladder $7 against $70. Wave's paid tier, standing at $19 a month, is advertised at exactly half. FreeAgent's four plans are each exactly half of theirs. When a discount lands on a clean half or a clean tenth across an entire ladder, the standing price is the designed number and the advertised one has been derived from it. Genuine price differences between tiers are not that tidy, and neither are genuine changes over time.
The second tell is what is missing, and here the record is blunt about a gap it cannot close. Of the vendors that discount, only Jetpack states the duration in the field beside the price: 'fifty per cent off the first year', a complete sentence you can act on. One other gets as far as 'for the first 6' and simply stops, six of a unit the field never names. The rest record no duration at all, because the page carried none. A discount without an end date is not a price, it is an invitation, and the record's job at that point is to say that the end date was not published rather than to guess a common one.
Sometimes the duration can be recovered by arithmetic even when it is not written. FreshBooks prints a saving figure beside each plan, and that figure is exactly three times the monthly gap between the advertised rate and the standing one. Three months, stated nowhere in words, recoverable from the vendor's own number. That is the sort of check worth running yourself: if a page shows a discounted rate, a standing rate and a total saving, the saving divided by the monthly gap is the length of the offer, and it will sometimes be shorter than you assumed.
Jetpack deserves naming as the counter-example, because the honest version of this costs a vendor nothing and is rare anyway. Its bundles print $9.95 a month beside a struck-through $19.95, under a plain statement that the discount covers the first year. Both figures are on the page, the standing one is legible rather than buried, and the duration is unambiguous. The top bundle does the same, $24.95 against $49.95. Nothing about that presentation stops anyone doing the arithmetic, which is presumably why so few pages present it that way.
The harder question is whether a discount is a discount at all, and that one cannot be answered by reading a page once. A badge saying seventy-five per cent off is a claim about time: it asserts a standing price that you are today being let off. A dated record can test the claim, because if the same badge and the same figure appear on every reading anyone has ever taken, then the badge is not describing a discount. It is describing the price, with decoration.
So we test it. The instrument behind is that discount real? counts, per vendor and per market, how many dated reads carried a given badge and how many carried the discounted figure underneath it. It refuses to answer in more situations than it answers in. A reading that cannot name the market it was taken in is excluded outright, because attributing a price to the wrong country is a worse error than having no data. A saving has to be quantified as a percentage or an amount before it counts as a badge at all, which keeps out the ROI tables some vendors publish, where a row reading 'savings, three year total' is a projection about your staffing costs and has nothing to do with the price of the software. And an annual-billing saving is excluded from the ranking entirely: 'save twenty per cent, pay once a year' describes how the vendor bills, is true every day by design, and reporting it as a promotion that has run for a week would be a category error dressed up as a finding.
What is left, after all that refusing, is one hundred and forty promotional findings across thirteen vendors. Eighty-six of them carried their badge on every single read. NordVPN's 'Save 75%' was present on all seven reads in the United States, all seven in Britain and all seven in Germany, the same offer running in three countries simultaneously without a day's interruption. Its other badge, a sixty-nine per cent one, did the same. ExpressVPN's business pages carried several at once on every read.
The window is seven days, and that has to be said plainly rather than buried, because it is the limit of what the claim can bear. Seven days of an unbroken badge does not prove a permanent promotion; it proves the offer did not move while we were watching, which is a smaller statement. What makes even that worth something is that the instrument does detect movement when there is any. Brevo's discounted figure was absent on five of the seven days. Mailchimp's dropped out for a day in four different markets. Mangools lost both its badge and its figure on the second of September and got them back afterwards. The measure is not blind, so an unbroken run is evidence rather than an artefact of never looking.
The practical upshot is a short list of things to do before you enter a card number. Find the renewal price, and if the page does not print one, treat that as the finding it is rather than an oversight. Divide any advertised saving by the monthly gap to get the length of the offer. Check whether the discount is a clean half or a clean tenth, which tells you which of the two numbers the vendor actually designed. And budget on the standing rate, because that is the one you will be paying for every year but the first.