Guide — Method

When a price disappears: telling a withdrawn tier from a reshuffled page from a rebuilt one

Our change record holds 63 confirmed changes; 22 of them removed a figure, across 14 vendors. Exactly half of those removals put nothing back and half swapped one set of figures for another. The largest single removal took 237 figures off one page in one day and repriced nothing. This guide sets out how to tell those cases apart, and why a removal is confirmed only after several clean reads agree.

A price that vanishes is the one change a pricing record cannot hold. Every other movement leaves something behind to point at — a higher figure, a new tier, a renamed plan — but a withdrawal leaves an absence, and an absence looks the same whatever caused it. The plan may be gone. The page may have been reorganised around the same plans. Or the page may simply render differently today than it did yesterday, with every price unchanged behind it.

We keep the removals separately for that reason. Across the confirmed changes in the record as it stands on 5 September 2026, 63 changes are confirmed and 22 of them removed at least one figure, spread across fourteen vendors: aimlapi, Asana, Bitdefender, Intercom, LastPass, Notion, Private Internet Access, RoboForm, SendPulse, Surfer SEO, Teachable, Wix, WP Engine and Zendesk. That is a substantial minority of everything that moved, and it is the part most likely to be misread.

The first division is the useful one: did anything come back? Of those 22 removals, **eleven put nothing in the place of what went**, and **eleven swapped** — figures left and other figures arrived in the same read. Those two halves mean opposite things. A removal with nothing behind it is a tier that is no longer sold, or no longer shown to the country we read from. A removal with arrivals beside it is a ladder being rearranged, and the buyer's question is not "what was withdrawn" but "what does the equivalent seat cost now".

The second division is size, and it is where the reading most often goes wrong. The largest removals in this record belong to one vendor, aimlapi, whose pricing page dropped 237 figures in one confirmed change and 205 in another, on the same day, in dollars, with nothing added in either. Read as pricing events those are extraordinary. They are not pricing events. A page carrying several hundred numbers is not a ladder of several hundred plans; it is a table of rates, and when a table like that is rebuilt every number in it leaves the record at once. Nothing there tells you a single price fell.

Compare that with WP Engine on 2 September, where one confirmed change removed fourteen figures and added eight, and another removed eleven and added seventeen. Those are reshuffles: the ladder moved, the count moved with it, and the honest summary is that the shape changed rather than that eleven or fourteen prices were cut. And compare both with the smallest removals in the record — Teachable, Wix and Intercom each losing exactly one figure with nothing added. One figure, gone, nothing in its place, is the clean case: a tier that was for sale and now is not.

The third division is where it happened. Ten of the 22 removals were read in euros and twelve in dollars, which matters because a figure can leave one market's page while standing perfectly still in another. A euro price disappearing is not evidence that a dollar price moved. This is the same discipline the rest of the site runs on — a ladder belongs to the market it was read in — and it applies with more force to absences, because an absence in one market is the easiest thing in the world to mistake for a fact about the product.

None of this is recorded on sight. A removal is only confirmed after repeated clean reads agree that the figure is still gone, and the number of reads varies with how the page behaves: in this record the confirmations range from two clean reads to twelve, with the majority settled by two or four. A page that renders reliably earns a fast confirmation. A page that intermittently fails to paint its prices — which is common, and is a rendering fault rather than a commercial one — has to be read many more times before an absence counts as real, because the alternative is publishing a withdrawal that was a bad afternoon on someone's server.

That waiting period is the reason this record can be quoted at all, and it is also its main limitation: a genuine withdrawal is reported here days after it happened, not the morning it happened. We would rather be late and right. A tier that returns on the third read never enters the record, and no reader is told a plan was discontinued because a page failed to load twice.

What none of this can tell you is why. A vendor does not annotate a removal, and we do not infer one. A tier withdrawn from public pricing may still be sold on request, may have been folded into the tier above it, may have been renamed, or may be gone. Those are different outcomes for a buyer and the pricing page distinguishes none of them. Where the record says a figure was withdrawn, it means exactly that we read the page, the figure was there, we read it again on several later days and it was not. Everything past that is the vendor's to answer.

The practical reading, then. Count what left and count what arrived: nothing arriving is a withdrawal, something arriving is a reshuffle. Look at the size before drawing a conclusion, because a removal in the hundreds is a page being rebuilt and a removal of one is a plan being retired. Check which market it was read in before generalising. And treat the confirmation date as what it is — the day we became sure, not the day it changed.

Nothing arrived, so a tier went

Eleven of the 22 removals put nothing back. That is the clean case: a plan that was on the page is not on the page, and the buyer's question is what replaces it.

Something arrived, so the ladder moved

The other eleven removed and added in the same read. WP Engine did both on 2 September — fourteen out and eight in, then eleven out and seventeen in. The shape changed; that is not the same as prices falling.

Hundreds at once is a page, not a ladder

aimlapi's page dropped 237 figures in one confirmed change and 205 in another, on the same day, with nothing added. A rate table being rebuilt takes every number with it and reprices none of them.

One figure gone is the strongest signal

Teachable, Wix and Intercom each lost exactly one figure with nothing in its place. Small removals are the ones most likely to be a real commercial decision.

An absence belongs to one market

Ten of the removals were read in euros and twelve in dollars. A price leaving a euro page says nothing about the dollar page, and we do not let it.

Confirmed by repetition, not by sight

Confirmations in this record took between two and twelve clean reads, most often two or four. A page that renders unreliably has to say the same thing many more times before an absence is believed.

Questions people actually search

Does a withdrawn price mean the plan was discontinued?

Not necessarily, and we do not say so. It means the figure was on the vendor's public pricing page when we read it, and was not there across several later reads. The plan may have been folded into another tier, renamed, restricted to existing customers, or moved behind a sales conversation. A pricing page does not distinguish those and neither do we.

Why does the record show hundreds of prices removed from one page?

Because some pages are rate tables rather than plan ladders, and rebuilding one takes every number off at once. aimlapi's page accounts for the two largest removals in the record — 237 figures and 205 figures, the same day, nothing added — and neither is evidence that any single rate changed. Size is the first thing to check before reading a removal as a price cut.

How long before a disappearance is recorded?

As long as it takes for repeated clean reads to agree. In this record confirmations took between two and twelve clean reads, and the pages needing the most are the ones that intermittently fail to render their prices at all. The cost of that discipline is that we report a genuine withdrawal some days after it happens; the benefit is that we do not report one that was a page failing to load.

If a price disappears in Germany but not in the United States, what changed?

Only the German page, as far as we can say. Ten of the 22 removals in this record were euro reads and twelve were dollar reads, and we keep them apart deliberately. Vendors withdraw, restrict and re-tier by country, so an absence in one market is a fact about that market's page and nothing more until the other market's page says the same.

Can I see the figure that was withdrawn?

The change record holds it, with the day it was last read and the day the removal was confirmed, and the change pages publish it. What we will not do is carry a withdrawn figure forward into a comparison as though it were still purchasable — a price that is no longer offered is a historical fact, not an option, and mixing the two is how a price record starts quoting numbers nobody can buy.

Every count on this page comes from our own confirmed change record as it stood on 5 September 2026: 63 confirmed changes, of which 22 removed at least one figure across fourteen vendors, eleven with nothing added and eleven alongside additions, ten read in euros and twelve in dollars. Confirmation counts (two to twelve clean reads) are the record's own held_clean_reads values. Individual vendors are named because the record names them and dates them; the withdrawn amounts themselves are published on the change pages rather than repeated here, because a figure that can no longer be bought does not belong beside figures that can. No reason for any withdrawal is stated, because no vendor stated one. The method →

Found a price that no longer matches the vendor’s page? Tell us — corrections are dated and stay on the page.