Automation tools: a saving that does not exist, and a unit that gets dearer in bulk

Start with the claim, because it is the plainest thing in our records. IFTTT's page prints a saving of forty per cent beside its professional plan's annual option. That plan's monthly price multiplied by twelve equals the printed annual total exactly — to the penny — so the annual option is not forty per cent off, or four per cent off. As printed, it is nothing off. The tier above behaves the same way and makes no claim at all, which suggests the arithmetic is the vendor's normal practice and the badge is the anomaly.

It is worth being careful about what that does and does not establish. We are reading two numbers the vendor published on the same page on the same day, and dividing one into the other. There may be a discount elsewhere, on a term we did not see, or the badge may be stale. What can be said is that the two figures printed beside the badge do not support it, and that a reader doing the obvious arithmetic would find nothing. Where a claim and the numbers under it disagree, we record both and let the disagreement stand.

Zapier, in the same category, shows how the honest version looks. Its page claims a saving of thirty-three per cent for annual billing, and both of its priced tiers work out at almost exactly that — a third off, to within a hundredth of a point on each. The claim is accurate, it holds on every tier it covers rather than on the best one, and it can be checked in seconds from figures printed on the same page. Two vendors in one category, one claim confirmed to two decimal places and one contradicted entirely.

The second finding here is about the unit rather than the discount, and it inverts something this whole series has taken for granted. Everywhere else we have measured, buying more makes each unit cheaper: storage falls by more than half per terabyte, monitoring by more than half per monitor, credits stay roughly flat. ActivePieces goes the other way. Its larger paid plan costs about ten times its smaller one and carries five times the credit allowance, so each credit on the larger plan costs a little over twice what a credit on the smaller one costs.

That has an honest explanation and it is worth giving. The larger plan includes a substantial number of user seats where the smaller one does not, so some of that tenfold price increase is buying people rather than credits. The comparison is still worth making, because the credit allowance is the headline unit on both cards and a buyer reading them as a volume ladder would reach exactly the wrong conclusion. But it is a caution about reading a single dimension of a multi-dimensional price rather than proof that the vendor is charging more for less.

Make presents a subtler version of the same problem. Its paid tiers are all quoted at one volume point — the same credit allowance on each — so its ladder is not a volume ladder at all. Moving up it buys features, not capacity. That is entirely reasonable and it means comparing its tiers against another vendor's volume tiers compares two different things. Zapier's published prices, meanwhile, are floors: a slider above the plans sets how many tasks you need, and the printed figure is the bottom of that range.

Which brings up the thing that makes this category genuinely hard to compare. One vendor bills in tasks and another in credits, and those are not the same unit — a point one of the vendors makes on its own page. Setting one headline figure against the other compares different quantities of different work, in the same way credits could not be compared across AI tools. Any table that ranks automation platforms by monthly price without normalising the unit is ranking them by something other than cost.

One last detail worth carrying away, because it is the kind of thing that decides whether a free tier is usable. On one platform the free plan's binding constraint is not its credit allowance but a minimum interval between runs — a quarter of an hour. If you need something to happen more often than that, no amount of remaining allowance helps you. The limit that binds first is frequently not the one denominated in money, and on free tiers it usually is not.