A browser provider's bill did not move after your veterinary booking suite's sessions got much shorter. Why?

## What a flat bill is actually evidence of Every spend lever acts on exactly one quantity. Shortening a session is a **duration** lever, and a duration lever can only move a charge that is a function of duration. So when the veterinary booking suite's sessions genuinely got shorter and the provider charge did not follow, the first thing to notice is that this is information — evidence about the meter rather than a failed optimisation. The question to answer is: *which quantity is this bill a function of?* Four explanations cover almost every real case, and they call for completely different responses. ## The charge is not duration-shaped at all Two of the common metering shapes are structurally immune to a duration lever. - **Reserved parallelism.** The provider holds a width of simultaneous sessions available to your account and charges for the promise. That charge is settled before the period begins, so nothing happening during it is an input. Shorter sessions return the width to idleness sooner, and idleness is exactly what you already pay for. - **Access counting.** The charge follows the group of people admitted to the account. It is a function of your organisation rather than your workload, so a suite that ran all night and a suite that never ran at all produce the same charge. If you are on either shape, the correct conclusion is not *try harder*. It is that the lever you chose cannot work, and a different one has to be found. ## The charge is duration-shaped but sits inside a floor An arrangement can be metered and still fail to respond, if you had already committed to a minimum. Consumption falls, the charge stays where the commitment put it, and the reduction is absorbed rather than realised. This is the most misleading of the four, because the meter *is* counting duration and would respond — just not from where you are standing. The saving becomes real only once consumption climbs back above the commitment, or when the commitment itself is renegotiated. ## You shortened a part nobody was counting Establish where the counted span starts and stops before assuming your change was inside it. If the span runs from the moment the session is created to the moment it is released, everything in between counts: 1. **Setup** — the browser starting, the first page loading, the receptionist login the suite performs before it can reach the appointment calendar at all. 2. **The test's own work** — the part teams instinctively optimise, and frequently the smaller share of the whole span. 3. **Teardown and release** — the fixture closing, plus any stretch where the session is finished but has not yet been handed back. A team that made assertions faster inside an unchanged setup, an unchanged teardown and an unchanged idle gap has shortened a slice of the counted span and left the rest alone. The measured reduction is real; the share of the metered interval it represents is small. ## Telling the four apart The distinguishing method is a controlled change, not a theory. - Start with the arrangement itself and whatever record of consumption the account gives you. Often it names the quantity outright and the investigation ends there. - If it does not, change one input at a time and watch the charge. A period with a narrower reserved width, or a period in which nobody new is admitted, gives a clean signal that duration alone never will. - Compare a session's counted span against the wall-clock your own harness reports for the same run. A large gap between them says the metered interval is wider than the work. - Look for a step rather than a slope. Charges that jump between periods and then sit flat are commitment-, reservation- or access-shaped — an access-counted charge steps only when someone is admitted or removed — while charges that drift with activity are duration-shaped. ## What to do once you know, and what you gained anyway Match the lever to the shape and stop guessing. Each counted quantity answers to exactly one lever: a duration-shaped charge to fewer and shorter counted spans, setup and teardown included; a reserved width to a narrower reservation sized from your genuine peak; an access-counted charge to a narrower group holding direct access. The fourth case is not a fourth shape at all — a committed minimum is a floor under a duration-shaped charge, so the lever you already hold is the right one, and what it needs is either a renegotiated commitment or consumption grown back above the line. One thing is worth saying to the team out loud: a flat charge does not mean the optimisation was pointless. Shorter sessions still cut wall-clock, still shrink the window in which the far side can terminate a run underneath you, and still reduce how much is lost when a single session goes wrong. Under a duration-shaped charge those benefits arrive with money attached; under the other shapes they arrive alone. The failure was never the engineering. It was choosing a lever before establishing what the bill is a function of, which is a half-day of reading and considerably cheaper than a sprint of optimisation that could not bite.