Show Customer Value Without Dressing Assumptions as ROI
Show Customer Value Without Dressing Assumptions as ROI
Two and a half minutes. Multiply by the bookings a scheduler handles in a week, by the number of schedulers, by fifty-two weeks, by a loaded hourly rate. The arithmetic is easy to follow, and that is precisely what makes it persuasive. The first number may have been measured. Everything after it is a decision someone made about the future, arranged in a column so it looks like the same kind of fact.
There are three defensible things a sales team can do in that position, and they are not interchangeable. It can describe the change it observed, in the terms it observed, and stop. It can build a conditional scenario that exposes exactly which assumptions the value depends on, labeled as a scenario and nothing more. Or, when no usable baseline exists, it can propose a bounded way to produce one. What it should not do is present the second or third as though they were the first.
The distinction that stays hidden inside a single spreadsheet cell is the one worth carrying through the whole conversation: work that actually changed, capacity that became available, and money that left a budget. Most inflated value claims come from collapsing the first into the third without ever showing the second.
Establish what is being compared
Take an invented case, because the arithmetic is easier to inspect when no real customer is attached to it. Halden Physio Group is a fictional three-clinic practice. It is evaluating a scheduling assistant that hides appointment slots likely to create a clash.
The customer task is narrow and concrete. A scheduler books a follow-up appointment. Checking for clashes means confirming the clinician is free, the room is free, and any travel between clinics fits. Today the scheduler scans two calendar views and, when a slot looks ambiguous, calls the clinician. The check, including any call, runs about four minutes at the median. Roughly one booking in three needs a call.
Halden runs a two-week trial at one of its three clinics. Four of the clinic's seven schedulers use the assistant; the other three keep the old process. The assistant removes most potential clashes before the scheduler sees the slot list, and the scheduler calls only on exceptions.
In the trial, the whole check — review plus call — falls to a median of about a minute and a half. Calls drop from roughly one booking in three to about one in five. The site administrator, meanwhile, begins spending about twenty minutes each morning keeping room and travel rules current, and a scheduler occasionally has to rebook when a travel rule turns out to be wrong.
That is a recorded observation with a stated scope: a named step, four named people, one clinic, two weeks, with three colleagues still working the old way. Before it becomes a value claim, two questions have to be answered honestly.
Is the comparison equivalent work under equivalent conditions? Here, mostly yes, and the three non-adopting schedulers are what make it checkable. Had all seven switched at once, a drop in check time could have come from seasonal booking patterns, a new patient mix, or the simple fact that people perform differently when watched.
And has work moved rather than disappeared? The twenty minutes of daily rule maintenance did not exist before. It is small next to what the trial claims to save, so it will not sink the case on its own — but omitting it makes the remaining number slightly too clean and much less believable. A faster selected step is not automatically a cheaper workflow when preparation or checking has simply relocated.
Follow released time to the use it could actually have
Two and a half minutes saved per booking is a real observation. It is also, by itself, nothing you can spend. Follow it and see where it goes.
Each scheduler at that clinic handles roughly 120 bookings a week. Multiply the observed reduction by that figure and you get about five hours per scheduler per week. Across the seven schedulers, if all of them adopted and the reduction held, the clinic would recover about thirty-five hours a week in that one step. The arithmetic is fine.
The question is what the clinic could do with thirty-five hours a week, and every answer needs different evidence.
Complete more work. The clinic has a waiting list, so freed scheduler time could become more appointments. That requires the assistant's benefit to survive into booking volume, which requires that rooms and clinicians are actually free in the hours the schedulers recover — if rooms are the binding constraint at peak times, more scheduler time produces nothing except a longer list of patients who are ready to be booked. You would need room and clinician availability to check this, not scheduler time.
Relieve a bottleneck. If the clinic's real constraint is answering patient calls or chasing recalls, the recovered time can go there instead. In the trial, that is exactly what happened: the four schedulers used most of the recovered time to work down a backlog of recall calls, which was genuinely worth doing. It is also finite. Backlog absorption is a one-time benefit wearing the costume of a recurring one, and it will look very different in week twenty than in week two.
Change service quality. Shorter checks might reduce booking errors, cut rebooking calls to patients, or shorten the wait before an appointment is confirmed. These are plausible and none of them is in the trial data. Each would need its own observation.
Possibly reduce a cost. This is the largest claim and the one with the least support in the example. It is possible. It has not been shown.
The UK Government Digital Service's service manual guidance on measuring service benefits is useful here precisely because of how it separates things: cashable benefits, non-cashable benefits, and wider economic benefits are treated as different animals, and teams are told to scrutinize their assumptions and show how much they expect to improve things by. It is public-service planning guidance, and its vocabulary is not a universal corporate taxonomy — a private customer's result cannot be classified into its categories and called done. But the underlying discipline transfers. Saving a scheduler's time is a non-cashable benefit until something specific converts it. If the monetization is unknown, say so and keep the operational benefit. Do not upgrade it to a cash saving because the operational version feels like a weaker slide.
Bring the customer's effort into the same comparison
Value claims usually subtract nothing. The customer's costs show up in a footnote, or a separate slide, or nowhere. Bring them into the same comparison, over the same horizon.
At Halden, the transition work is concrete: two half-days of training for each participating scheduler, and a per-seat fee. Continuing effort includes handling the one booking in five that still needs a call, the administrator's twenty minutes each morning keeping room and travel rules current, and the occasional rebooking when a travel rule is wrong.
Then the conditions of the comparison itself. Four of seven schedulers adopted, at one of three clinics, for two weeks in one season. Travel and room rules change with the season. A trial in a quiet fortnight is not a trial in a busy one. Spreading the observed reduction across all three clinics and all seven schedulers is not an observation; it is an extrapolation with conditions attached, and those conditions are part of the claim rather than a caveat underneath it.
The ratio of observed to assumed effort matters as much as the total. Training is observed — it happened, and someone can say how long it took. Future supervision is assumed. When a comparison quietly converts the second into the first, the customer is being asked to accept a projection as a measurement.
Three honest ways to present value now
Given this evidence, Halden's vendor has three credible routes. Which one it picks should be decided by the customer's pending decision and by the available support, not by which produces the largest number.
A bounded operational account. State the measured change inside its scope and stop. For four schedulers at one clinic over two weeks, the clash-check step, including calls, fell from a median of about four minutes to about a minute and a half. Rule maintenance of about twenty minutes each morning moved to the site administrator. The recovered time went to a recall backlog. No financial result is claimed, because none has been established.
This is the most defensible form and the one that answers the buyer's question least. If the buyer's actual decision is whether to fund a line item, an operational account will not close it, and pretending otherwise wastes everyone's time.
A conditional scenario. Show how value depends on assumptions the customer can inspect. If all seven schedulers at the clinic adopted the assistant, and if the observed reduction held, the step would release roughly thirty-five hours a week. Those hours could become additional appointments from the waiting list in the hours where a room and a clinician are both free; at the group's three clinics on the same assumptions, roughly three times that. Each clause is an assumption, and the scenario is not a forecast.
The test of a well-built scenario is that changing one assumption visibly changes the answer. If the clinic cannot free a room at the relevant hours, the whole benefit drops toward the value of the reduced calls alone. A scenario that stays impressive under every assumption is a forecast in disguise.
A proposal to establish the missing baseline. When there is no usable baseline at all — no measured check time, no comparison group, no idea how often the task occurs — the honest output is a bounded measurement with a decision at the end, not an estimate with a decimal point. The differences between that kind of exercise and a general pilot are worth their own treatment; what matters here is that proposing measurement is a legitimate answer, not a failure to produce one.
If the deck contains a fourth slide — "saves about one full-time post, $X a year" — look at how it was built. Two and a half minutes across 840 bookings a week is about 1,800 hours a year, roughly one full-time job on paper. The multiplication works. The conversion fails, for reasons the customer already suspects: nobody's hours or pay are going to change, so no money leaves the budget; the recovered time arrives in small scattered pieces, not as a post that can be declined; the administrator's continuing minutes and the per-seat fee were subtracted from nothing; and the entire figure rests on four people at one clinic extrapolated to three.
The problem is not the size of the number. It is that the slide presents a modeled conversion as though it were the same order of evidence as the measured minutes that opened it.
Keep the assumptions where the claim is read
The assumptions that decide the claim belong beside the claim, not in a disclaimer the reader reaches after the comparison. A "results may vary" line at the bottom of a slide does not carry the assumption that rooms are free at peak hours, and a footnote does not carry the assumption that all seven schedulers will adopt at the same rate as four volunteers.
Name the assumption whose failure would change the customer's decision. At Halden, that is probably room and clinician availability at the hours freed, because if the clinic cannot convert recovered scheduler time into appointments, the case rests entirely on a modest reduction in calls. Everything else can be handled with a sentence. That one deserves a slide.
Then have the right people look at it. If a real monetary interpretation is going into a customer document, it should be reviewed by the customer's own finance or operations owner and by whoever owns the model on the vendor side. This is not a financial recommendation and there is no universal calculator that will produce one. It is a presentation decision about a single value claim, made by people who can see the customer's actual records.
What to put on the last slide
Choose the form your evidence can carry: an observed operational change inside its stated scope, a conditional scenario whose assumptions the customer can push on, or a bounded measurement when the baseline is missing. Put the customer's effort and the observation horizon in the same comparison as the benefit. Keep the deciding assumption next to the claim.
Then write the limit into the same slide, in ordinary language. At Halden that sentence is something like: the check step got faster for four schedulers over two weeks; whether that becomes appointments depends on room and clinician availability, which we have not measured.
An operational improvement that is honestly described will sometimes lose to a fabricated return. That is a real commercial cost, and it is smaller than it looks. A customer who discovers that a modest, well-supported claim was true will take the next call. A customer who discovers that a confident number was a multiplication problem wearing a lab coat will not.
Frequently asked questions
What are three honest ways to present customer value when evidence is mixed?
Describe the observed change in the terms observed and stop; build a conditional scenario that exposes the assumptions the value depends on and label it as a scenario; or, when no usable baseline exists, propose a bounded way to produce one. Do not present a scenario or a measurement proposal as though it were a recorded observation.
What did the Halden trial actually observe?
Four of seven schedulers at one of three clinics used the assistant for two weeks while three kept the old process. The whole clash check, including calls, fell from a median of about four minutes to about a minute and a half, and calls dropped from roughly one booking in three to about one in five. The site administrator began spending about twenty minutes each morning maintaining room and travel rules, and a scheduler occasionally had to rebook when a travel rule was wrong.
Why does two and a half minutes saved per booking not become a cash saving by itself?
It is released time, not money. Something specific must convert it. Completing more work needs room and clinician availability in the recovered hours; relieving a bottleneck can absorb a recall backlog but that is finite; service-quality changes were not measured; reducing cost is possible but has the least support. A non-cashable operational benefit should stay operational until a conversion is evidenced.
Which assumption belongs next to the value claim?
Name the assumption whose failure would change the customer's decision. At Halden that is likely room and clinician availability at the hours freed, because if recovered scheduler time cannot become appointments, the case rests mainly on reduced calls. Put the customer's effort and the observation horizon in the same comparison: training, per-seat fee, remaining calls, rule maintenance, rebookings, four of seven schedulers, one of three clinics, two weeks, and one season.
How should a slide claiming it saves about one full-time post be evaluated?
Two and a half minutes across 840 bookings a week is about 1,800 hours a year, roughly one full-time job on paper. The multiplication works, but the conversion fails: nobody's hours or pay are going to change, no money leaves the budget, recovered time arrives in small scattered pieces, the administrator's continuing minutes and the per-seat fee were subtracted from nothing, and the figure rests on four people at one clinic extrapolated to three. It presents a modeled conversion as the same order of evidence as the measured minutes.