Big ROI numbers can cost you the room. Work out what a leadership programme must achieve to pay for itself, then test that against a conservative method.
Anyone who has taken a leadership programme through a budget review knows the moment. The design is agreed and the sponsor is keen. Then someone from finance asks what the company gets back, and the room turns to you.
The usual answer is an ROI figure: a percentage, often a large one, built from attrition savings and productivity gains. There are plenty of guides to producing one. This article argues for a different answer. It is smaller, easier to defend, and it holds up better when finance tests it.
In 1994, Gary Latham and Glen Whyte ran an experiment with 143 experienced managers. All of them were shown a recommendation to adopt a validated HR procedure. Some also saw a utility analysis, the formal method for putting a dollar value on an HR programme. In the authors' words, utility analysis "reduced the support of managers for implementing a valid selection procedure, even though the analysis indicated that the net benefits from the new procedure were substantial." [1]
Two limits matter here. The procedure was a selection test, not a training programme. And the finding has not held everywhere: Macan and Foster (2004) report that later work, including a 1998 study by Carson, Becker and Henderson that I have not read directly, found small positive effects of utility information on decision-makers [2].
So the fair reading is not "never show a number." It is narrower and more useful: a big dollar figure is not reliably persuasive, and with experienced managers it can backfire. Anyone whose job is to be sceptical of forecasts has seen a lot of large projected returns. An ROI of several hundred percent on a leadership programme does not read to them as evidence. It reads as a claim they now have to take apart.
Break-even turns the question round. You stop asking "what will this return?" and ask "what would this have to achieve to pay for itself?"
That question has three useful properties. Finance can check every input. It needs no attribution argument at the start. And it moves the debate from whether your number is inflated to whether the threshold is realistic, which is a conversation between two people looking at the same business.
Here is a worked example. Every figure is hypothetical. Replace each one with the client's own.
The programme. 40 first-line managers, together responsible for 320 staff.
Step 1: Fully load the cost. The ROI Institute's guiding principles say to "fully load all costs of a solution, project, or program when analyzing ROI" [3]. Finance will do this whether you do or not, so do it first.
| Cost line | Basis | Amount |
|---|---|---|
| External design and delivery fee | quoted | $60,000 |
| Participant time | 40 managers × 24 hours × $60 loaded hourly cost | $57,600 |
| Internal coordination | HR business partner and admin time | $8,000 |
| Evaluation | surveys, data pulls, analysis | $4,400 |
| Total | $130,000 |
Participant time is the line most proposals leave out, and it is nearly half the total here. If you leave it out, finance will add it back, and your whole case loses credibility at that moment.
Step 2: Price one unit of the outcome. Take the outcome the sponsor already reports and is already worried about: here, regretted attrition. Use the client's own replacement-cost figure if they have one. If they don't, Boushey and Glynn's review of 30 case studies from 11 research papers found that "the typical (median) cost of turnover was 21 percent" of salary for all positions except executives and physicians, rising to as much as 213% for executive roles [4]. At an average salary of $55,000, one departure costs about $11,550.
Step 3: Divide. $130,000 ÷ $11,550 = 11.3. The programme breaks even if it prevents 12 regretted departures in the year.
Step 4: Put the threshold in context. At the current 18% regretted attrition, the 320 staff will produce about 58 departures this year. Twelve is roughly one in five of them, or 3.75 percentage points off the rate.
Now you have a sentence finance can test:
This programme pays for itself if it prevents about 12 of the 58 regretted departures we expect among these managers' teams this year.
Nobody in the room has to accept a forecast. They only have to decide whether one in five is plausible, given what they know about why people leave. That is a judgement the business is better placed to make than you are, and asking them to make it moves them onto your side of the table.
Break-even says what the programme must achieve. It does not show that it will. After the programme runs, someone will ask how much of any improvement it actually caused. The ROI Institute's published method answers that with a set of isolation techniques: control groups, trend lines, mathematical modelling, and estimates from participants, managers, experts or customers [3]. When estimates are used, the guide says: "because they are not always accurate, the estimates are adjusted for error, using a confidence percentage" [3].
Run the example through it. Say attrition falls from 18% to 12% over the year: 19 fewer departures. There was no control group, so managers are asked to estimate how much of that fall was down to the programme. Their average answer is 50%, and their average confidence in that answer is 70%.
Attrition fell by a third and the conservative method still reports a loss. That is not an unusual result. It is what happens when a programme's value case rests on one outcome line and the method discounts it twice, as it is designed to. The guide's principles make it stricter still: "when analyzing data, select the most conservative alternative"; "use only the first year of annual benefits in ROI analysis of short-term solutions"; and "if no improvement data are available for a population or from a specific source, assume that little or no improvement has occurred" [3].
Each of those rules is what a careful CFO would apply anyway. The practical lesson is to run this calculation on your own design at the proposal stage, not in month nine. If the plausible outcome doesn't clear break-even on the outcomes you can actually measure, you have three honest options:
What you should not do is go into the budget meeting with a number the conservative method would take apart.
Put three things in the business case, in this order:
A proposal built this way asks the CFO to judge whether a threshold is plausible, not to believe a projected return. Because the evaluation method is declared in advance, at its most conservative, whatever result comes back at month twelve has already been agreed as fair. That is worth more than any percentage you could put on a slide.
Before the numbers, the conversation. Break-even only works if you know which outcome the sponsor already watches and which one keeps them up at night. That comes out of the discovery call, not the spreadsheet. The free Diagnostic Conversation Coach runs you through that conversation, with no signup and no email.
[1] Latham, G. P. & Whyte, G. (1994). "The futility of utility analysis." Personnel Psychology, 47, 31–46. DOI: 10.1111/j.1744-6570.1994.tb02408.x. Abstract quoted from the copy hosted at Academia.edu.
[2] Macan, T. H. & Foster, J. (2004). "Managers' reactions to utility analysis and perceptions of what influences their decisions." Journal of Business and Psychology, 19(2), 241–253. DOI: 10.1007/s10869-004-0550-x. Carson, Becker & Henderson (1998), Journal of Applied Psychology, 83, 84–96, is cited as reported in this abstract; not read directly.
[3] ROI Institute (2019). Application Guide: The ROI Methodology in 12 Easy Steps. PDF.
[4] Boushey, H. & Glynn, S. J. (2012, 16 November). There Are Significant Business Costs to Replacing Employees. Center for American Progress. PDF.