What 45 L&D Professionals Think About ROI

I recently posted a question on LinkedIn that generated more response than almost anything I've written there. The setup was simple:

Your organization invests approximately $150,000 in a training program for 100 people. One participant has an insight during or after the training that leads to a behavior change, which generates a one-time revenue gain of $450,000. There is no other measurable evidence of impact. Was this a successful project?

I asked people to lead with a yes or no before qualifying their answer. In the first day it was up, forty-five comments came in. The debate was genuine, the range of views was wide, and the themes that emerged are worth documenting — both for what they reveal about how the field thinks about ROI, and for what they suggest about where the conversation needs to go next.

Here's what the comments contained (note: AI helped generate this summary).

Theme 1: Yes, because the ROI math works

The most common "yes" position was also the most straightforward: 3x return on a $150K investment is a 3x return. Several senior practitioners landed here, some with caveats about attribution but still concluding that the financial outcome justified the investment. This is the position that takes the scenario at face value and applies standard return-on-investment logic.

Theme 2: No, because attribution is impossible

The most common "no" position centered on a problem the scenario deliberately leaves open: you can't credibly claim the training caused the insight. The participant might have had it anyway. The behavior change might have been prompted by something else entirely. Claiming credit for the $450K without being able to demonstrate causation is, in this view, exactly the kind of ROI overclaiming that undermines L&D's credibility as a function.

Theme 3: ROI can be distorted by a single anomalous outcome

A smaller but intellectually more interesting cluster argued that the scenario exposes something more fundamental: ROI may be structurally the wrong tool for evaluating learning investments. The specific problem isn't just that it misses unmeasurable outcomes — it's that it can overprivilege a single financial result that may be rare, unreplicable, and unrepresentative of whether the program actually works. A 3x return generated by one person's insight passes the ROI sniff test. It tells you almost nothing about whether the program would produce anything similar if run again. That's a significant limitation for a tool being used to make investment decisions.

Theme 4: It depends on intent and design

Several commenters introduced design questions the scenario deliberately omits. Was the program designed to produce this kind of breakthrough insight? Was the one participant's outcome the intended goal or a happy accident? If the organization explicitly commissioned a high-investment, moonshot-style program in the hopes of generating at least one major financial result, a 3x return is hard to call anything other than a success. If the outcome was accidental, the calculus changes.

Theme 5: What about the other 99?

A recurring thread challenged the framing from a different angle: even if one person generated $450K, the program reached 100 people. What happened to the other 99? Some argued that a program which demonstrably failed to impact 99% of its participants has failed as a program, regardless of the net financial return. Others pushed back that we don't know the 99 generated nothing — only that nothing was measured.

Theme 6: The hedgers

A significant number of commenters led with qualifications rather than a yes or no, some explicitly acknowledging they were "cheating" by not answering directly. The pattern itself is data: the field is genuinely uncomfortable with binary judgments about program success, even in a hypothetical with a deliberately clean fact pattern. That discomfort is worth examining.

Theme 7: The measurement problem

Several commenters identified what may be the most important issue in the scenario: not the $450K, but the absence of any measurement infrastructure for the other 99 participants. The problem isn't the outcome — it's that the program was built with no way to know whether it worked for anyone other than the person whose impact happened to be visible. "No other evidence of impact" may mean no impact. Or it may mean no measurement. Those are very different situations, and the scenario doesn't let you tell them apart.

My view, and where this goes next

My answer to the original question is no — with the caveat that context and intent matter, as Theme 4 commenters rightly noted.

Going off the simple fact pattern as stated, declaring this a success would mean accepting that any program touching enough people is justified by the chance that someone might do something valuable afterward. That's not a standard. It's a rationalization.

But the more important conversation — the one that Themes 3 and 7 are pointing toward — is about whether ROI is the right construct for evaluating learning investments at all. Several commenters got there independently, which suggests the field may be readier for that conversation than it sometimes appears. I'll be developing that argument in a follow-up Practice Notes piece shortly.