The spreadsheet said the new system saved eleven hours a week. Nice, defensible, boring. What the spreadsheet didn't say: the ops manager had stopped flinching at 4:55 p.m., when the end-of-day export used to fail in a new and creative way each afternoon. Her shoulders came down from her ears sometime in week three. Show me the cell for that.

This is the soft ROI of software projects: the morale, trust, and sanity gains that are completely real and completely allergic to pivot tables. Vendors have abused the category so thoroughly that honest people now under-report it, which is its own kind of loss. So this is a field guide to measuring the soft stuff honestly, with proxy metrics you can defend in a budget meeting and a firm rule about never cooking the books.

The one-sentence answer: soft ROI is real value measured through proxies (behavior change, survey deltas, error patterns) rather than direct dollars, and it stays credible only if you document it with the same rigor as the hard numbers and never, ever blend the two columns together.

The win that won't fit in a cell

Hard ROI is easy to love: hours saved, errors avoided, headcount not hired. You can put it in a deck and watch a CFO nod. The standard FDE ROI calculation handles that column well, and you should do it first, always.

But sit with the users for a week and a different ledger appears. The warehouse lead who no longer prints the pick list "just in case the system is lying." The account manager who stops CCing herself on everything because the audit trail finally exists. The Friday afternoon that becomes an actual Friday afternoon. None of these ship with an invoice line. All of them show up in retention, in error rates six months later, and in whether the renewal conversation is easy or excruciating.

Why the soft ROI of software projects gets a bad name

Blame the vendors who treated "soft benefits" as a junk drawer for value they couldn't prove. Every failed rollout in history was defended with "strategic alignment," "improved collaboration," and my personal favorite, "future optionality," which is Latin for "we cannot find the benefit either."

So the skepticism is earned, and the response isn't to insist harder. It's to hold soft ROI to a rule that kills the junk drawer: soft ROI is real, vague ROI is not. If you can't name a proxy metric, a baseline, and an after-measurement, it doesn't go in the report. "The team seems happier" is a vibe. "Voluntary after-hours logins dropped from 38 per week to 9" is a measurement of the same thing, dressed for court.

A quick audit for junk-drawer claims, before they reach a deck: can you name who changed their behavior, what they stopped doing, and when you'd expect to see it? If a benefit survives those three questions, it's soft ROI and belongs in the report with a proxy attached. If it dissolves ("the organization is more agile now" — agile how, Susan), it's confetti, and confetti belongs at launch parties, not in financial documents.

Measuring morale without being weird about it

You don't need an engagement consultancy. You need two questions and the discipline to ask them before the project starts, not after it succeeds.

The two-question survey

Before kickoff and again at week eight, ask the affected team, anonymously: "How much of your week feels like fighting your tools?" and "How confident are you that the numbers you see are right?" Ten-point scales, thirty seconds to answer. On one engagement (illustrative numbers, composite client), the tools-fighting score moved from 7.8 to 3.1 and the confidence score from 4.2 to 8.6. That's morale, quantified well enough for a budget appendix.

Behavior tells the rest. Voluntary overtime patterns, the tone of the Monday meeting, whether anyone has invented a personal shadow spreadsheet lately. Shadow spreadsheets are the canary in the coal mine: people build them when they don't trust the system, and they quietly delete them when they start to. Count the canaries.

Measuring trust and sanity

Trust in a system has a beautiful, measurable failure mode: double-entry. When people don't trust the software, they keep their own records beside it, in a notebook, a spreadsheet, or an email folder named IMPORTANT DO NOT DELETE. The double-entry rate is the single best soft metric I know. On that composite engagement, roughly 40% of orders were being manually re-verified before the rebuild. Eight weeks after, it was under 5%, and the remaining 5% was one guy named Dale who re-verifies everything, including his own birthday. Dale is a rounding error with a badge.

Sanity is after-hours messages, weekend tickets, and the Sunday-evening dread index. Count "urgent" messages sent outside working hours before and after; on a healthy project that number craters, because the system stops generating emergencies as a hobby. None of this requires a survey platform. It requires reading the metadata of the tools you already have, because the expense of manual double-checking was always there, just filed under "normal."

One more proxy worth stealing: reopen rates and meeting drift. When people trust a system, tickets about it get shorter and duller, "add a column" instead of "the numbers are wrong again." When they don't, every status meeting grows a ten-minute segment where someone re-litigates whether the data is real. Time that segment. It feels awkward the first week and devastatingly clear by the fourth, and it converts "the team doesn't trust it" from an accusation into a trend line you can act on.

The honest documentation template

Every soft claim goes in the report in exactly five fields:

Then the rule that saves your reputation: soft numbers never enter the hard-ROI total. The breakeven math, the payback period, all of it runs on dollars alone, exactly the discipline in finding the FDE breakeven point and the cost framing in the FDE cost breakdown. The soft column sits beside the hard one, clearly labeled, doing what it's supposed to do: explaining why the hard numbers will hold.

When soft ROI decides the renewal

Here's the pattern I've watched across renewals, including the ones where the fee conversation got spicy: the CFO renews on the hard column, but the team demands the renewal on the soft one. If the dispatchers, the accountants, and the warehouse lead all walk into the meeting saying the thing made their lives better, the CFO's spreadsheet gains ten pounds of credibility. If they walk in neutral, no spreadsheet survives.

So keep both columns, keep them separate, and measure the soft one with the same seriousness as the first. Morale, trust, and sanity aren't the mushy stuff you mention when the numbers are thin. They're the leading indicators of whether the numbers will still be there next year. The flinch test, quantified, is still the flinch test. It just fits in a deck now.