Every automation pitch comes with a number attached — hours saved, cost reduced, revenue unlocked. Most of those numbers are made up, or at least optimistically rounded, because "the automation saves 6 hours a week" is a much easier sentence to write than to actually measure. We've tried to hold ourselves to a stricter standard, and it's worth explaining what that standard actually is.
The rule: never invent a number
Every figure in every case study we publish is drawn from a real, specific process before and after — not extrapolated, not rounded up for effect. When an automated invoicing workflow says it removes about six hours of admin per pay cycle, that's the difference between the old manual process — reconciling a calendar, a schedule sheet, and a no-API booking platform by hand — and the new one, where a scheduled workflow does the reconciliation and produces ready-to-send invoice drafts. It's a specific, bounded task with a clear before and after, which is exactly the kind of number worth trusting.
Compare that to the weekly marketing-analytics automation, where the client's marketing team was assembling the same mobile-app performance report by hand every week — pulling metrics, updating a tracking sheet, building a slide deck. Two hours, every week, on a task specific enough to time honestly. Automating it didn't just save the two hours; it also meant the report started arriving with AI-written commentary on the trends, which didn't exist in the manual version at all — a real change, but one we keep separate from the time-saved figure rather than folding it into a bigger, vaguer number.
Why some case studies don't have a time figure at all
Not every automation reduces to "hours saved," and forcing that framing onto something that measures differently produces exactly the kind of soft number we're trying to avoid. The AI website integration audit removes roughly 25 minutes of manual audit work per lead — a real time figure, because the old process was a person manually checking a prospect's site. But its bigger effect is response speed: prospects hear back in minutes instead of days, which is a conversion-rate lever, not a time-saved one, and we report it as what it actually is rather than converting it into a number it isn't.
What "honest ROI" actually looks like in a case study
- A specific task, timed before and after — not "we estimate this saves the team significant time," but a bounded process with a real duration on both sides.
- Qualitative outcomes stay qualitative. If a change improves consistency, response speed, or data trust rather than removing a specific number of hours, we say that — we don't manufacture a percentage to make it feel more rigorous than it is.
- No implied scale. We don't say "hundreds of customers" or "a growing team" unless we actually know that number, because vague scale language is often doing the same rhetorical work as an invented statistic.
Why this matters for you as a buyer
If a vendor's case studies are full of suspiciously round, uniformly impressive numbers — "50% faster," "10x ROI," across every single project — that's worth a second look, not more confidence. Real automation work produces uneven results: some processes save six hours, some save twenty-five minutes, and some don't produce a time number at all because the value shows up somewhere else entirely. That unevenness is a sign the numbers are real. Suspiciously perfect consistency usually means someone rounded up.