Arguments about predictive versus preventive maintenance are usually about philosophy and are settled by cost, so it is worth being precise about the shape of each. Preventive costs are recurring and roughly linear in the number of assets and visits: more units, more hours, more money, forever. Predictive costs are mostly upfront and per monitored asset, then small. Those are different curves, they cross at different points for different assets, and the crossing point is the actual decision.
The preventive cost is a rate, and it is easy to compute
Assets multiplied by visits a year multiplied by hours a visit multiplied by an hourly rate. That is the whole of it, and any site can work it out from figures it already has. It scales linearly and it never stops, which is both its weakness and the reason it is so easy to budget for and so hard to argue against.
The predictive cost is a step, then a trickle
Instrumentation, installation, a baseline period and either a person who can interpret readings or a service that does. That is a step cost per monitored asset and a smaller running cost afterwards. It only makes sense where the avoided failure is expensive enough to clear the step, which is why it concentrates on large rotating plant and rarely reaches the rest of a site.
The number that decides it is the one nobody has
What a failure of this specific asset actually costs, including the emergency response, the downtime and the collateral. Sites that have kept a maintenance record can estimate it from their own history; sites that have not are guessing, and guessing usually favours whichever option was proposed most recently. That is the strongest practical argument for keeping the record from day one.
Questions people ask about predictive vs preventive maintenance
Is predictive maintenance cheaper in the long run?
On the assets it suits, often. Across a whole site, rarely, because most assets do not clear the step cost. Treat any general claim either way with suspicion; the answer is per asset.
How do we estimate what a failure costs?
From your own history where you have it: response time, downtime hours, what the downtime was worth, and what the emergency repair cost against the planned one. Where you have no history, estimate and mark it as an estimate.
Does software change the cost comparison?
It reduces the administrative cost of the preventive side, which matters most where the plan is large. It does not change the labour of the visits themselves, which is where the money actually is.