TL;DR: Ship performance monitoring is worth what its output can survive. Judge it on three things: whether the result arrives as a decision rather than a parameter, whether you can see which readings were excluded and why, and whether the model underneath has been verified by somebody who does not sell you the software.
A fleet manager asks a performance platform a simple question: is this ship burning more than she should, and by how much. What comes back is a screen of engine parameters.
That gap is what stalls most evaluations. The data is real and the platform works, but the output is written in a language only the technical team speaks, and the person who has to sign for it cannot see what they are buying.
So the question worth asking before you buy is not how much data the system collects. It is what it hands you at the end, and whether that output will hold up when somebody disputes it.
What the output should actually be
The output you should expect is a quantified deviation from expected behaviour, not a reading from a sensor.
Engine parameters are an input. The useful result states how the vessel performed against how she should have performed in the conditions she was actually in, expressed in tonnes of fuel or in money. Making that conversion needs a modelled expectation that changes with operating condition, which is the part that separates a monitoring bill from a management tool.
ZeroNorth ship performance monitoring is built around that conversion. SMARTShip monitors machinery condition continuously, validates incoming reports automatically, flags readings that sit outside expected behaviour, and resolves the result into one dataset both teams read the same way, so the technical and commercial sides of the business stop working from different numbers. SMARTShip Prime does this with no onboard hardware, working from the data the vessel already produces. SMARTShip Professional adds real-time telemetry across more than 5,000 onboard data points per vessel where machinery-level resolution is what you need.
Why you have to be able to see the working
If you cannot see which readings were excluded and why, you cannot defend the figure that remains.
Every performance system filters. Good-weather windows, steady-state periods, manoeuvring, time in shallow water: something has to decide which hours count towards a performance figure and which are set aside. Those choices decide the answer as much as the raw data does, and two platforms applying different rules to the same voyage will report different results without either being wrong.
So ask three things of any system you are evaluating. What is the exclusion rule, stated plainly. How many data points did it remove on a given voyage. And what does the same voyage look like with the filter off. A platform that cannot show you is asking you to accept its arithmetic on trust, and that trust gets tested the first time the number costs somebody money. The related question of which source to believe when reported and sensed data disagree is covered in why your noon reports and your sensors disagree.
Which level of monitoring matches the decision you are funding
Match the monitoring level to the decision it has to support, because each decision carries a different evidential burden.
What makes a performance figure defensible outside your own company
A performance figure becomes defensible when the model behind it has been checked by somebody who does not sell you the software.
Data you keep for your own management can be approximate, because the only person who has to accept it is you. A figure you put in front of a charterer, a financier or an auditor cannot be, and the difference is independent verification. ZeroNorth's AI-enabled fuel model achieves a median accuracy above 90% across the fleet, on a platform that has processed more than 2.7 million voyages.
That is also the fair test to put to anyone else at this stage. Not how many data points they ingest, but which of their outputs somebody outside their company has checked, and what the verified figure actually was.
How to prove the value to whoever signs for it
Value is proved by measuring a small number of things the owner already cares about, starting before the system goes live.
The usual failure is not that the system delivered nothing. It is that nobody recorded the before, so at renewal the case rests on anecdote and the finance question gets asked again from scratch. Capture a baseline in the weeks ahead of go-live, agree three measures with whoever is funding it, and report the same three every month without changing the definitions.
Three that travel well: fuel consumed per day at comparable loading and weather, time spent outside the expected performance band, and the proportion of reports that needed correction before anyone could use them. The last one is the most persuasive to an owner, because it is the one they can feel without reading a chart.
