The request usually arrives in a reasonable form: we already track the machines, so let us do the attachments as well. It sounds like an extension of something that works, which is why it is so often the point at which a rollout starts going wrong. From the outside the two problems look identical. Underneath they have different physics, and most products in this market were designed for the other one.
The first question is where it gets its power, and it decides almost everything else. The great majority of telematics hardware assumes a harness: a permanent supply from the machine it is fitted to, and therefore no meaningful constraint on how often it reports or how much it does. An attachment has no supply of its own. A breaker, a bucket, a grapple or a shear is usually a lump of steel with hydraulic connections and nothing electrical to tap. Any product whose answer to this question involves wiring is either not going to fit or is going to require somebody to invent a supply, which is where installations quietly stall.
If the answer is that it runs on its own battery, the follow up matters more than the headline. Ask what determines the life: is it a period of time, or is it a number of transmissions. These give very different answers in the field. A figure quoted in years without a stated reporting rate is not really an answer, because the same unit will last a very different length of time at one report a day and at one an hour. The SolidTrack Field Unit typically supports about 4,000 transmissions at one transmission a day, with signal conditions and temperature both moving the number. That is a transmission budget rather than a guaranteed service life, and it lets you work out your own answer for your own reporting rate, which is the only version that is any use.
Third, ask what it survives, and insist on numbers rather than adjectives. Ingress rating, operating temperature range, and a maximum shock figure. Be particularly careful with shock claims: a heroic sounding number attached to no standard is a warning sign rather than a reassurance, and an attachment environment will find out the truth within a season. Our own unit is rated IP69K, its specified operating temperature range is -30 °C to +85 °C, and it is rated for up to 8 g of shock and vibration. Those are three separate specifications rather than one claim of ruggedness, and the 8 g figure is a rating rather than a guarantee for every mounting position, which is why the unit belongs away from the point of direct impact and why a protective cover exists for high impact work. A supplier who cannot tell you where their limit is has probably not tested against one.
Fourth, how does it fix on, and can it come off again. Attachments rotate. They move between carriers, between sites and between customers, and a fixing method that assumes a permanent home creates work every time an attachment changes hands. A screwed fixing that can be undone and refitted answers this; a permanent one does not. Weld on fixings are worth treating with suspicion for a sensor, because they commit you permanently and they involve heat near electronics. Be equally wary of anything held on by magnets alone on a percussive tool: vibration is exactly the load that defeats friction.
Fifth, and this one catches people out, ask what the word hours means inside the product. There are three common answers: engine time, hydraulic flow, or motion. For an attachment the first is impossible, because there is no engine, so if a product offers you attachment hours and is not measuring motion then it is almost certainly reporting the carrier's number under a different label. That number is not wrong exactly, it is just an answer to a different question, and it will follow the carrier around rather than the tool. Motion is the answer that fits an attachment, but be clear about what it is: recorded activity of that kind is derived from movement rather than read off a physical counter, which makes it good for utilization and for comparison, and a poor substitute for the OEM meter in a warranty conversation.
Sixth, can it tell work from noise. A unit that only knows moving from not moving will give you utilization, and utilization on its own is a useful but shallow number. What sits behind wear on a breaker is not how long it ran but how it ran: whether firing bursts were short and controlled or long and committed, and how much of the period was classified as blank firing. Ask whether impacts are classified and how the bands are defined. If the answer is that everything above a threshold counts as an impact, you have a counter rather than an analysis. Ask as well what the classification is claiming, because a burst band and a blank firing share are both read from the motion pattern: they are indicators to compare and act on rather than a physical diagnosis of the tool, and a supplier who volunteers that distinction is telling you something useful about the rest of their numbers.
Seventh, what happens when there is no coverage, which in this industry is a routine condition rather than an edge case. The answer you want is that readings are held on the unit and sent when the network returns, with a stated buffer size, and a clear account of how the reported position degrades when GNSS has no view of the sky. Ask what happens when an outage outlasts the buffer as well, because a buffer is a finite number of records rather than a promise that the history will come back complete. A supplier who treats coverage as a solved problem has not worked in a pit.
Eighth, whose number is it and how do you get it out. Ask about export, and about integration with whatever you already run for maintenance or hire management. This matters less on day one and a great deal in year three, when somebody wants to reconcile usage against invoices and discovers that the data is only visible inside one web page. SolidTrack answers the first half: reports arrive by email, Enterprise adds daily reports delivered as PDF, and the lists you work with can be downloaded. It does not answer the second half on its own, because there is no public customer API today, so an integration is a conversation with us rather than a self service exercise, and it is better had before purchase than after.
The cheapest way to answer all of this is not a longer evaluation but a narrower one. Pick the small number of assets that already generate arguments in your business, the ones everyone can name without checking, and instrument those. You will find out within one maintenance cycle whether the hours are believable, whether the install survives, whether the coverage story holds up on your sites, and whether anybody actually opens the reports. That is a more honest test than any specification comparison, and it costs very little to be wrong about.