If you cover off-highway equipment, you already know the problem. In the U.S. and a handful of other markets, registration data gives you a clean read on who's buying what. Everywhere else, that data source just doesn't exist in usable form, or it's fragmented across provincial agencies that don't talk to each other, or it only covers on-road units and misses the construction and ag iron you care about. You're left trying to build a market share number with no denominator underneath it.
This is the quiet failure point in a lot of regional strategy decks. The numerator (your own shipment or revenue data) is solid. The denominator (total installed base, competitors included) is a guess dressed up with a footnote.
Why dealer surveys don't fix this
The standard workaround is a dealer survey: call around, ask channel partners what they're seeing on lots and in the field, triangulate a number. It works, sort of, for the segment of the fleet that moves through dealers who answer your calls. It misses gray-market imports, equipment bought direct from a competitor's regional distributor, machines still running past their normal replacement cycle, and anything owned by a fleet operator who buys opportunistically and doesn't talk to anyone's channel team.
Dealer surveys also carry an obvious bias. A dealer tends to know their own brand's install base well and everyone else's install base poorly. Ask five dealers for a competitor's unit count in their territory and you'll get five different guesses, none of which is really a count.
None of this means the survey approach is worthless. It's often the only option when there's genuinely no other data source. The issue is that it produces a number you can't really defend when someone on the board asks how you got it.
Building a denominator from what's actually sitting on the ground
The alternative is to stop trying to infer the fleet from paperwork and count it directly. High-resolution satellite imagery can resolve individual machines at construction sites, quarries, ag operations, and equipment yards. That gives you a raw site-level count: how many loaders, excavators, tractors, or whatever category you're tracking are visible at a sampled set of locations across a region.
Turning a sample of sites into a regional total is the hard part of this work. You need to correct for coverage (you didn't image every site in the region, so you're scaling up from a subset) and for occlusion (machines parked under cover, stacked behind other equipment, or partially hidden by terrain don't show up in a raw count even when they're there). Skip either correction and you've just replaced one unreliable number with another one that merely looks more rigorous because it came from a satellite.
This is the specific problem Installed Fleet Estimation is built around: scaling site-level machine counts up to a regional installed base with coverage and occlusion correction built into the math, refreshed on a quarterly cadence so the denominator moves with the fleet instead of going stale between annual updates.
What this gets you that a survey doesn't
A survey gives you an estimate with no visible method behind it. A coverage-corrected count gives you a number you can walk a skeptical colleague through: here's the sample, here's the correction applied, here's why the total scales the way it does. That's the difference between a market share figure that survives a planning review and one that gets quietly revised every time someone asks a follow-up question.
It's worth being honest about where this stands today. The scaling method is still being validated region by region, which is exactly why it isn't live as a published index yet. Analysts who want in are working with it directly, on a request-access basis, while that validation continues.
If your next regional plan needs a market share denominator you can actually defend in the room, it's worth finding out whether your region is one where this approach is ready to run.