Dealer survey estimate vs satellite fleet count: how the two methods differ

Tractors and farm machinery parked inside a large indoor storage shed.

If you've built a market-share model off a dealer survey estimate, you already know the uncomfortable part: the denominator is only as good as what your dealers are willing to tell you, and what they can see from their own lot.

Satellite-derived fleet counts build that denominator a different way. Neither method is wrong on its face. They're just counting different things, through different windows, on different timelines. Here's what separates them.

What a dealer survey estimate measures

A dealer survey rolls up through your regional sales team. Dealers report their own sell-through, estimate competitor share based on trade-ins and conversations at the counter, and sometimes drive a competitor's lot to eyeball what's parked there. A regional manager aggregates those reports into a fleet estimate, usually once or twice a year.

The number that comes out is really a measure of what your dealer network happens to be able to see. That has three structural gaps worth naming:

Dealer coverage maps to whatever territory that dealer network holds, and that's often narrower than the territory your OEM competes in. A county with no active dealer relationship shows up as a blank, or gets filled in with a rough extrapolation from a neighboring region.

The survey reflects relationships, not inventory. A dealer who's lost share to a competitor has less incentive, and less access, to report that competitor's fleet accurately. Self-reported estimates tend to understate the players a dealer doesn't deal with.

And the timing lag is real. A fleet that turns over mid-year won't show up until the next survey cycle, so your denominator is always a few months to a year behind whatever's sitting on the ground.

What a satellite-derived fleet count measures instead

A satellite-derived count starts somewhere different: physical units sitting on sites across a region, read directly off high-resolution imagery, regardless of who sold them or whether a dealer has ever set foot on that property. It counts machines by site, independent of channel relationships.

That raw count isn't the finished number, though. Imagery doesn't cover every parcel in a region on the same pass, and some units sit under roofed storage, in shadow, or stacked close enough to hide one another from an overhead view. Getting from what the imagery shows to what's actually out there means correcting for both of those gaps, coverage and occlusion, before the count can stand in as a regional installed-base figure. That correction step is the real work, and it's also the part still being proven across regions rather than treated as a settled, off-the-shelf output.

Where the two numbers diverge, and why it matters for your denominator

The practical disagreement shows up in a few predictable places. Dealer surveys tend to undercount secondary and gray-market units that moved through channels your network doesn't touch. They tend to lag a new entrant's share gains, because that entrant's growth hasn't worked its way into dealer conversations yet. And they're vulnerable to the reporting dealer's own bias toward or against a given brand.

A satellite-derived count doesn't know who financed a unit, what brand loyalty looks like, or what a dealer thinks about a competitor. It only knows what's physically present on a site, scaled up with a stated correction method. For a denominator you have to defend in front of your own leadership or a board, that's a meaningful difference: the dealer number is defensible because of who collected it, while the satellite number is defensible because of how the count was corrected and can be re-run.

Neither replaces field knowledge of pricing, financing terms, or channel strategy. What a scaled, corrected site-level count gives you is a denominator that doesn't rise and fall with which dealers answered the phone this quarter.

If your current market-share figure is still built on a dealer survey estimate and you want to see what a site-level count scaled to a regional installed base looks like against it, that's the comparison worth running next.

Request early access

← Back to the blog