16 Nov 2026

Field notes: this quarter's portfolios, and why only one earned a walk-down

Aerial view of a dense commercial and industrial district with many flat and low-slope rooftops, parking lots and warehouses.

This quarter, three regional portfolios came across the desk screen in the same six-week stretch. All three looked workable on paper: decent roof stock, retail and light-industrial mix, the kind of C&I footprint that gets a developer excited before anyone's driven a single site. Only one made it to a walk-down.

That ratio isn't unusual. What's worth writing down is why the other two didn't.

The first portfolio was a grocery-and-pharmacy chain, forty-one rooftops across two states. The desk screen killed it on orientation before obstruction even entered the conversation. A run of sites built in the same era by the same architect, all with the same low-slope mansard detail that puts half the usable roof plane on a north-facing pitch. On a satellite pass you can see it immediately once you know to look: the ridgelines all run the same way because one firm designed every store off the same template.

Thirty-one of the forty-one roofs had less than half their area in a workable azimuth band. The portfolio owner had been pricing this as if every roof pulled its full square footage toward installable kWp. It doesn't work that way, and no PPA bid survives a correction that steep once the EPC's own site assessment catches it three months later.

The second was a distribution and cold-storage portfolio, smaller, nineteen sites, and this one was an obstruction story rather than an orientation story. Cold storage roofs carry a lot of rooftop mechanical: condenser arrays, cooling units, the occasional second-tier HVAC retrofit bolted on after the original build. None of that shows up in a drawing set that's twelve years old, which is what most portfolio owners hand over at the start of a screen. A VHR pass at sub-half-metre resolution picks up every one of those units as a hard obstruction, and on this portfolio the obstruction load ate close to a third of the gross roof area across the board. Combined with setback requirements around the units themselves, the installable kWp number came in low enough that the per-site economics didn't clear the threshold the developer had already quoted to two of their investors.

The third portfolio is the one that went to walk-down, and it's instructive mostly because it wasn't the cleanest-looking roof stock of the three. Mixed warehouse and parking canopy, a logistics operator's distribution network, roofs added in phases over fifteen years so nothing matched. But the azimuth spread was wide enough that even the worst-oriented buildings still had meaningful south-facing area, and the obstruction load was light: a few skylights, standard HVAC spacing, nothing structural in the way. The desk screen ranked the sites, the ranking matched what the field team found once boots hit three of the top five roofs, and the portfolio moved to bid prep without a surprise.

Square footage was never the number that mattered here, and most legacy screening still reports it anyway. A roof can be enormous and still be a poor site once you subtract the north-facing third and the mechanical clutter. Installable kWp per site is the number that survives contact with an EPC's own walk-down, and it's the number an annual VHR re-rank can hold steady as roofs change, new units get added, or a canopy goes up in the parking lot nobody flagged in the original site list. That's the case for running the shortlist through a tool built around that output rather than a spreadsheet of rooftop area, which is the approach behind this site's rooftop ranking pass.

Two portfolios out of three getting cut at the desk is exactly what the screen is for: catching the mismatch before anyone spends a field crew's time finding out the hard way. The expensive version of this story is the one where the walk-down happens first.

If your queue has a portfolio that hasn't been screened past gross roof area yet, that's worth fixing before the next bid cycle.

Start a pilot

← Back to the blog