Where Demand Outruns Supply. (Sometimes.)

The Idea

Establishment counts alone don't tell you if a market is saturated.

Car washes have, by a wide margin, the most search interest relative to how many businesses exist to catch it. At 0.79 interest points per establishment, car washes lead the next-closest trade (finish carpentry, 0.34) by better than 2 to 1. This lines up with something the original Trade Data project already found from a completely different data source: car wash wages grew 67.5% over five years (the fastest of any trade tracked) while the number of car washes grew only 12%. Two independent datasets, BLS wage data and Google search interest, are both pointing at the same thing: real demand outpacing the businesses available to meet it.

Roofing stays near the top even though it's already the most saturated trade in Tulsa. Roofing has the highest location quotient of any of the 15 trades (1.88, 88% more roofing contractors per capita than the national average) and it still posts the third-highest interest-per-establishment ratio. That's not really a contradiction: Oklahoma's hail corridor drives genuinely high demand for roofing work, high enough that even a crowded market hasn't caught up to it.

Site preparation contractors show essentially zero direct search interest, confirming it's a subcontractor trade, not a searched-for service. The interest index rounds to 0.0. That's not a data gap; it's consistent with how the trade actually works: site prep is booked through general contractors and developers, not homeowners typing a search into Google. It's also the most under-represented trade by location quotient (0.44), which now reads less like an opportunity and more like a trade that was never going to show up in consumer search in the first place.

Finish carpentry pairs a below-average business count with the second-highest demand signal on the list. Its location quotient (0.80) already flagged it as under-represented in the original Trade Data findings; this adds a second signal pointing the same direction: real search interest, relatively few businesses to answer it.

Electrical, plumbing/HVAC, and janitorial services sit at the bottom of the ratio despite real establishment counts, worth reading carefully, not literally. These three have some of the highest raw establishment counts in the dataset (267, 337, and 160) but the lowest interest-per-business ratios. That could mean those markets are genuinely well-supplied. It could also mean the single search term used for each (see the methodology below) undercounts real demand that shows up as more specific searches: "water heater replacement," "ac not cooling," "office cleaning company," rather than the broad occupational term tracked here. I'd treat this one as a real limitation of the method, not a confident finding either way.

Google Trends reports relative interest on a 0-100 scale within whatever set of terms you compare at once, it does not give raw search volume for free, and every comparison batch gets rescaled to its own highest term. To compare all 15 trades on one consistent scale, I ran them in four batches of up to five terms each, with "electrician" included in every batch as a shared anchor, then rescaled every batch back to a single common index using that anchor's value. Data is weekly, Tulsa DMA (Google's ad-market geography, not a Census boundary, more on that below), averaged from September 2021 through September 2026.

Term selection mattered more than expected, and got a couple of things wrong on the first pass. The obvious search term isn't always the real one. "Roofer" returned an interest index of 0.14; "roofing" returned 40.23, a 280× difference for what's supposed to be the same trade. "House painter" returned 0.05; "painter" returned 8.8. And "mason" looked promising at 57.89 until it became clear that's mostly picking up searches for the first name Mason and Mason jars, not masonry contractors: "masonry contractor" and "brick repair" both came back under 0.1, confirming the true number is much closer to the low end. Every term used in the final table was spot-checked against at least one alternative phrasing before being locked in; the ones that looked wrong got swapped.

The geography is a real mismatch, stated plainly instead of glossed over. Google Trends' "Tulsa OK" region is a Designated Market Area (DMA): a 22-county media market spanning parts of Oklahoma and Kansas, built for TV ad buying, not a Census or BLS boundary. The establishment counts and location quotients it's compared against here are Tulsa County only, the same single-county figures used throughout the Trade Data project. That means this page is comparing a broader regional demand signal against a narrower, single-county supply count. The core of that DMA's population sits in the Tulsa metro, so the comparison is still directionally useful, but it is not a precise county-to-county match, and shouldn't be read as one.

The Interest / Establishment ratio is a simple division, not a hidden formula. Search interest index ÷ establishment count, nothing else folded in. It's not a composite score or a ranking of "best trade to start", both inputs are shown in the table above so anyone can recompute it, check a different trade's ratio, or disagree with how much weight to put on it.

Full detail on the establishment counts, location quotients, and BLS sourcing this page depends on lives in the Trade Data project's own methodology, this page only covers what's new here: the Trends side of the comparison.