Two engineers in February. Eight in September. $1M in work from the web.

One firm site can't rank in 24 states. The people who buy arc flash, short-circuit, and coordination studies search the service with their state attached, and the firms that come up are local. So the next day, Ohio got its own site: ohiopowerstudies.com. By June 13 every state TPS is licensed in had one, 24 in all, from Alabama to Wisconsin.

All of them come from one template, and forking a state means changing one file. That's what keeps the network maintainable. It's also the fastest way to build a doorway network, which is the thing search engines punish: the same page, over and over, with the place name swapped. We treated that as the main design problem, not an afterthought.

The answer is that each state's file holds that state's actual power landscape. The Ohio site knows Ohio sits inside PJM, that AEP Ohio, FirstEnergy, AES Ohio, and Duke Energy Ohio serve it, and that with no state OSHA plan, Ohio employers answer to federal OSHA. The Texas site knows ERCOT carries about 90% of Texas load, that the Panhandle sits in SPP, Beaumont in MISO, and El Paso in the Western Interconnection. Each site also carries its state's BLS establishment counts by sector and its own FAQ. A doorway page swaps the city name. These swap the grid.

One rule came from the client: these sites are concept-driven. No named clients, no named projects, ever. That rules out the easy proof, so the depth has to come from knowing the state, which is harder to fake and, it turns out, more convincing to an engineer.

• 24 states, 24 domains, one template, one config file per state
• Ohio answers to PJM and federal OSHA; Texas to ERCOT, SPP, MISO, and WECC
• No named clients or projects, by client rule

In August we built a second network, for TPS's mechanical, electrical, and plumbing design work, and the audience changed. Power studies are bought by facility owners. MEP design is bought by architects and developers, people who spend their working day reading construction documents. So the 10 MEP sites, live since August 14, don't look like marketing sites. Each one is a drawing set.

Every section is a sheet with a heavy border and a title strip: sheet number, title, state, issue, firm. The cover sheet carries a real sheet index. G-001 is the code summary, M-101 the services schedule, then the legend, keynotes, and general notes, and the contact form is T-001, the transmittal. An architect knows how to read it before they've read a word.

The code summary is the sheet that matters. An architect searching Kentucky mechanical code is searching for an edition, and a wrong edition on an engineering site is a liability. Third-party code aggregators got Kentucky wrong three ways: they claimed the 2018 IMC (it's the 2015), NEC 2017 (it's 2023), and a 2018 International Plumbing Code for a state that writes its own plumbing code under 815 KAR Chapter 20. So every edition on every MEP site was read from the state's own publication, and each sheet is stamped with the date it was verified and a link to the source.

The drawing-set template is also the fastest thing in the network: mep-oh.com scores 91 on mobile Lighthouse, the median of three runs, with zero layout shift.

• Sheet G-001, the code summary: every edition from the state's own publication
• Aggregators had Kentucky's IMC, NEC, and plumbing code all wrong
• The contact form is sheet T-001, the transmittal

A network this size is only useful if TPS can tell which doors people come through. All 35 sites run on one Sonor project with one API key, and each has its own managed form with its own slug. Every submission is stamped with the host it came from, so the business development team works one queue and still sees that a lead came from Maryland rather than the main site.

The forms don't use CAPTCHA. Sonor only accepts a submission that carries proof a browser actually rendered the page, which stops the bots that post straight at the endpoint without asking a real buyer to pick out traffic lights.

Here's what that pipeline has logged since June 1: 20 form inquiries that passed spam screening. Six came through truepowersystems.com. The other 14 came through 12 different microsites, from Georgia and North Carolina to Washington and Oklahoma. That's 70% of the form inquiries arriving through a door that didn't exist in April.

Forms aren't the whole funnel, and we won't pretend they are. Every site carries TPS's phone number, and calls and direct emails aren't in these counts. Nor is every inquiry a qualified lead. The count shows where people came in, not what they bought.

• 35 managed forms, one Sonor project, every lead tagged with its host
• 14 of 20 form inquiries came through a microsite
• No CAPTCHA: Sonor wants proof the page rendered instead