Why does month one matter so much?
Two numbers frame it. The Bureau of Labor Statistics' survival tables show 77.9% of private-sector establishments open after one year and 51.4% after five. A startup doesn't get many years to spend, and a year lost to a foundation that has to be rebuilt is a large share of them.
The other is where buyers now look. Google said at I/O in May 2026 that AI Overviews has more than 2.5 billion monthly users and AI Mode more than a billion. Pew Research found people clicked a traditional search result in 8% of visits that showed an AI summary, against 15% when there was none. A new company that the answer engines can't read is starting behind, and the reasons it can't be read are almost always decided at setup.
The decisions we make first
The domain, in the company's name
Registered to the business, with a login the founders control. A domain sitting in a contractor's account is one of the slowest things to recover, because a change of registrant triggers a transfer lock.
The platform
We build static sites served from Cloudflare's network: each page is generated ahead of time with its title, description, structured data and FAQ already in it. Nothing on those pages waits for a plugin update or a database. We stopped building on WordPress, and we've written down why. Where a product needs an application, we build that with Next.js on the same network, which is how this site runs.
Who the accounts belong to
Google Ads, GA4, Search Console, Bing Webmaster Tools and the Business Profile, if you have one, are opened in the company's name. We're added as users. When an investor or a future hire asks for access, you grant it yourself.
Whether the AI engines can read you
Each engine fetches pages with its own crawler, and each publishes how to allow it. Anthropic documents the crawlers Claude uses to search and answer, and says blocking them may reduce your visibility. Perplexity says its crawler exists to surface and link sites in its results. We allow the search crawlers, publish an llms.txt that describes the business plainly, and mark up every page with structured data tying it to one identity.
Measurement from the first day
Search Console, GA4 and Bing are verified on launch day, and every site we build also logs its own traffic at the edge. You can't backfill the months before tracking existed, and those are the months that tell you which first channel worked.
What does a wrong early choice cost later?
Our proof comes from established small businesses we rebuilt, not from startups. The mistakes are the same ones, made earlier.
- An identity the engines can't pin down. When we first asked five engines about Paich & Co. Tax Services by name, on September 24, 2026, ChatGPT, Claude and Perplexity described a similarly named firm in another town. Her old GoDaddy builder site had no business schema and no llms.txt to say which firm she was. A new company with a common name has the same problem from day one.
- Crawlers shut out by default. Southern Belle Construction's old WordPress site blocked the AI crawlers. Nobody noticed, because nothing on the page looks different.
- A platform that has to be replaced. Moving to a new site means new addresses and, in Google's words, visibility that “may fluctuate temporarily during the move.” When we migrate a site we keep or redirect every old address, as we did for Spring Haven Mansion. Better never to need it.
What we'd leave for later
Paid ads before tracking is verified. A content calendar before you know which questions customers ask. Social accounts on every network. A big site when five honest pages would do. Most of what a startup is sold in month one can wait for month four, and the proposal will say which parts.
Before any of it, run your current site or your pre-launch page through the website scan. It runs the same checks every site we build has to pass before it goes live.