Can AI grow a brand-new local travel site? The Crank experiment, five weeks in
AI plans, a local approves, budget near zero. 3,470 impressions and 53 clicks since August 26, which pages earned them, and what we can't tell yet.
Proof, published
We rebuild our own sites first and publish what happens: timings, costs, traffic, mistakes included. Every number a client ever hears from us gets earned here first. New notes as the work happens. Measured, not promised.
AI plans, a local approves, budget near zero. 3,470 impressions and 53 clicks since August 26, which pages earned them, and what we can't tell yet.
Two reference sites, 997 quote-bound claims and 212 dark-sky places, one rule: no source, no sentence. Why we think sourcing is the search strategy.
A coaching practice's site went offline with no source and no backups. We rebuilt it from the Wayback Machine, kept every URL, and had it live in 3.5 days.
45,600 impressions, 481 clicks, average position 21.7. Three months of Search Console data on a bilingual directory built on a domain with no history.
7,996 AI crawler requests in 20 days, each checked against the vendor's published addresses. Who showed up, what they read, and what it means for blocking bots.
58 findings, 8 figures deleted, two conflicts left in plain view. What a primary-source check of an AI-researched site found, and where the approval gate went.
A dense, citation-heavy research site comparing the AI boom to 200 years of bubbles. Built as a working demonstration of the methodology we sell.
110 boutiques, 65 cities, two languages, zero plugins. What we built, how the stack held up, and which numbers we're watching next.
Every agency says "trust us." We'd rather show receipts. Here's the plan: five of our own sites, rebuilt in public, numbers included.
Want your site in here someday?
Free, no call required, and you get a written plan either way. The numbers come later, and they'll be real.