Case study · live
We run Rank Sphere on Rank Sphere.
No screenshots from a happy customer, yet. Instead, the real numbers from our own workspace, read from the database every hour. When they are small, they are small. Watch them move.
10
blog articles published by the director
in 53 days, on blog.ranksphereai.com
35
comparison and industry pages written by the engine
served on this site under /compare and /seo-for
86/100
site health score
from 91 at the first audit
52
keywords tracked daily
Google and Bing, plus AI Overview presence
207
Google impressions, latest 28-day window
0 clicks, from Search Console
462
AI crawler visits to our blog, 30 days
Anthropic ClaudeBot 287 · OpenAI GPTBot 67 · Amazon Alexa 43
What we learned so far
Our first articles came out with the year wrong in the title. The writer had no idea what day it was. Every customer's writer now gets today's date. That fix came from using the product, not from a bug report.
A comparison page written by the engine reads better than the hand-written one it replaced, because it has to pass the same gates our customers' articles pass: sources checked live, a question-and-answer block, a readability floor, and a human-sounding score.
Publishing our own landing pages taught us something customers already pay us for: a script that skips the "only gate-passing drafts go live" rule ships thin pages. Fourteen guides went out under the bar, we pulled them the same night, and the engine gained a structure-repair step so a draft missing its answer paragraph or question block gets fixed instead of stored.
The learning loop now runs every Monday: articles that earn clicks are compared with the ones that don't, and the winning length and structure become the writer's target for everyone. This page will say what it found.