We Published One Article a Day for 90 Days. Here Is What Happened to Traffic
By SiteRanking team,
A transparent look at a 90-day daily publishing experiment on a small business site: impressions, clicks, AI mentions, leads, and what we would do differently.
Everyone says "publish consistently". Few show the numbers. So we ran the experiment on a real small business site (a regional home services company, two locations) and tracked everything in Search Console, Analytics and a weekly AI-mention check.
The setup
- Starting point: 38 pages, 6 blog posts, about 320 impressions and 15 clicks a week.
- The rule: one article a day, every day, for 90 days. Each article answered one real customer question, 900 to 1,400 words, answer-first, with an FAQ and internal links.
- Technical fixes in week one: titles, sitemap, schema, speed.
- No ads, no link buying. Two partner links and a local news mention came naturally.
Impressions: the first thing that moved
Impressions doubled in week two, before any ranking was on page one. Google started testing the new pages for long-tail queries almost immediately.
Clicks: three weeks behind
Clicks lagged impressions by roughly three weeks. By week 12 the site had 18x the clicks it started with. About 60 percent came from articles published in the first month, which had time to mature.
Shortcut: SiteRanking does all of this on autopilot: it audits your site, finds the keywords and questions your customers ask, writes and publishes one article a day, and shows you a before-and-after chart. Your first article is free.
AI mentions: zero to real
We asked ChatGPT, Perplexity and Gemini 20 local buyer questions every Friday.
Perplexity moved first, because it cites fresh pages aggressively. ChatGPT followed once pages ranked on Google.
Leads: the number that matters
Form submissions and calls from organic search went from 4 in the month before to 23 in month three. Three leads mentioned "the AI said you were the best option". Cost per lead from ads at the same company: $86. Cost per organic lead after the experiment: the time to approve an article.
What we would do differently
- Start the technical fixes two weeks before publishing, not during.
- Write comparison and pricing pages in week one; they converted best.
- Refresh the top ten pages at day 60 instead of day 90.
Do it on autopilot
Everything above is work you can do by hand. Most founders do not have the hours. SiteRanking runs the whole loop for you:
- Scans your site and your competitors, finds the gaps.
- Checks what ChatGPT, Claude, Gemini and Perplexity say when someone asks for a business like yours.
- Writes, optimizes and publishes one article a day, in your language, for your country.
- Shows the before-and-after: clicks, impressions, AI mentions and revenue, on one Home page.
Your first article is free, no card needed. Start here.
FAQ
Does this only work for home services? No. SaaS, e-commerce, agencies and clinics show the same shape; the slope differs by competition.
Is one a day too much? The point is consistency. Three a week works; one a month does not.
Did you write them by hand? The first ten yes. The rest were drafted by SiteRanking and approved by the owner in about ten minutes a day.