The new-site boost ended. Three small tools then beat my 312 content pages in a week

This is a follow-up to my last post, where I rewrote 78 meta descriptions and ruined my own control group. The project is a tarot reference in four languages: https://tarot-magic.com. It has 312 card pages (78 cards in each language) and a handful of interactive pages.

I pulled fresh Search Console data this week. Two things stood out, and neither of them was the experiment.

The boost I mistook for ranking

Two equal 28-day windows:

Same clicks, a third of the impressions. The daily chart explained it. From August 25 to August 31 the site got roughly 700 to 970 impressions a day at an average position of 12 to 17. On September 7 it was down to 239, and from September 10 on it mostly stayed between 50 and 110 a day.

That week in late August was not my site being good. Google tried the new pages high, saw how people reacted, and moved them back. People call it the new site honeymoon. I had read about it and still spent September comparing every number against those few days.

If you launch something, write down your first two weeks and then stop looking at them. They are a test, not a baseline.

Three tools, one week

On October 10 I added two small interactive pages: a love reading with three cards and a page that pulls one random card. On October 16 a birth card calculator went live. You enter a date of birth and it adds the digits down to one of the 22 Major Arcana. Each exists in four languages, so 12 new URLs in total.

Then came the week of October 12 to 18:

Impressions doubled in a week. Here is the 1,238 split by page type:

The Russian version of the birth card calculator alone got 196 impressions and 3 clicks at an average position of 23.5. It had been live for three days. No single card page got more than 34 impressions that week.

Why I think it happened

Three days of data is thin, so this is a guess. My best explanation is intent. A card page answers "what does the Ten of Swords mean", and that question already has plenty of good answers on sites that are years older than mine. "Calculate my birth arcana from my date of birth" asks the page to do something. A page that actually does it seems to get a chance much sooner, even on a young domain.

The queries point the same way. Most of the calculator's impressions came from dozens of phrasings of one request (find my arcana by date of birth, arcana calculator, calculate it online), each with one or two impressions.

Position 23.5 is page three, so there is nothing to celebrate yet. Still, most card pages needed weeks to get that far, and many of them never did.

What I'm changing

I stopped comparing against August. The baseline is now the first half of October. New work goes to small tools that answer a "do something" query, and to linking those tools from articles I already have; the card pages stay as they are.

In two weeks the calculator will have a full 14 days of data. I'll run the same split again, and if it falls back to card page level, I'll write that too.

Has anyone seen tool pages on a new domain pick up faster than content pages? I'm curious whether it holds after the first few weeks.