Search went from nothing to the biggest channel.

Search went from a rounding error to the biggest channel in the store.

Phrase Studios is our own streetwear label, so nothing here is second hand. In June 2026 search sent 177 sessions a month. In September it sent 4 031, and search became 43% of everything arriving at the store, against 13 percent in June.

From June to August social stayed flat, 52 sessions to 78, and paid was off. Nothing else changed on the store. What moved was the writing, and the Journal has since grown to 101 articles.

One property, no sampling. The June figure is the baseline we started from, not a low point we picked. The chapters below were written from the August read, so their figures stop there.

We rewrote the pages that already existed before we wrote a single new one.

Before a single new article went live, we rewrote what was already there. The homepage, every collection page, the product pages, the about page, the FAQ, and an llms.txt so the AI crawlers get a clean read. Same products, same photography, different words.

phrasestudios.com

phrasestudios.com. The homepage after the rewrite, with the routes into the catalogue written as language people actually type.

This is the cheap half of the work and it is the half most brands skip. A store that already ranks for its own name is usually one rewrite away from ranking for the thing it sells.

Then 78 articles in three repeatable shapes, around one hub.

Then the engine. On 20 July we put 65 articles live in one go, 59 new and 6 rewritten, in English and Dutch. The Journal now holds 78 articles and roughly 150 000 words.

/pages/streetwear-guide

The hub. Every article links back to it, and it links out to all of them.

/blogs/journal

The Journal index, 78 articles deep and growing in two languages.

Country roundup

A country roundup, one of seventeen. These are the pages the AI engines quote most.

Brands like X

A brands like X article, one of twenty one. Built for the search that starts with a competitor.

524 links, all of them inside the prose.

An article nobody can navigate out of is a dead end. Every link below sits inside the prose, in the sentence where it makes sense, not in a related posts widget under the fold.

Links in prose

Links live in the sentence. A reader following one is already halfway to a product page.

FAQ and structured data

A question block on all 78 articles, marked up so Google and the AI engines can lift the answer whole.

Both batches were submitted to Google and Bing on the day they went live, and measured every week after that. Nothing was left to be discovered.

Live on 20 July. Quoted by Google on 24 July.

Six weeks from a quiet blog to the biggest channel in the store. This is the actual sequence, dated.

Search Console › Performance

Google Search Console, daily clicks and impressions across the summer of 2026. The step is the week the Journal went live.

The AI engines started answering with our pages.

This is the part most agencies cannot show you yet. Between June and August the articles were shown 27 657 times inside AI answers across 163 pages, and Copilot cited them 8 257 times in August alone against 219 in July.

Search Console › AI features

Google Search Console, AI features. 27 657 impressions between 31 May and 29 August 2026, and the pages Google chose to show.

The July AI figure of roughly 3 864 impressions is derived by subtraction from the three month total, not read off a dashboard. The June and August figures are read directly.

Sessions, baskets, orders and the email list all moved together.

Impressions are not the job. Here is what happened to the store underneath them.

On last click, search is credited with 3 orders in August against 0 in June. Last click credits the final tap and nothing before it, which is why the honest read is the one above: every channel in the store got bigger at once, in the month the writing landed.

What this looks like on your site.

The order matters more than the volume. This is the sequence we would run on your store, and it is the same one that produced everything on this page.

1. Audit the site and map the demand. What you already rank for, what you nearly rank for, and what the AI engines currently say when somebody asks for a brand like yours.
2. Rewrite what is already there. Homepage, collections, product pages, FAQ. The cheapest traffic in the account is the traffic your existing pages nearly earn.
3. Add the engine when it fits. A hub, then the two or three article shapes that suit your catalogue, published in batches rather than one a week.
4. Wire it in and measure it. Links inside the prose, structured data on every page, submitted the day it ships, then read weekly against the baseline we took in step one.

You do not have to take the whole thing. Most brands start at step two, because it pays for step three.