日本語版を読む / Read the Japanese edition
On day one, I documented the gap between a site's published articles and the traffic I could actually verify. Day two brought tangible changes: a separate English section went live, two English articles were published, and Google Analytics 4 (GA4) began giving us a clearer view of how the site was being opened.
One number looks encouraging. Pageviews increased from 36 on October 9 to 50 on October 10. But sessions dropped from seven to four. And because we spent the day testing the new pages ourselves, the higher pageview count is not evidence that the audience grew.
This is the second entry in a public experiment to rebuild ichigoro.com using measurable evidence rather than publication volume.
The numbers: more pageviews, fewer sessions
I checked the connected GA4 property at approximately 9:25 p.m. Japan time on October 10, 2026. October 10 was still in progress. These are provisional, same-day reports, and GA4 processing may change the final counts.
| GA4 metric | Oct 9 | Oct 10 (partial) |
|---|---|---|
| Pageviews | 36 | 50 |
| Sessions | 7 | 4 |
| Active users | 6 | 3 |
| Engagement rate | 85.7% | 75.0% |
| Key events | 0 | 0 |
Pageviews rose by 14, or about 39%; sessions fell by three, or about 43%. A pageview is a page display, not a person. Repeated visits, browser testing, and different devices can all affect these small numbers.
GA4 classified every recorded session across these two days as Direct, with source/medium (direct) / (none). That classification does not identify a successful organic-search acquisition channel.
A likely explanation: we were testing our own English pages
October 10's page-level report shows 13 pageviews for the English homepage /en/, eight for /en/career/, and six for /en/about/. Those three pages account for 27 of the day's 50 pageviews (54%).
That was also the day we checked the new bilingual navigation, category archives, images, and layouts at desktop and mobile widths. The overlap is suggestive, but it is not a user-level attribution: GA4 does not tell us from this report exactly which visits were internal tests. I cannot honestly call these views new English-speaking readers.
The first growth-log article also received post-publication browser QA visits on day one. We need to distinguish our own verification traffic from independent readers before interpreting a two-day trend.
What actually changed on day two
1. Japanese and English now have separate site sections
Previously, the English menu was essentially a link to an English category archive, not a matching English site experience. We launched a dedicated English homepage, English article listings, and topic pages while keeping Japanese posts out of the English lists and vice versa.
The release record shows 30 successful browser checks: 15 routes at two viewport widths (390px and 1365px), with HTTP 200 responses and no broken images or horizontal overflow in the tested pages. That's meaningful evidence of working navigation. It does not prove a search-ranking improvement.
2. Two career articles were adapted into English
The first, published at 6:57 a.m. JST on October 10, covers what to prepare before focusing on English when applying for IT jobs at foreign-affiliated companies in Japan. A five-step English illustration was added later that morning.
The second went live at 6:50 p.m. JST: how to align expectations ahead of performance reviews at foreign-affiliated IT companies in Japan. Both are grounded in existing Japanese articles, and the publishing records include checks on their content and language classification.
The public WordPress post count moved from 29 to 31. English-language posts went from three to five. Publishing two articles is an output, not an audience outcome. The second article had only been online for a few hours when the analytics snapshot was taken.
3. The publishing workflow became reproducible
We also documented the actual WordPress publishing path and its safeguards in GitHub: content checks, narrow backups, rollback steps, and mobile/desktop verification. This matters when using AI to assist with content. The goal is not to publish faster at any cost; it is to avoid losing the source, overwriting the wrong article, or turning unverified metrics into a success story.
What about Google Search traffic?
We do not have settled Search Console data for October 9 or October 10 yet. The latest settled date available when checked was October 6. Across the settled September 9–October 6 period, the property had one Google Search click and two property-level impressions. That click was attributed to an older article, not the new growth log or career translations.
Search Console and GA4 measure different things and follow different reporting schedules. The earlier one-click figure is a baseline for confirmed Google Search activity, not a statement that there was exactly one visitor during these two days. It would also be misleading to report zero October 10 search clicks before the data is settled.
What I will measure next
The next milestone is not another batch of translations. It is evidence that someone outside the testing process found a useful page.
- Keep track of owner and QA visits so they are not mistaken for external audience growth.
- Revisit Search Console when the October 9–10 data becomes settled, comparing queries and landing pages where the data is available.
- Track the two new English articles for at least seven days, focusing on meaningful external discovery and follow-on reading.
- Check whether relevant calls to action and inquiries are actually measured. A GA4 key-event count of zero is not the same as verified zero inquiries or revenue.
Day two produced a better bilingual site and two additional English articles. Those are concrete publishing results. Whether this has attracted real readers—or created commercial value—remains unproven. That's the question the next update needs to answer.
Data notes: GA4 dates use Asia/Tokyo, with October 10 captured at approximately 21:25 JST before the day ended. Google Search Console was settled only through October 6, using its America/Los_Angeles reporting basis. Publication facts were checked against live WordPress and the site's GitHub release logs. Evidence: Day-two report record, bilingual release checks, and English article release records.