DEV Community

Ian
Ian

Posted on

Index count went 62 91 in five days. Here's what I actually changed.

Small site, four locales, no ad budget. Search Console at the start of the week: 62 pages indexed, 50 impressions over three months, average position 50.

Five days later: 91 indexed, 115 impressions, 9 clicks.

I want to be careful here, because SEO posts are full of "I did X and traffic tripled" claims that don't survive scrutiny. So let me be precise about what changed and what I can't attribute.

What I changed

1. Compared content depth against competitors, per locale.

I searched the exact query my French page was ranking for (discours mariage frère, position 82) and measured body text on the top six results:

Site Body text
competitor A 21,746
competitor B 16,115
competitor C 10,971
competitor D 8,733
competitor E 7,809
mine 3,653

The page wasn't thin because I was lazy — it was thin because I'd written two examples and called it done, while every competitor was running eight.

Fix: expanded to four examples plus six FAQ entries, all answering questions the search results themselves implied (how long, what structure, can I mention this, what do I do if I cry). 3,653 → 7,437 characters.

2. Did the same audit on every locale.

Turns out the French pages were the good ones. Japanese pages averaged 1,455 characters with zero FAQ blocks. Same template, wildly different depth, because I'd written different amounts for different languages at different times.

Ten Japanese pages: 14,554 → 25,011 characters total.

3. Added FAQPage markup — but only where the FAQ was visible.

This is the part people get wrong. Structured data has to match rendered content. I generate both from one array so they can't drift:

{cat.faq?.length > 0 && <FaqJsonLd items={cat.faq} />}
Enter fullscreen mode Exit fullscreen mode

4. Fixed a locale that had no content at all.

The English example library was an empty array:

const exampleLibrary = {
  ja: jaCategories,
  fr: withExtraExamples(frCategories, extraExamples.fr),
  es: withExtraExamples(esCategories, extraExamples.es),
  en: [],   // ← empty
};
Enter fullscreen mode Exit fullscreen mode

Meanwhile Search Console showed best man speech generator at position 54 — my best ranking on the entire site — landing on a page with nothing under it. Added five English categories.

5. Pushed the URLs.

IndexNow (Bing, Yandex) accepted 28 URLs, HTTP 200. Google was left to the sitemap, since the Indexing API is for job postings and livestreams.

What I can't claim

  • Five days is too short. Google recrawls on its own schedule. Part of the 62 → 91 may be pages that were already queued.
  • Correlation with the content work is plausible, not proven. I changed five things at once. Classic confounded experiment.
  • Impressions going 50 → 115 is a 3-month cumulative figure, so it partly just means time passed.
  • 9 clicks is not a business. It's a rounding error.

What did move measurably

The keyword I worked on most specifically:

Before After
impressions 7 14
avg position 82.6 76.0
clicks 0 0

Position improved 6.6 places. Still page 8. Still zero clicks. That's the honest shape of this kind of work — you move, but not into the money.

What I'd tell someone starting

  1. Measure competitors per locale. Your site average hides which locale is empty.
  2. A page can rank and still be useless. Position 54 with no content behind it is worse than position 90 with content, because you're burning the one impression Google gave you.
  3. Structured data mirrors visible content, always. One source of truth.
  4. Expect weeks, not days. The feedback loop is the hard part, not the work.

If you want to check the same thing on your own site: pick the query where you rank worst-but-visible, measure the top five results' body text, and compare. If you're at a quarter of their length, you don't have a technical problem.

(Site in question: Speech Kobo. The open dataset behind its content checks: GitHub.)

Top comments (1)

Collapse
 
onelifevibe profile image
OneLifeVibe •

I appreciated how careful you were about the caveats, especially that five days is too short to prove the content changes caused the index jump. The per locale audit was the most useful part for me, since the French pages looked fine while the Japanese pages averaged only 1,455 characters with no FAQ blocks. Finding a page ranking at position 54 with nothing behind it is a strong reminder that ranking alone does not mean the page is useful.