Independent comparison reference
People needed to compare versions of a product. I built pages from shared records, measured the search results, and added checks to catch bad data.
I built and maintain this project. Its name and URL stay off this page; the numbers are its own.
See what I builtObserved result Apr to Oct 2026
17.1KGoogle and Bing search clicks
Apr seven complete months Oct
- Impressions
- 587K
- Visits from search
- 27.9K
32 Google clicks17K Bing clicks
Search: Google Search Console + Bing Webmaster Tools. Visits from search: Google Analytics 4. These count different things, and none of them shows what caused a change.
The reader's question
A reader had narrowed a choice but still needed to see which version offered which features. What helped was a side-by-side comparison with every feature defined the same way; another general overview wouldn't have answered it.
One record, several views
The first pages carried facts directly in their markup, which made corrections slow and inconsistent. I rebuilt them around structured records. A template now reads the same fields for the comparison, supporting explanation and FAQ; validators check missing values, ranges, duplicates and disagreements between sources before publishing.
- Definition
- One meaning
- Value
- Checked once
- Source
- Traceable
- Normalize the facts Move repeated facts out of individual pages into structured records.
- Build the comparison Use shared fields for side-by-side views, explanations and FAQs.
- Check every consumer Validate missing values, disagreement and thin output before release.
One later error showed that a shared field can still be read differently by two parts of a site. I traced the value through every consumer and added cross-view checks. The deploy also compares built content with the live version, so a data correction changes only the pages that actually need it.
What happened
From April through October 2026 it earned 17.1K search clicks, including 17K from Bing. That split matters: the result is substantial, but one search engine carries almost all of it.
- April 2026428
- May 20261.1K
- June 20262.1K
- July 20262.8K
- August 20264.1K
- September 20263.5K
- October 20262.9K
Bar lengths compare months within this project. The table gives exact values.
| Month | Search clicks | Impressions | Visits from search |
|---|---|---|---|
| 428 | 14,818 | 647 | |
| 1,101 | 31,826 | 1,809 | |
| 2,103 | 61,230 | 3,468 | |
| 2,846 | 93,820 | 4,577 | |
| 4,142 | 134,556 | 6,633 | |
| 3,515 | 128,254 | 5,921 | |
| 2,932 | 122,502 | 4,872 |
What I corrected
An early spot check found errors across the initial hand-entered pages. I stopped patching page by page, normalized the records, and rebuilt from one source. A later malformed record still produced an indexable page, so I moved the thin-page sweep ahead of release. The open risk is that validation can establish consistency without proving every external fact is correct.
What this evidence cannot prove
The search totals establish that people reached the site, and the organic sessions measure visits under GA4's channel rules. They do not establish why one engine favored it or prove that a particular change produced the traffic. I will not turn that uncertainty into a success story about a tactic.
Complete months, April through October 2026. Google data through Nov 5, 2026; Bing through Nov 4, 2026; Google Analytics through Nov 7, 2026.